One document matched: draft-ietf-nfsv4-rfc5666bis-04.xml
<?xml version='1.0' encoding='ascii'?>
<!DOCTYPE rfc SYSTEM "rfc2629.dtd">
<?xml-stylesheet type="text/xsl" href="rfc2629.xslt" ?>
<?rfc strict="yes" ?>
<?rfc toc="yes"?>
<?rfc tocdepth="2"?>
<?rfc symrefs="yes"?>
<?rfc sortrefs="yes" ?>
<?rfc compact="yes" ?>
<?rfc subcompact="no" ?>
<rfc category="std" ipr="trust200902" docName="draft-ietf-nfsv4-rfc5666bis-04" obsoletes="5666" updates="" submissionType="IETF" xml:lang="en">
<front>
<title abbrev="RDMA Transport for RPC">Remote Direct Memory Access Transport for Remote Procedure Call </title>
<author initials="C.L." surname="Lever" fullname="Charles Lever" role="editor">
<organization abbrev="Oracle">Oracle Corporation </organization>
<address>
<postal>
<street>1015 Granger Avenue</street>
<city>Ann Arbor</city>
<region>MI</region>
<code>48104</code>
<country>USA</country>
</postal>
<phone>+1 734 274 2396</phone>
<email>chuck.lever@oracle.com</email>
</address>
</author>
<author initials="W.A." surname="Simpson" fullname="William Allen Simpson">
<organization>DayDreamer </organization>
<address>
<postal>
<street>1384 Fontaine</street>
<city>Madison Heights</city>
<region>MI</region>
<code>48071</code>
<country>USA</country>
</postal>
<email>william.allen.simpson@gmail.com</email>
</address>
</author>
<author initials="T.T." surname="Talpey" fullname="Tom Talpey">
<organization abbrev="Microsoft">Microsoft Corp. </organization>
<address>
<postal>
<street>One Microsoft Way</street>
<city>Redmond</city>
<region>WA</region>
<code>98052</code>
<country>USA</country>
</postal>
<phone>+1 425 704-9945</phone>
<email>ttalpey@microsoft.com</email>
</address>
</author>
<date/>
<area>Transport</area>
<workgroup>Network File System Version 4</workgroup>
<abstract>
<t>This document specifies a protocol for conveying Remote Procedure Call (RPC) messages on physical transports capable of Remote Direct Memory Access (RDMA). It requires no revision to application RPC protocols or the RPC protocol itself. This document obsoletes RFC 5666. </t>
</abstract>
</front>
<middle>
<section title="Introduction" toc="default">
<t>This document obsoletes RFC 5666. However, the protocol specified by this document is based on existing interoperating implementations of the RPC-over-RDMA Version One protocol. </t>
<t>The new specification clarifies text that is subject to multiple interpretations, and removes support for unimplemented RPC-over-RDMA Version One protocol elements. It makes the role of Upper Layer Bindings an explicit part of the protocol specification. </t>
<t>In addition, this document introduces conventions that enable bi-directional RPC-over-RDMA operation, enabling operation of NFSv4.1 <xref target="RFC5661" pageno="false" format="default"/> on RDMA transports, and that enable the use of RPCSEC_GSS <xref target="RFC5403" pageno="false" format="default"/> on RDMA transports. </t>
<section title="Requirements Language" anchor="requirements-language" toc="default">
<t>The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in <xref target="RFC2119" pageno="false" format="default"/>. </t>
</section>
<section title="Remote Procedure Calls On RDMA Transports" toc="default">
<t>Remote Direct Memory Access (RDMA) <xref target="RFC5040" pageno="false" format="default"/> <xref target="RFC5041" pageno="false" format="default"/> <xref target="IB" pageno="false" format="default"/> is a technique for moving data efficiently between end nodes. By directing data into destination buffers as it is sent on a network, and placing it via direct memory access by hardware, the benefits of faster transfers and reduced host overhead are obtained. </t>
<t>Open Network Computing Remote Procedure Call (ONC RPC, or simply, RPC) <xref target="RFC5531" pageno="false" format="default"/> is a remote procedure call protocol that runs over a variety of transports. Most RPC implementations today use UDP <xref target="RFC0768" pageno="false" format="default"/> or TCP <xref target="RFC0793" pageno="false" format="default"/>. On UDP, RPC messages are encapsulated inside datagrams, while on a TCP byte stream, RPC messages are delineated by a record marking protocol. An RDMA transport also conveys RPC messages in a specific fashion that must be fully described if RPC implementations are to interoperate. </t>
<t>RDMA transports present semantics different from either UDP or TCP. They retain message delineations like UDP, but provide a reliable and sequenced data transfer like TCP. They also provide an offloaded bulk transfer service not provided by UDP or TCP. RDMA transports are therefore appropriately viewed as a new transport type by RPC. </t>
<t>In this context, the Network File System (NFS) protocols as described in <xref target="RFC1094" pageno="false" format="default"/>, <xref target="RFC1813" pageno="false" format="default"/>, <xref target="RFC7530" pageno="false" format="default"/>, <xref target="RFC5661" pageno="false" format="default"/>, and future NFSv4 minor verions are obvious beneficiaries of RDMA transports. A complete problem statement is discussed in <xref target="RFC5532" pageno="false" format="default"/>, and NFSv4-related issues are discussed in <xref target="RFC5661" pageno="false" format="default"/>. Many other RPC-based protocols can also benefit. </t>
<t>Although the RDMA transport described here can provide relatively transparent support for any RPC application, this document also describes mechanisms that can optimize data transfer further, given more active participation by RPC applications. </t>
</section>
</section>
<section title="Changes Since RFC 5666" toc="default">
<section title="Changes To The Specification" toc="default">
<t>The following alterations have been made to the RPC-over-RDMA Version One specification. The section numbers below refer to <xref target="RFC5666" pageno="false" format="default"/>. <list style="symbols"><t>Section 2 has been expanded to introduce and explain key RPC, XDR, and RDMA terminology. These terms are now used consistently throughout the specification. </t><t>Section 3 has been re-organized and split into sub-sections to help readers locate specific requirements and definitions. </t><t>Sections 4 and 5 have been combined to improve the organization of this information. </t><t>The specification of the optional Connection Configuration Protocol has been removed from the specification. </t><t>A section consolidating requirements for Upper Layer Bindings has been added. </t><t>A section discussing RPC-over-RDMA protocol extensibility has been added. </t><t>A section specifying conventions for bi-directional RPC operation on RPC-over-RDMA Version One has been added. </t><t>The "Security Considerations" section has been expanded to include a discussion of how RPC-over-RDMA security depends on features of the underlying RDMA transport. A subsection specifying conventions for using RPCSEC_GSS with RPC-over-RDMA Version One has been added. </t></list> </t>
</section>
<section title="Changes To The XDR Definition" toc="default">
<t>The XDR changes described in this section do not alter the over-the-wire message format described in <xref target="RFC5666" pageno="false" format="default"/>. Changes made to the XDR which do alter the over-the-wire message format (i.e., to make it match actual interoperating implementations) are discussed in <xref target="changes-to-the-protocol" pageno="false" format="default"/>. </t>
<t>These alterations make it easier to extend the RPC-over-RDMA protocol. They also better organize the definition, making the protocol elements more consonant with actual protocol function. The specific changes are: <list style="symbols"><t>The XDR description has been given an extraction script using a sentinel sequence, matching the approach used in <xref target="RFC5662" pageno="false" format="default"/>. </t><t>XDR data types which need to be the same in all RPC-over-RDMA versions have been moved to a separate section and given names that are not version-specific. </t><t>To allow extensions without modification to the existing XDR, the header types previously defined as members of the enum rpcrdma1_proc have been defined as constants, the union rpcrdma1_body was deleted, and RDMA_ERR_CHUNK has been renamed as RDMA_ERR_BADHEADER. </t></list> </t>
</section>
<section title="Changes To The Protocol" anchor="changes-to-the-protocol" toc="default">
<t>Although the protocol described herein interoperates with existing implementations of <xref target="RFC5666" pageno="false" format="default"/>, the following changes have been made relative to the protocol described in that document: <list style="symbols"><t>Support for the Read-Read transfer model has been removed. Read-Read is a slower transfer model than Read-Write, thus implementers have chosen not to support it. Removal simplifies explanatory text, and support for the RDMA_DONE procedure is no longer necessary. </t><t>The specification of RDMA_MSGP in <xref target="RFC5666" pageno="false" format="default"/> and current implementations of it are incomplete. Even if completed, benefit for protocols such as NFSv4.0 <xref target="RFC7530" pageno="false" format="default"/> is doubtful. Therefore the RDMA_MSGP message type is no longer supported. </t><t>Technical errors with regard to handling RPC-over-RDMA header errors have been corrected. </t><t>Specific requirements related to handling XDR round-up and complex XDR data types have been added. </t><t>Explicit guidance is provided for sizing Write chunks, managing multiple chunks in the Write list, and handling unused Write chunks. </t><t>Clear guidance about Send and Receive buffer size has been added. This enables better decisions about when to provide and use the Reply chunk. </t></list> The protocol version number has not been changed because the protocol specified in this document fully interoperates with implementations of the RPC-over-RDMA Version One protocol specified in <xref target="RFC5666" pageno="false" format="default"/>. </t>
</section>
</section>
<section title="Terminology" toc="default">
<section title="Remote Procedure Calls" toc="default">
<t>This section introduces key elements of the Remote Procedure Call <xref target="RFC5531" pageno="false" format="default"/> and External Data Representation <xref target="RFC4506" pageno="false" format="default"/> protocols, upon which RPC-over-RDMA Version One is constructed. </t>
<section title="Upper Layer Protocols" toc="default">
<t>Remote Procedure Calls are an abstraction used to implement the operations of an "Upper Layer Protocol," or ULP. The term Upper Layer Protocol refers to an RPC Program and Version tuple, which is a versioned set of procedure calls that comprise a single well-defined API. One example of an Upper Layer Protocol is the Network File System Version 4.0 <xref target="RFC7530" pageno="false" format="default"/>. </t>
</section>
<section title="Requesters And Responders" toc="default">
<t>Like a local procedure call, every Remote Procedure Call (RPC) has a set of "arguments" and a set of "results". A calling context is not allowed to proceed until the procedure's results are available to it. Unlike a local procedure call, the called procedure is executed remotely rather than in the local application's context. </t>
<t>The RPC protocol as described in <xref target="RFC5531" pageno="false" format="default"/> is fundamentally a message-passing protocol between one server and one or more clients. ONC RPC transactions are made up of two types of messages: <list style="hanging"><t hangText="CALL Message"><vspace blankLines="0"/> A CALL message, or "Call", requests that work be done. A Call is designated by the value zero (0) in the message's msg_type field. An arbitrary unique value is placed in the message's xid field in order to match this CALL message to a corresponding REPLY message. </t><t hangText="REPLY Message"><vspace blankLines="0"/> A REPLY message, or "Reply", reports the results of work requested by a Call. A Reply is designated by the value one (1) in the message's msg_type field. The value contained in the message's xid field is copied from the Call whose results are being reported. </t></list> </t>
<t>The RPC client endpoint, or "requester", serializes an RPC Call's arguments and conveys them to a server endpoint via an RPC Call message. This message contains an RPC protocol header, a header describing the requested upper layer operation, and all arguments. </t>
<t>The RPC server endpoint, or "responder", deserializes the arguments and processes the requested operation. It then serializes the operation's results into another byte stream. This byte stream is conveyed back to the requester via an RPC Reply message. This message contains an RPC protocol header, a header describing the upper layer reply, and all results. </t>
<t>The requester deserializes the results and allows the original caller to proceed. At this point the RPC transaction designated by the xid in the Call message is complete, and the xid is retired. </t>
</section>
<section title="RPC Transports" toc="default">
<t>The role of an "RPC transport" is to mediate the exchange of RPC messages between requesters and responders. An RPC transport bridges the gap between the RPC message abstraction and the native operations of a particular network transport. </t>
<t>RPC-over-RDMA is a connection-oriented RPC transport. When a connection-oriented transport is used, requesters initiate transport connections, while responders wait passively for incoming connection requests. </t>
</section>
<section title="External Data Representation" toc="default">
<t>One cannot assume that all requesters and responders internally represent data objects the same way. RPC uses eXternal Data Representation, or XDR, to translate data types and serialize arguments and results <xref target="RFC4506" pageno="false" format="default"/>. </t>
<t>The XDR protocol encodes data independent of the endianness or size of host-native data types, allowing unambiguous decoding of data on the receiving end. RPC Programs are specified by writing an XDR definition of their procedures, argument data types, and result data types. </t>
<t>XDR assumes that the number of bits in a byte (octet) and their order are the same on both endpoints and on the physical network. The smallest indivisible unit of XDR encoding is a group of four octets in little-endian order. XDR also flattens lists, arrays, and other complex data types so they can be conveyed as a stream of bytes. </t>
<t>A serialized stream of bytes that is the result of XDR encoding is referred to as an "XDR stream." A sending endpoint encodes native data into an XDR stream and then transmits that stream to a receiver. A receiving endpoint decodes incoming XDR byte streams into its native data representation format. </t>
<section title="XDR Opaque Data" toc="default">
<t>Sometimes a data item must be transferred as-is, without encoding or decoding. Such a data item is referred to as "opaque data." XDR encoding places opaque data items directly into an XDR stream without altering their content in any way. Upper Layer Protocols or applications perform any needed data translation in this case. Examples of opaque data items include the content of files, or generic byte strings. </t>
</section>
<section title="XDR Round-up" toc="default">
<t>The number of octets in a variable-size opaque data item precedes that item in an XDR stream. If the size of an encoded data item is not a multiple of four octets, octets containing zero are added to the end of the item as it is encoded so that the next encoded data item starts on a four-octet boundary. The encoded size of the item is not changed by the addition of the extra octets, and the zero bytes are not exposed to the Upper Layer. </t>
<t>This technique is referred to as "XDR round-up," and the extra octets are referred to as "XDR padding". </t>
</section>
</section>
</section>
<section title="Remote Direct Memory Access" toc="default">
<t>RPC requesters and responders can be made more efficient if large RPC messages are transferred by a third party such as intelligent network interface hardware (data movement offload), and placed in the receiver's memory so that no additional adjustment of data alignment has to be made (direct data placement). Remote Direct Memory Access enables both optimizations. </t>
<section title="Direct Data Placement" toc="default">
<t>Typically, RPC implementations copy the contents of RPC messages into a buffer before being sent. An efficient RPC implementation sends bulk data without copying it into a separate send buffer first. </t>
<t>However, socket-based RPC implementations are often unable to receive data directly into its final place in memory. Receivers often need to copy incoming data to finish an RPC operation; sometimes, only to adjust data alignment. </t>
<t>In this document, "RDMA" refers to the physical mechanism an RDMA transport utilizes when moving data. Although this may not be efficient, before an RDMA transfer a sender may copy data into an intermediate buffer before an RDMA transfer. After an RDMA transfer, a receiver may copy that data again to its final destination. </t>
<t>This document uses the term "direct data placement" (or DDP) to refer specifically to an optimized data transfer where it is unnecessary for a receiving host's CPU to copy transferred data to another location after it has been received. Not all RDMA-based data transfer qualifies as Direct Data Placement, and DDP can be achieved using non-RDMA mechanisms. </t>
</section>
<section title="RDMA Transport Requirements" anchor="rdma-transport-requirements" toc="default">
<t>The RPC-over-RDMA Version One protocol assumes the physical transport provides the following abstract operations. A more complete discussion of these operations is found in <xref target="RFC5040" pageno="false" format="default"/>. <list style="hanging"><t hangText="Registered Memory"><vspace blankLines="0"/> Registered memory is a segment of memory that is assigned a steering tag that temporarily permits access by the RDMA provider to perform data transfer operations. The RPC-over-RDMA Version One protocol assumes that each segment of registered memory MUST be identified with a steering tag of no more than 32 bits and memory addresses of up to 64 bits in length. </t><t hangText="RDMA Send"><vspace blankLines="0"/> The RDMA provider supports an RDMA Send operation, with completion signaled on the receiving peer after data has been placed in a pre-posted memory segment. Sends complete at the receiver in the order they were issued at the sender. The amount of data transferred by an RDMA Send operation is limited by the size of the remote pre-posted memory segment. </t><t hangText="RDMA Receive"><vspace blankLines="0"/> The RDMA provider supports an RDMA Receive operation to receive data conveyed by incoming RDMA Send operations. To reduce the amount of memory that must remain pinned awaiting incoming Sends, the amount of pre-posted memory is limited. Flow-control to prevent overrunning receiver resources is provided by the RDMA consumer (in this case, the RPC-over-RDMA Version One protocol). </t><t hangText="RDMA Write"><vspace blankLines="0"/> The RDMA provider supports an RDMA Write operation to directly place data in remote memory. The local host initiates an RDMA Write, and completion is signaled there. No completion is signaled on the remote. The local host provides a steering tag, memory address, and length of the remote's memory segment. <vspace blankLines="1"/> RDMA Writes are not necessarily ordered with respect to one another, but are ordered with respect to RDMA Sends. A subsequent RDMA Send completion obtained at the write initiator guarantees that prior RDMA Write data has been successfully placed in the remote peer's memory. </t><t hangText="RDMA Read"><vspace blankLines="0"/> The RDMA provider supports an RDMA Read operation to directly place peer source data in the read initiator's memory. The local host initiates an RDMA Read, and completion is signaled there; no completion is signaled on the remote. The local host provides steering tags, memory addresses, and a length for the remote source and local destination memory segments. <vspace blankLines="1"/> The remote peer receives no notification of RDMA Read completion. The local host signals completion as part of a subsequent RDMA Send message so that the remote peer can release steering tags and subsequently free associated source memory segments. </t></list> </t>
<t>The RPC-over-RDMA Version One protocol is designed to be carried over RDMA transports that support the above abstract operations. This protocol conveys to the RPC peer information sufficient for that RPC peer to direct an RDMA layer to perform transfers containing RPC data and to communicate their result(s). For example, it is readily carried over RDMA transports such as Internet Wide Area RDMA Protocol (iWARP) <xref target="RFC5040" pageno="false" format="default"/> <xref target="RFC5041" pageno="false" format="default"/>. </t>
</section>
</section>
</section>
<section title="RPC-Over-RDMA Protocol Framework" toc="default">
<section title="Transfer Models" toc="default">
<t>A "transfer model" designates which endpoint is responsible for performing RDMA Read and Write operations. To enable these operations, the peer endpoint first exposes segments of its memory to the endpoint performing the RDMA Read and Write operations. <list style="hanging"><t hangText="Read-Read"><vspace blankLines="0"/> Requesters expose their memory to the responder, and the responder exposes its memory to requesters. The responder employs RDMA Read operations to pull RPC arguments or whole RPC calls from the requester. Requesters employ RDMA Read operations to pull RPC results or whole RPC relies from the responder. </t><t hangText="Write-Write"><vspace blankLines="0"/> Requesters expose their memory to the responder, and the responder exposes its memory to requesters. Requesters employ RDMA Write operations to push RPC arguments or whole RPC calls to the responder. The responder employs RDMA Write operations to push RPC results or whole RPC relies to the requester. </t><t hangText="Read-Write"><vspace blankLines="0"/> Requesters expose their memory to the responder, but the responder does not expose its memory. The responder employs RDMA Read operations to pull RPC arguments or whole RPC calls from the requester. The responder employs RDMA Write operations to push RPC results or whole RPC relies to the requester. </t><t hangText="Write-Read"><vspace blankLines="0"/> The responder exposes its memory to requesters, but requesters do not expose their memory. Requesters employ RDMA Write operations to push RPC arguments or whole RPC calls to the responder. Requesters employ RDMA Read operations to pull RPC results or whole RPC relies from the responder. </t></list> <xref target="RFC5666" pageno="false" format="default"/> specifies the use of both the Read-Read and the Read-Write Transfer Model. All current RPC-over-RDMA Version One implementations use only the Read-Write Transfer Model. Therefore the use of the Read-Read Transfer Model by RPC-over-RDMA Version One implementations is no longer supported. Other Transfer Models may be used by a future version of RPC-over-RDMA. </t>
</section>
<section title="Message Framing" toc="default">
<t>On an RPC-over-RDMA transport, each RPC message is encapsulated by an RPC-over-RDMA message. An RPC-over-RDMA message consists of two XDR streams. <list style="hanging"><t hangText="RPC Payload Stream"><vspace blankLines="0"/> The "Payload stream" contains the encapsulated RPC message being transferred by this RPC-over-RDMA message. This stream always begins with the XID field of the encapsulated RPC message. </t><t hangText="Transport Header Stream"><vspace blankLines="0"/> The "Transport stream" contains a header that describes and controls the transfer of the Payload stream in this RPC-over-RDMA message. This header is analogous to the record marking used for RPC over TCP but is more extensive, since RDMA transports support several modes of data transfer. </t></list> </t>
<t>In its simplest form, an RPC-over-RDMA message consists of a Transport stream followed immediately by a Payload stream conveyed together in a single RDMA Send. To transmit large RPC messages, a combination of one RDMA Send operation and one or more RDMA Read or Write operations is employed. </t>
<t>RPC-over-RDMA framing replaces all other RPC framing (such as TCP record marking) when used atop an RPC-over-RDMA association, even when the underlying RDMA protocol may itself be layered atop a transport with a defined RPC framing (such as TCP). </t>
<t>It is however possible for RPC-over-RDMA to be dynamically enabled in the course of negotiating the use of RDMA via an Upper Layer Protocol exchange. Because RPC framing delimits an entire RPC request or reply, the resulting shift in framing must occur between distinct RPC messages, and in concert with the underlying transport. </t>
</section>
<section title="Managing Receiver Resources" anchor="managing-receiver-resources" toc="default">
<t>It is critical to provide RDMA Send flow control for an RDMA connection. If no pre-posted receive buffer is large enough to accept an incoming RDMA Send, the RDMA Send operation fails. If a pre-posted receive buffer is not available to accept an incoming RDMA Send, the RDMA Send operation can fail. Repeated occurrences of such errors can be fatal to the connection. This is a departure from conventional TCP/IP networking where buffers are allocated dynamically as part of receiving messages. </t>
<t>The longevity of an RDMA connection requires that sending endpoints respect the resource limits of peer receivers. To ensure messages can be sent and received reliably, there are two operational parameters for each connection. </t>
<section title="Credit Limit" toc="default">
<t>The number of pre-posted RDMA Receive operations is sometimes referred to as a peer's "credit limit." Flow control for RDMA Send operations directed to the responder is implemented as a simple request/grant protocol in the RPC-over-RDMA header associated with each RPC message. <xref target="credit-value" pageno="false" format="default"/> has further detail. <list style="symbols"><t>The RPC-over-RDMA header for RPC Call messages contains a requested credit value for the responder. This is the maximum number of RPC replies the requester can handle at once, independent of how many RPCs are in flight at that moment. The requester MAY dynamically adjust the requested credit value to match its expected needs. </t><t>The RPC-over-RDMA header for RPC Reply messages provides the granted result. This is the maximum number of RPC calls the responder can handle at once, without regard to how many RPCs are in flight at that moment. The granted value MUST NOT be zero, since such a value would result in deadlock. The responder MAY dynamically adjust the granted credit value to match its needs or policies (e.g. to accommodate the available resources in a shared receive queue). </t></list> </t>
<t>The requester MUST NOT send unacknowledged requests in excess of this granted responder credit limit. If the limit is exceeded, the RDMA layer may signal an error, possibly terminating the connection. If an RDMA layer error does not occur, the responder MAY handle excess requests or return an RPC layer error to the requester. </t>
<t>While RPC calls complete in any order, the current flow control limit at the responder is known to the requester from the Send ordering properties. It is always the lower of the requested and granted credit values, minus the number of requests in flight. Advertised credit values are not altered when individual RPCs are started or completed. </t>
<t>On occasion a requester or responder may need to adjust the amount of resources available to a connection. When this happens, the responder needs to ensure that a credit increase is effected (i.e. RDMA Receives are posted) before the next reply is sent. </t>
<t>Certain RDMA implementations may impose additional flow control restrictions, such as limits on RDMA Read operations in progress at the responder. Accommodation of such restrictions is considered the responsibility of each RPC-over-RDMA Version One implementation. </t>
</section>
<section title="Inline Threshold" toc="default">
<t>A receiver's "inline threshold" value is the largest message size (in octets) that the receiver can accept via an RDMA Receive operation. Each connection has two inline threshold values, one for each peer receiver. </t>
<t>Unlike credit limits, inline threshold values are not advertised to peers via the RPC-over-RDMA Version One protocol, and there is no provision for the inline threshold value to change during the lifetime of an RPC-over-RDMA Version One connection. </t>
</section>
<section title="Initial Connection State" anchor="initial-connection-state" toc="default">
<t>When a connection is first established, peers might not know how many receive buffers the other has, nor how large these buffers are. </t>
<t>As a basis for an initial exchange of RPC requests, each RPC-over-RDMA Version One connection provides the ability to exchange at least one RPC message at a time that is 1024 bytes in size. A responder MAY exceed this basic level of configuration, but a requester MUST NOT assume more than one credit is available, and MUST receive a valid reply from the responder carrying the actual number of available credits, prior to sending its next request. </t>
<t>Receiver implementations MUST support an inline threshold of 1024 bytes, but MAY support larger inline thresholds values. A mechanism for discovering a peer's inline threshold value before a connection is established may be used to optimize the use of RDMA Send operations. In the absense of such a mechanism, senders MUST assume a receiver's inline threshold is 1024 bytes. </t>
</section>
</section>
<section title="XDR Encoding With Chunks" toc="default">
<t>When a direct data placement capability is available, during XDR encoding it can be determined that an XDR data item is large enough that it might be more efficient if the transport placed the content of the data item directly in the receiver's memory. </t>
<section title="Reducing An XDR Stream" toc="default">
<t>RPC-over-RDMA Version One provides a mechanism for moving part of an RPC message via a data transfer separate from an RDMA Send/Receive. The sender removes one or more XDR data items from the Payload stream. They are conveyed via one or more RDMA Read or Write operations. The receiver inserts the data items into the Payload stream before passing it to the Upper Layer. </t>
<t>A contiguous piece of a Payload stream that is split out and moved via separate RDMA operations is known as a "chunk." A Payload stream after chunks have been removed is referred to as a "reduced" Payload stream. </t>
</section>
<section title="DDP-Eligibility" toc="default">
<t>Only an XDR data item that might benefit from Direct Data Placement may be reduced. The eligibility of particular XDR data items to be reduced is not specified by this document. </t>
<t>To maintain interoperability on an RPC-over-RDMA transport, a determination must be made of which XDR data items in each Upper Layer Protocol are allowed to use Direct Data Placement. Therefore an additional specification is needed that describes how an Upper Layer Protocol enables Direct Data Placement. The set of requirements for an Upper Layer Protocol to use an RPC-over-RDMA transport is known as an "Upper Layer Binding specification," or ULB. </t>
<t>An Upper Layer Binding specification states which specific individual XDR data items in an Upper Layer Protocol MAY be transferred via Direct Data Placement. This document will refer to XDR data items that are permitted to be reduced as "DDP-eligible". All other XDR data items MUST NOT be reduced. RPC-over-RDMA Version One uses RDMA Read and Write operations to transfer DDP-eligible data that has been reduced. </t>
<t>Detailed requirements for Upper Layer Bindings are discussed in full in <xref target="upper-layer-binding-specifications" pageno="false" format="default"/>. </t>
</section>
<section title="RDMA Segments" toc="default">
<t>When encoding a Payload stream that contains a DDP-eligible data item, a sender may choose to reduce that data item. It does not place the item into the Payload stream. Instead, the sender records in the RPC-over-RDMA header the actual address and size of the memory region containing that data item. </t>
<t>The requester provides location information for DDP-eligible data items in both RPC Calls and Replies. The responder uses this information to initiate RDMA Read and Write operations to retrieve or update the specified region of the requester's memory. </t>
<t>An "RDMA segment", or a "plain segment", is an RPC-over-RDMA header data object that contains the precise co-ordinates of a contiguous memory region that is to be conveyed via one or more RDMA Read or RDMA Write operations. The following fields are contained in each segment. <figure title="" suppress-title="false" align="left" alt="" width="" height=""><artwork xml:space="preserve" name="" type="" align="left" alt="" width="" height="">
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Handle |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Length |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| |
+ Offset +
| |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
</artwork></figure> </t>
<t><list style="hanging"><t hangText="Handle"><vspace blankLines="0"/> Steering tag (STag) or handle obtained when the segment's memory is registered for RDMA. Also known as an R_key, this value is generated by registering this memory with the RDMA provider. </t><t hangText="Length"><vspace blankLines="0"/> The length of the memory segment, in octets. </t><t hangText="Offset"><vspace blankLines="0"/> The offset or beginning memory address of the segment. </t></list> See <xref target="RFC5040" pageno="false" format="default"/> for further discussion of the meaning of these fields. </t>
</section>
<section title="Chunks" toc="default">
<t>In RPC-over-RDMA Version One, a "chunk" refers to a portion of the Payload stream that is moved via RDMA Read or Write operations. Chunk data is removed from the sender's Payload stream, transferred by separate RDMA operations, and then re-inserted into the receiver's Payload stream. </t>
<t>Each chunk consists of one or more RDMA segments. Each segment represents a single contiguous piece of that chunk. Segments MAY divide a chunk on any boundary that is convenient to the requester. </t>
<t>Except in special cases, a chunk contains exactly one XDR data item. This makes it straightforward to remove chunks from an XDR stream without affecting XDR alignment. Not every RPC-over-RDMA message has chunks associated with it. </t>
<section title="Counted Arrays" toc="default">
<t>If a chunk contains a counted array data type, the count of array elements MUST remain in the Payload stream, while the array elements MUST be moved to the chunk. For example, when encoding an opaque byte array as a chunk, the count of bytes stays in the Payload stream, while the bytes in the array are removed from the Payload stream and transferred within the chunk. </t>
<t>Any byte count left in the Payload stream MUST match the sum of the lengths of the segments making up the chunk. If they do not agree, an RPC protocol encoding error results. </t>
<t>Individual array elements appear in a chunk in their entirety. For example, when encoding an array of arrays as a chunk, the count of items in the enclosing array stays in the Payload stream, but each enclosed array, including its item count, is transferred as part of the chunk. </t>
</section>
<section title="Optional-data" toc="default">
<t>If a chunk contains an optional-data data type, the "is present" field MUST remain in the Payload stream, while the data, if present, MUST be moved to the chunk. </t>
</section>
<section title="XDR Unions" toc="default">
<t>A union data type should never be made DDP-eligible, but one or more of its arms may be DDP-eligible. </t>
</section>
</section>
<section title="Read Chunks" toc="default">
<t>A "Read chunk" represents an XDR data item that is to be pulled from the requester to the responder using RDMA Read operations. </t>
<t>A Read chunk is a list of one or more RDMA segments. Each RDMA segment in a Read chunk is a plain segment which has an additional Position field. <figure title="" suppress-title="false" align="left" alt="" width="" height=""><artwork xml:space="preserve" name="" type="" align="left" alt="" width="" height="">
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Position |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Handle |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Length |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| |
+ Offset +
| |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
</artwork></figure> </t>
<t><list style="hanging"><t hangText="Position"><vspace blankLines="0"/> The byte offset in the Payload stream where the receiver re-inserts the data item conveyed in a chunk. The Position value MUST be computed from the beginning of the Payload stream, which begins at Position zero. All RDMA segments belonging to the same Read chunk have the same value in their Position field. </t></list> </t>
<t>While constructing an RPC-over-RDMA Call message, a requester registers memory segments containing data in Read chunks. It advertises these chunks in the RPC-over-RDMA header of the RPC Call. </t>
<t>After receiving an RPC Call sent via an RDMA Send operation, a responder transfers the chunk data from the requester using RDMA Read operations. The responder reconstructs the transferred chunk data by concatenating the contents of each segment, in list order, into the received Payload stream at the Position value recorded in the segment. </t>
<t>Put another way, a receiver inserts the first segment in a Read chunk into the Payload stream at the byte offset indicated by its Position field. Segments whose Position field value match this offset are concatenated afterwards, until there are no more segments at that Position value. The next XDR data item in the Payload stream follows. </t>
<section title="Read Chunk Round-up" toc="default">
<t>XDR requires each encoded data item to start on four-byte alignment. When an odd-length data item is encoded, its length is encoded literally, while the data is padded so the next data item in the XDR stream can start on a four-byte boundary. Receivers ignore the content of the pad bytes. </t>
<t>After an XDR data item has been reduced, all data items remaining in the Payload stream must continue to adhere to these padding requirements. Thus when an XDR data item is moved from the Payload stream into a Read chunk, the requester MUST remove XDR padding for that data item from the Payload stream as well. </t>
<t>The length of a Read chunk is the sum of the lengths of the read segments that comprise it. If this sum is not a multiple of four, the requester MAY choose to send a Read chunk without any XDR padding. If the requester provides no actual round-up in a Read chunk, the responder MUST be prepared to provide appropriate round-up in the reconstructed call XDR stream </t>
<t>The Position field in a read segment indicates where the containing Read chunk starts in the Payload stream. The value in this field MUST be a multiple of four. Moreover, all segments in the same Read chunk share the same Position value, even if one or more of the segments have a non-four-byte aligned length. </t>
</section>
<section title="Decoding Read Chunks" toc="default">
<t>While decoding a received Payload stream, whenever the XDR offset in the Payload stream matches that of a Read chunk, the transport initiates an RDMA Read to pull the chunk's data content into registered memory on the responder. </t>
<t>The responder acknowledges its completion of use of Read chunk source buffers when it sends an RPC Reply to the requester. The requester may then release Read chunks advertised in the request. </t>
</section>
</section>
<section title="Write Chunks" toc="default">
<t>A "Write chunk" represents an XDR data item that is to be pushed from a responder to a requester using RDMA Write operations. </t>
<t>A Write chunk is an array of one or more plain RDMA segments. Write chunks are provided by a requester long before the responder has prepared the reply Payload stream. Therefore RDMA segments in a Write chunk do not have a Position field. </t>
<t>While constructing an RPC Call message, a requester also prepares memory regions to catch DDP-eligible reply data items. A requester does not know the actual length of the result data item to be returned, thus it MUST register a Write chunk long enough to accommodate the maximum possible size of the returned data item. </t>
<t>A responder copies the requester-provided Write chunk segments into the RPC-over-RDMA header that it returns with the reply. The responder MUST NOT change the number of segments in the Write chunk. </t>
<t>The responder fills the segments in array order until the data item has been completely written. The responder updates the segment length fields to reflect the actual amount of data that is being returned in each segment. If a Write chunk segment is not filled by the responder, the updated length of the segment SHOULD be zero. </t>
<t>The responder then sends the RPC Reply via an RDMA Send operation. After receiving the RPC Reply, the requester reconstructs the transferred data by concatenating the contents of each segment, in array order, into RPC Reply XDR stream. </t>
<section title="Write Chunk Round-up" toc="default">
<t>XDR requires each encoded data item to start on four-byte alignment. When an odd-length data item is encoded, its length is encoded literally, while the data is padded so the next data item in the XDR stream can start on a four-byte boundary. Receivers ignore the content of the pad bytes. </t>
<t>After a data item is reduced, data items remaining in the Payload stream must continue to adhere to these padding requirements. Thus when an XDR data item is moved from a reply Payload stream into a Write chunk, the responder MUST remove XDR padding for that data item from the reply Payload stream as well. </t>
<t>A requester SHOULD NOT provide extra length in a Write chunk to accommodate XDR pad bytes. A responder MUST NOT write XDR pad bytes for a Write chunk. </t>
</section>
<section title="Unused Write Chunks" toc="default">
<t>There are occasions when a requester provides a Write chunk but the responder does not use it. </t>
<t>For example, an Upper Layer Protocol may define a union result where some arms of the union contain a DDP-eligible data item while other arms do not. The requester is required to provide a Write chunk in this case, but if the responder returns a result that uses an arm of the union that has no DDP-eligible data item, the Write chunk remains unused. </t>
<t>When forming an RPC-over-RDMA Reply message with an unused Write chunk, the responder MUST set the length of all segments in the chunk to zero. </t>
<t>Unused write chunks, or unused bytes in write chunk segments, are not returned as results. Their memory is returned to the Upper Layer as part of RPC completion. However, the RPC layer MUST NOT assume that the buffers have not been modified. </t>
</section>
</section>
</section>
<section title="Message Size" toc="default">
<t>A receiver of RDMA Send operations is required by RDMA to have previously posted one or more adequately sized buffers. Memory savings can be achieved on both requesters and responders by leaving the inline threshold small. However, not all RPC messages are small. </t>
<section title="Short Messages" toc="default">
<t>RPC messages are frequently smaller than typical inline thresholds. For example, the NFS version 3 GETATTR request is only 56 bytes: 20 bytes of RPC header, plus a 32-byte file handle argument and 4 bytes for its length. The reply to this common request is about 100 bytes. </t>
<t>Since all RPC messages conveyed via RPC-over-RDMA require an RDMA Send operation, the most efficient way to send an RPC message that is smaller than the receiver's inline threshold is to append the Payload stream directly to the Transport stream. An RPC-over-RDMA header with a small RPC Call or Reply message immediately following is transferred using a single RDMA Send operation. No RDMA Read or Write operations are needed. </t>
</section>
<section title="Chunked Messages" toc="default">
<t>If DDP-eligible data items are present in a Payload stream, a sender MAY reduce the Payload stream to enable the use of RDMA Read or Write operations to move the reduced data items. The Transport stream with the reduced Payload stream immediately following is transferred using a single RDMA Send operation. </t>
<t>After receiving the Transport and Payload streams of a Chunked RPC-over-RDMA Call message, the responder uses RDMA Read operations to move reduced data items in Read chunks. Before sending the Transport and Payload streams of a Chunked RPC-over-RDMA Reply message, the responder uses RDMA Write operations to move reduced data items in Write and Reply chunks. </t>
</section>
<section title="Long Messages" toc="default">
<t>When a Payload stream is larger than the receiver's inline threshold, the Payload stream is reduced by removing DDP-eligible data items and placing them in chunks to be moved separately. If there are no DDP-eligible data items in the Payload stream, or the Payload stream is still too large after it has been reduced, the RDMA transport MUST use RDMA Read or Write operations to convey the Payload stream itself. This mechanism is referred to as a "Long Message." </t>
<t>To transmit a Long Message, the sender conveys only the Transport stream with an RDMA Send operation. The Payload stream is not included in the Send buffer in this instance. Instead, the requester provides chunks that the responder uses to move the Payload stream. <list style="hanging"><t hangText="Long RPC Call"><vspace blankLines="0"/> To send a Long RPC-over-RDMA Call message, the requester provides a special Read chunk that contains the RPC Call's Payload stream. Every segment in this Read chunk MUST contain zero in its Position field. Thus this chunk is known as a "Position Zero Read chunk." </t><t hangText="Long RPC Reply"><vspace blankLines="0"/> To send a Long RPC-over-RDMA Reply message, the requester provides a single special Write chunk in advance, known as the "Reply chunk", that will contain the RPC Reply's Payload stream. The requester sizes the Reply chunk to accommodate the maximum expected reply size for that Upper Layer operation. </t></list> Though the purpose of a Long Message is to handle large RPC messages, requesters MAY use a Long Message at any time to convey an RPC Call. </t>
<t>A responder chooses which form of reply to use based on the chunks provided by the requester. If Write chunks were provided and the responder has a DDP-eligible result, it first reduces the reply Payload stream. If a Reply chunk was provided and the reduced Payload is larger than the requester's inline threshold, the responder MUST use the provided Reply chunk for the reply. </t>
<t>Because these special chunks contain a whole RPC message, any XDR data item MAY appear in one of these special chunks without regard to its DDP-eligibility. DDP-eligible data items MAY be removed from these special chunks and conveyed via normal chunks, but non-eligible data items MUST NOT appear in normal chunks. </t>
</section>
</section>
</section>
<section title="RPC-Over-RDMA In Operation" toc="default">
<t>Every RPC-over-RDMA Version One message has a header that includes a copy of the message's transaction ID, data for managing RDMA flow control credits, and lists of RDMA segments used for RDMA Read and Write operations. All RPC-over-RDMA header content is contained in the Transport stream, and thus MUST be XDR encoded. </t>
<t>RPC message layout is unchanged from that described in <xref target="RFC5531" pageno="false" format="default"/> except for the possible reduction of data items that are moved by RDMA Read or Write operations. </t>
<t>The RPC-over-RDMA protocol passes RPC messages without regard to their type (CALL or REPLY) or direction (forwards or backwards). Both endpoints of a connection MAY send any RPC-over-RDMA message header type at any time (subject to credit limits). </t>
<section title="XDR Protocol Definition" toc="default">
<t>This section contains a description of the core features of the RPC-over-RDMA Version One protocol, expressed in the XDR language <xref target="RFC4506" pageno="false" format="default"/>. </t>
<t>This description is provided in a way that makes it simple to extract into ready-to-compile form. The reader can apply the following shell script to this document to produce a machine-readable XDR description of the RPC-over-RDMA Version One protocol without any OPTIONAL extensions. </t>
<figure title="" suppress-title="false" align="left" alt="" width="" height="">
<artwork xml:space="preserve" name="" type="" align="left" alt="" width="" height="">
<CODE BEGINS>
#!/bin/sh
grep '^ *///' | sed 's?^ /// ??' | sed 's?^ *///$??'
<CODE ENDS>
</artwork>
</figure>
<t>That is, if the above script is stored in a file called "extract.sh" and this document is in a file called "spec.txt" then the reader can do the following to extract an XDR description file: </t>
<figure title="" suppress-title="false" align="left" alt="" width="" height="">
<artwork xml:space="preserve" name="" type="" align="left" alt="" width="" height="">
<CODE BEGINS>
sh extract.sh < spec.txt > rpcrdma_corev1.x
<CODE ENDS>
</artwork>
</figure>
<t>As described in <xref target="rpc-over-rdma-version-one-extension-practices" pageno="false" format="default"/>, extensions to RPC-over-RDMA Version One, published as Proposed Standards, will have similar means of providing an XDR description appropriate to those extensions. Once XDR for extensions is also extracted, it can be appended to the XDR description file extracted from this document to produce a consolidated XDR description file reflecting all extensions selected for an RPC-over-RDMA implementation. </t>
<t>RPC-over-RDMA is not a stand-alone RPC Program. To enable protocol extension, there is no single XDR entity which describes the format of RPC-over-RDMA headers. Instead, implementers need to follow the instructions in <xref target="use-of-xdr-descriptions" pageno="false" format="default"/> to appropriately encode and decode protocol messages. </t>
<section title="Code Component License" toc="default">
<t>Code components extracted from this document must include the following license text. When the extracted XDR code is combined with other complementary XDR code which itself has an identical license, only a single copy of the license text need be preserved. <figure title="" suppress-title="false" align="left" alt="" width="" height=""><artwork xml:space="preserve" name="" type="" align="left" alt="" width="" height="">
<CODE BEGINS>
/// /*
/// * Copyright (c) 2010, 2015 IETF Trust and the persons
/// * identified as authors of the code. All rights reserved.
/// *
/// * The authors of the code are:
/// * B. Callaghan, T. Talpey, C. Lever, and D. Noveck.
/// *
/// * Redistribution and use in source and binary forms, with
/// * or without modification, are permitted provided that the
/// * following conditions are met:
/// *
/// * - Redistributions of source code must retain the above
/// * copyright notice, this list of conditions and the
/// * following disclaimer.
/// *
/// * - Redistributions in binary form must reproduce the above
/// * copyright notice, this list of conditions and the
/// * following disclaimer in the documentation and/or other
/// * materials provided with the distribution.
/// *
/// * - Neither the name of Internet Society, IETF or IETF
/// * Trust, nor the names of specific contributors, may be
/// * used to endorse or promote products derived from this
/// * software without specific prior written permission.
/// *
/// * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS
/// * AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED
/// * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
/// * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
/// * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
/// * EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
/// * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
/// * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
/// * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
/// * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
/// * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
/// * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
/// * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
/// * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
/// * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
/// */
<CODE ENDS>
</artwork></figure> </t>
</section>
<section title="XDR Applying To All Versions Of RPC-Over-RDMA" toc="default">
<t>XDR data items defined in this section describe elements of the RPC-over-RDMA protocol that are not subject to change in subsequent versions. A full discussion of the extensibility model is in <xref target="protocol-extensibility" pageno="false" format="default"/>. <figure title="" suppress-title="false" align="left" alt="" width="" height=""><artwork xml:space="preserve" name="" type="" align="left" alt="" width="" height="">
<CODE BEGINS>
/// typedef uint32 rdma_htype;
///
/// struct rpcrdma_prefix {
/// uint32 rdma_xid;
/// uint32 rdma_version;
/// uint32 rdma_credits;
/// rpcrdma_htype rdma_htype;
/// };
///
/// /*
/// * Mandatory RPC-over-RDMA message header types
/// */
/// const RDMA_MSG = 0;
/// const RDMA_NOMSG = 1;
/// const RDMA_ERROR = 4;
///
/// struct rpcrdma_err_vers {
/// uint32 rdma_vers_low;
/// uint32 rdma_vers_high;
/// };
<CODE ENDS>
</artwork></figure> </t>
</section>
<section title="XDR Applying To Version One Of RPC-Over-RDMA" toc="default">
<t>XDR data items defined in this section are subject to change in subsequent RPC-over-RDMA versions. </t>
<t>Even though the names of structures and unions begin "rpcrdma1_" these are not restricted to use in RPC-over-RDMA Version One. Structure definitions may be carried over unchanged to subsequence versions, but unions are subject to extension according to the rules for compatible XDR extension as discussed in <xref target="protocol-extensibility" pageno="false" format="default"/>. Comments identify items that cannot be changed in subsequent versions. </t>
<t><figure title="" suppress-title="false" align="left" alt="" width="" height=""><artwork xml:space="preserve" name="" type="" align="left" alt="" width="" height="">
<CODE BEGINS>
/// /*
/// * Version One reserved message types
/// */
/// const RDMA_MSGP = 2;
/// const RDMA_DONE = 3;
///
/// struct rpcrdma1_segment {
/// uint32 rdma_handle;
/// uint32 rdma_length;
/// uint64 rdma_offset;
/// };
///
/// struct rpcrdma1_read_segment {
/// uint32 rdma_position;
/// struct rpcrdma1_segment rdma_target;
/// };
///
/// struct rpcrdma1_read_list {
/// struct rpcrdma1_read_segment rdma_entry;
/// struct rpcrdma1_read_list *rdma_next;
/// };
///
/// struct rpcrdma1_write_chunk {
/// struct rpcrdma1_segment rdma_target<>;
/// };
///
/// struct rpcrdma1_write_list {
/// struct rpcrdma1_write_chunk rdma_entry;
/// struct rpcrdma1_write_list *rdma_next;
/// };
///
/// struct rpcrdma1_chunks {
/// struct rpcrdma1_read_list *rdma_reads;
/// struct rpcrdma1_write_list *rdma_writes;
/// struct rpcrdma1_write_chunk *rdma_reply;
/// };
///
/// struct rpcrdma1_padded {
/// uint32 rdma_align;
/// uint32 rdma_thresh;
/// rpcrdma1_chunks rdma_chunks;
/// };
///
/// enum rpcrdma1_errcode {
/// RDMA_ERR_VERS = 1,
/// RDMA_ERR_BADHEADER = 2
/// };
///
/// union rpcrdma1_error switch (rpcrdma1_errcode rdma_err) {
/// case RDMA_ERR_VERS:
/// rpcrdma_err_vers rdma_vrange; /* Immutable */
/// case RDMA_ERR_BADHEADER:
/// void;
/// };
<CODE ENDS>
</artwork></figure> </t>
</section>
<section title="Use Of XDR Descriptions" anchor="use-of-xdr-descriptions" toc="default">
<t>Though it is described by XDR, RPC-over-RDMA is not an RPC Program. Certain functions normally provided by RPC need to be addressed by the RPC-over-RDMA definition itself. In particular, the following functions normally provided by RPC need to be provided for as part of the RPC-over-RDMA XDR description: <list style="symbols"><t>negotiation of RPC-over-RDMA protocol version </t><t>Identifying RPC-over-RDMA header types that are followed by a Payload stream </t></list> </t>
<t>In <xref target="RFC5666" pageno="false" format="default"/> the XDR description did not take account of the natural layering between the part of RPC-over-RDMA functionality that performed RPC-layer like functions described above and that which implemented individual transport functions. As a result: <list style="symbols"><t>The four 32-bit words which must be the same in all versions of RPC-over-RDMA are split, with three of those words in struct rpcrdma1_header and the remaining word part of union rpcrdma1_body, together with each of the message bodies. </t><t>It is impossible, within the resulting structure, to add a new message type without modifying the existing XDR description. </t></list> </t>
<t>The XDR description introduced in this document reorganizes the XDR in line with this natural layering, while maintaining over-the-wire equivalence. As a result, the 32-bit big-endian field strating twelve bytes into the header is no longer the discriminator field of union rpcrdma1_body. Instead it is the last 32-bit word within struct rpcrdma_header which define the common (i.e., for all RPC-over-RDMA versions) header prefix. It retains its role of indicating the message type and deciding which particular header body is to follow. </t>
<t>As a result there is no longer a single XDR item that encompasses the entire RPC-over-RDMA header. Instead, each RPC-over-RDMA meassage consists of up to three items and those using XDR encode and decode must be aware that they proceed in sequence as follows: <list style="numbers"><t>A struct rpcrdma_prefix </t><t>Depending on the rdma_which field of the prefix, the appropriate header body for that message type as given by Table 1. In cases in which there is an undefined header type, this is to be treated as an XDR encode/decode error. </t><t>If allowed for that header type as defined in Table 1, an XDR stream for the RPC message being transported </t></list> <figure title="" suppress-title="false" align="left" alt="" width="" height=""><artwork xml:space="preserve" name="" type="" align="left" alt="" width="" height="">
+--------------+------------------------+-------------------+
| Message Type | Body | Payload stream? |
+--------------+------------------------+-------------------+
| RDMA_MSG | struct rpcrdma1_chunks | Yes |
+--------------+------------------------+-------------------+
| RDMA_NOMSG | struct rpcrdma1_chunks | No |
+--------------+------------------------+-------------------+
| RDMA_ERROR | union rpcrdma1_error | No |
+--------------+------------------------+-------------------+
</artwork><postamble>Table 1. Header Type Characteristics</postamble></figure> </t>
</section>
</section>
<section title="Fixed Header Fields" toc="default">
<t>The RPC-over-RDMA header begins with four fixed 32-bit fields that control the RDMA interaction. These four fields, which must remain with the same meanings and in the same positions in all subsequent versions of the RPC-over-RDMA protocol, are described below. <figure title="" suppress-title="false" align="left" alt="" width="" height=""><artwork xml:space="preserve" name="" type="" align="left" alt="" width="" height="">
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| XID |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Version Number |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Credit Value |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Procedure Number |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
</artwork></figure> </t>
<section title="Transaction ID (XID)" toc="default">
<t>The XID generated for the RPC Call and Reply. Having the XID at a fixed location in the header makes it easy for the receiver to establish context as soon as each RPC-over-RDMA message arrives. This XID MUST be the same as the XID in the RPC message. The receiver MAY perform its processing based solely on the XID in the RPC-over-RDMA header, and thereby ignore the XID in the RPC message, if it so chooses. </t>
</section>
<section title="Version Number" toc="default">
<t>For RPC-over-RDMA Version One, this field MUST contain the value one (1). Rules regarding changes to this transport protocol version number can be found in <xref target="rpc-over-rdma-version-numbering" pageno="false" format="default"/>. </t>
</section>
<section title="Credit Value" anchor="credit-value" toc="default">
<t>When sent in an RPC Call message, the requested credit value is provided. When sent in an RPC Reply message, the granted credit value is returned. RPC Calls SHOULD NOT be sent in excess of the currently granted limit. Further discussion of how the credit value is determined can be found in <xref target="managing-receiver-resources" pageno="false" format="default"/>. </t>
</section>
<section title="Procedure number" toc="default">
<t><list style="symbols"><t>RDMA_MSG = 0 indicates that chunk lists and a Payload stream follow. The format of the chunk lists is discussed below. </t><t>RDMA_NOMSG = 1 indicates that after the chunk lists there is no Payload stream. In this case, the chunk lists provide information to allow the responder to transfer the Payload stream using RDMA Read or Write operations. </t><t>RDMA_MSGP = 2 is reserved. </t><t>RDMA_DONE = 3 is reserved. </t><t>RDMA_ERROR = 4 is used to signal an encoding error in the RPC-over-RDMA header. </t></list> </t>
<t>An RDMA_MSG procedure conveys the Transport stream and the Payload stream via an RDMA Send operation. The Transport stream contains the four fixed fields, followed by the Read and Write lists and the Reply chunk, though any or all three MAY be marked as not present. The Payload stream then follows, beginning with its XID field. If a Read or Write chunk list is present, a portion of the Payload stream has been excised and is conveyed separately via RDMA Read or Write operations. </t>
<t>An RDMA_NOMSG procedure conveys the Transport stream via an RDMA Send operation. The Transport stream contains the four fixed fields, followed by the Read and Write chunk lists and the Reply chunk. Though any of these MAY be marked as not present, one MUST be present and MUST hold the Payload stream for this RPC-over-RDMA message. If a Read or Write chunk list is present, a portion of the Payload stream has been excised and is conveyed separately via RDMA Read or Write operations. </t>
<t>An RDMA_ERROR procedure conveys the Transport stream via an RDMA Send operation. The Transport stream contains the four fixed fields, followed by formatted error information. No Payload stream is conveyed in this type of RPC-over-RDMA message. </t>
<t>A gather operation on each RDMA Send operation can be used to combine the Transport and Payload streams, which might have been constructed in separate buffers. However, the total length of the gathered send buffers MUST NOT exceed the peer receiver's inline threshold. </t>
</section>
</section>
<section title="Chunk Lists" toc="default">
<t>The chunk lists in an RPC-over-RDMA Version One header are three XDR optional-data fields that follow the fixed header fields in RDMA_MSG and RDMA_NOMSG procedures. Read Section 4.19 of <xref target="RFC4506" pageno="false" format="default"/> carefully to understand how optional-data fields work. Examples of XDR encoded chunk lists are provided in <xref target="xdr-examples" pageno="false" format="default"/> as an aid to understanding. </t>
<section title="Read List" toc="default">
<t>Each RDMA_MSG or RDMA_NOMSG procedure has one "Read list." The Read list is a list of zero or more Read segments, provided by the requester, that are grouped by their Position fields into Read chunks. Each Read chunk advertises the location of argument data the responder is to retrieve via RDMA Read operations. The requester has removed the data in these chunks from the call's Payload stream. </t>
<t>Via a Position Zero Read Chunk, a requester may provide an RPC Call message as a chunk in the Read list. </t>
<t>If the RPC Call has no argument data that is DDP-eligible and the Position Zero Read Chunk is not being used, the requester leaves the Read list empty. </t>
</section>
<section title="Write List" anchor="write-list" toc="default">
<t>Each RDMA_MSG or RDMA_NOMSG procedure has one "Write list." The Write list is a list of zero or more Write chunks, provided by the requester. Each Write chunk is an array of RDMA segments, thus the Write list is a list of counted arrays. Each Write chunk advertises receptacles for DDP-eligible data to be pushed by the responder via RDMA Write operations. If the RPC Reply has no possible DDP-eligible result data items, the requester leaves the Write list empty. </t>
<t>*** This section needs to specify when a requester must provide Write chunks, and how many chunks must be provided. *** </t>
<t>When a Write list is provided for the results of an RPC Call, the responder MUST provide data corresponding to DDP-eligible XDR data items via RDMA Write operations to the memory referenced in the Write list. The responder removes the data in these chunks from the reply's Payload stream. </t>
<t>When multiple Write chunks are present, the responder fills in each Write chunk with a DDP-eligible result until either there are no more results or no more Write chunks. The requester may not be able to predict which DDP-eligible data item goes in which chunk. Thus the requester is responsible for allocating and registering Write chunks large enough to accommodate the largest XDR data item that might be associated with each chunk in the list. </t>
<t>The RPC Reply conveys the size of result data items by returning each Write chunk to the requester with the segment lengths rewritten to match the actual data transferred. Decoding the reply therefore performs no local data copying but merely returns the length obtained from the reply. </t>
<t>Each decoded result consumes one entry in the Write list, which in turn consists of an array of RDMA segments. The length of a Write chunk is therefore the sum of all returned lengths in all segments comprising the corresponding list entry. As each Write chunk is decoded, the entire Write list entry is consumed. </t>
<t>A requester constructs the Write list for an RPC transaction before the responder has formulated its reply. When there is only one DDP-eligible result data item, the requester inserts only a single Write chunk in the Write list. If the responder populates that chunk with data, the requester knows with certainty which result data item is contained in it. </t>
<t>However, Upper Layer Protocol procedures may allow replies where more than one result data item is DDP-eligible. For example, an NFSv4 COMPOUND procedure is composed of individual NFSv4 operations, more than one of which may have a reply containing a DDP-eligible result. </t>
<t>As stated above, when multiple Write chunks are present, the responder reduces DDP-eligible result until either there are no more results or no more Write chunks. Then, as the requester decodes the reply Payload stream, it is clear from the contents of the reply which Write chunk contains which data item. </t>
</section>
<section title="Reply Chunk" toc="default">
<t>Each RDMA_MSG or RDMA_NOMSG procedure has one "Reply chunk." The Reply chunk is a Write chunk, provided by the requester. The Reply chunk is a single counted array of RDMA segments. </t>
<t>A requester MUST provide a Reply chunk whenever the maximum possible size of the reply is larger than its own inline threshold. The Reply chunk MUST be large enough to contain a Payload stream (RPC message) of this maximum size. If the actual reply Payload stream is smaller than the requester's inline threshold, the responder MAY return it as a Short message rather than using the Reply chunk. </t>
</section>
</section>
<section title="Memory Registration" toc="default">
<t>RDMA requires that data is transferred between only registered memory segments at the source and destination. All protocol headers as well as separately transferred data chunks must reside in registered memory. </t>
<t>Since the cost of registering and de-registering memory can be a significant proportion of the RDMA transaction cost, it is important to minimize registration activity. For memory that is targeted by RDMA Send and Receive operations, a local-only registration is sufficient and can be left in place during the life of a connection without any risk of data exposure. </t>
<section title="Registration Longevity" toc="default">
<t>Data transferred via RDMA Read and Write can reside in a memory allocation not in the control of the RPC-over-RDMA transport. These memory allocations can persist outside the bounds of an RPC transaction. They are registered and invalidated as needed, as part of each RPC transaction. </t>
<t>The requester endpoint must ensure that memory segments associated with each RPC transaction are properly fenced from responders before allowing Upper Layer access to the data contained in them. Moreover, the requester must not access these memory segments while the responder has access to them. </t>
<t>This includes segments that are associated with canceled RPCs. A responder cannot know that the requester is no longer waiting for a reply, and might proceed to read or even update memory that the requester might have released for other use. </t>
</section>
<section title="Communicating DDP-Eligibility" toc="default">
<t>The interface by which an Upper Layer Protocol implementation communicates the eligibility of a data item locally to its local RPC-over-RDMA endpoint is not described by this specification. </t>
<t>Depending on the implementation and constraints imposed by Upper Layer Bindings, it is possible to implement reduction transparently to upper layers. Such implementations may lead to inefficiencies, either because they require the RPC layer to perform expensive registration and de-registration of memory "on the fly", or they may require using RDMA chunks in reply messages, along with the resulting additional handshaking with the RPC-over-RDMA peer. </t>
<t>However, these issues are internal and generally confined to the local interface between RPC and its upper layers, one in which implementations are free to innovate. The only requirement is that the resulting RPC-over-RDMA protocol sent to the peer is valid for the upper layer. </t>
</section>
<section title="Registration Strategies" toc="default">
<t>The choice of which memory registration strategies to employ is left to requester and responder implementers. To support the widest array of RDMA implementations, as well as the most general steering tag scheme, an Offset field is included in each segment. </t>
<t>While zero-based offset schemes are available in many RDMA implementations, their use by RPC requires individual registration of each segment. For such implementations, this can be a significant overhead. By providing an offset in each chunk, many pre-registration or region-based registrations can be readily supported. By using a single, universal chunk representation, the RPC-over-RDMA protocol implementation is simplified to its most general form. </t>
</section>
</section>
<section title="Error Handling" toc="default">
<t>A receiver performs basic validity checks on the RPC-over-RDMA header and chunk contents before it passes the RPC message to the RPC consumer. If errors are detected in an RPC-over-RDMA header, an RDMA_ERROR procedure MUST be generated. Because the transport layer may not be aware of the direction of a problematic RPC message, an RDMA_ERROR procedure MAY be generated by either a requester or a responder. </t>
<t>To form an RDMA_ERROR procedure: The rdma_xid field MUST contain the same XID that was in the rdma_xid field in the failing request; The rdma_vers field MUST contain the same version that was in the rdma_vers field in the failing request; The rdma_proc field MUST contain the value RDMA_ERROR; The rdma_err field contains a value that reflects the type of error that occurred, as described below. </t>
<t>An RDMA_ERROR procedure indicates a permanent error. Receipt of this procedure completes the RPC transaction associated with XID in the rdma_xid field. A receiver MUST silently discard an RDMA_ERROR procedure that cannot be decoded. </t>
<section title="Header Version Mismatch" toc="default">
<t>When a receiver detects an RPC-over-RDMA header version that it does not support (currently this document defines only Version One), it MUST reply with an RDMA_ERROR procedure and set the rdma_err value to RDMA_ERR_VERS, also providing the low and high inclusive version numbers it does, in fact, support. </t>
</section>
<section title="XDR Errors" anchor="xdr-errors" toc="default">
<t>A receiver might encounter an XDR parsing error that prevents it from processing the incoming Transport stream. Examples of such errors include an invalid value in the rdma_proc field, an RDMA_NOMSG message that has no chunk lists, or the contents of the rdma_xid field might not match the contents of the XID field in the accompanying RPC message. If the rdma_vers field contains a recognized value, but an XDR parsing error occurs, the responder MUST reply with an RDMA_ERROR procedure and set the rdma_err value to RDMA_ERR_BADHEADER. </t>
<t>When a responder receives a valid RPC-over-RDMA header but the responder's Upper Layer Protocol implementation cannot parse the RPC arguments in the RPC Call message, the responder SHOULD return a RPC_GARBAGEARGS reply, using an RDMA_MSG procedure. This type of parsing failure might be due to mismatches between chunk sizes or offsets and the contents of the Payload stream, for example. A responder MAY also report the presence of a non-DDP-eligible data item in a Read or Write chunk using RPC_GARBAGEARGS. </t>
</section>
<section title="Responder RDMA Operational Errors" toc="default">
<t>In RPC-over-RDMA Version One, it is the responder which drives RDMA Read and Write operations that target the requester's memory. Problems might arise as the responder attempts to use requester-provided resources for RDMA operations. For example: <list style="symbols"><t>Chunks can be validated only by using their contents to form RDMA Read or Write operations. If chunk contents are invalid (say, a segment is no longer registered, or a chunk length is too long), a Remote Access error occurs. </t><t>If a requester's receive buffer is too small, the responder's Send operation completes with a Local Length Error. </t><t>If the requester-provided Reply chunk is too small to accommodate a large RPC Reply, a Remote Access error occurs. A responder can detect this problem before attempting to write past the end of the Reply chunk. </t></list> </t>
<t>RDMA operational errors are typically fatal to the connection. To avoid a retransmission loop and repeated connection loss that deadlocks the connection, once the requester has re-established a connection, the responder should send an RDMA_ERROR reply with an rdma_err value of RDMA_ERR_BADHEADER to indicate that no RPC-level reply is possible for that XID. </t>
</section>
<section title="Other Operational Errors" toc="default">
<t>While a requester is constructing a Call message, an unrecoverable problem might occur that prevents the requester from posting further RDMA Work Requests on behalf of that message. As with other transports, if a requester is unable to construct and transmit a Call message, the associated RPC transaction fails immediately. </t>
<t>After a requester has received a reply, if it is unable to invalidate a memory region due to an unrecoverable problem, the requester MUST close the connection to fence that memory from the responder before the associated RPC transaction is complete. </t>
<t>While a responder is constructing a Reply message or error message, an unrecoverable problem might occur that prevents the responder from posting further RDMA Work Requests on behalf of that message. If a responder is unable to construct and transmit a Reply or error message, the responder MUST close the connection to signal to the requester that a reply was lost. </t>
</section>
<section title="RDMA Transport Errors" toc="default">
<t>The RDMA connection and physical link provide some degree of error detection and retransmission. iWARP's Marker PDU Aligned (MPA) layer (when used over TCP), Stream Control Transmission Protocol (SCTP), as well as the InfiniBand link layer all provide Cyclic Redundancy Check (CRC) protection of the RDMA payload, and CRC-class protection is a general attribute of such transports. </t>
<t>Additionally, the RPC layer itself can accept errors from the link level and recover via retransmission. RPC recovery can handle complete loss and re-establishment of the link. </t>
<t>The details of reporting and recovery from RDMA link layer errors are outside the scope of this protocol specification. See <xref target="security-considerations" pageno="false" format="default"/> for further discussion of the use of RPC-level integrity schemes to detect errors. </t>
</section>
</section>
<section title="Protocol Elements No Longer Supported" toc="default">
<t>The following protocol elements are no longer supported in RPC-over-RDMA Version One. Related enum values and structure definitions remain in the RPC-over-RDMA Version One protocol for backwards compatibility. </t>
<section title="RDMA_MSGP" toc="default">
<t>The specification of RDMA_MSGP in Section 3.9 of <xref target="RFC5666" pageno="false" format="default"/> is incomplete. To fully specify RDMA_MSGP would require: <list style="symbols"><t>Updating the definition of DDP-eligibility to include data items that may be transferred, with padding, via RDMA_MSGP procedures </t><t>Adding full operational descriptions of the alignment and threshold fields </t><t>Discussing how alignment preferences are communicated between two peers without using CCP </t><t>Describing the treatment of RDMA_MSGP procedures that convey Read or Write chunks </t></list> </t>
<t>The RDMA_MSGP message type is beneficial only when the padded data payload is at the end of an RPC message's argument or result list. This is not typical for NFSv4 COMPOUND RPCs, which often include a GETATTR operation as the final element of the compound operation array. </t>
<t>Without a full specification of RDMA_MSGP, there has been no fully implemented prototype of it. Without a complete prototype of RDMA_MSGP support, it is difficult to assess whether this protocol element has benefit, or can even be made to work interoperably. </t>
<t>Therefore, senders MUST NOT send RDMA_MSGP procedures. When receiving an RDMA_MSGP procedure, receivers SHOULD reply with an RDMA_ERROR procedure, setting the rdma_err field to RDMA_ERR_BADHEADER. </t>
</section>
<section title="RDMA_DONE" toc="default">
<t>Because no implementation of RPC-over-RDMA Version One uses the Read-Read transfer model, there is never a need to send an RDMA_DONE procedure. </t>
<t>Therefore, senders MUST NOT send RDMA_DONE messages. When receiving an RDMA_DONE procedure, receivers SHOULD reply with an RDMA_ERROR procedure, setting the rdma_err field to RDMA_ERR_BADHEADER. </t>
</section>
</section>
<section title="XDR Examples" anchor="xdr-examples" toc="default">
<t>RPC-over-RDMA chunk lists are complex data types. In this section, illustrations are provided to help readers grasp how chunk lists are represented inside an RPC-over-RDMA header. </t>
<t>An RDMA segment is the simplest component, being made up of a 32-bit handle (H), a 32-bit length (L), and 64-bits of offset (OO). Once flattened into an XDR stream, RDMA segments appear as <figure title="" suppress-title="false" align="left" alt="" width="" height=""><artwork xml:space="preserve" name="" type="" align="left" alt="" width="" height="">
HLOO
</artwork></figure> </t>
<t>A Read segment has an additional 32-bit position field. Read segments appear as <figure title="" suppress-title="false" align="left" alt="" width="" height=""><artwork xml:space="preserve" name="" type="" align="left" alt="" width="" height="">
PHLOO
</artwork></figure> </t>
<t>A Read chunk is a list of Read segments. Each segment is preceded by a 32-bit word containing a one if there is a segment, or a zero if there are no more segments (optional-data). In XDR form, this would look like <figure title="" suppress-title="false" align="left" alt="" width="" height=""><artwork xml:space="preserve" name="" type="" align="left" alt="" width="" height="">
1 PHLOO 1 PHLOO 1 PHLOO 0
</artwork></figure> where P would hold the same value for each segment belonging to the same Read chunk. </t>
<t>The Read List is also a list of Read segments. In XDR form, this would look like a Read chunk, except that the P values could vary across the list. An empty Read List is encoded as a single 32-bit zero. </t>
<t>One Write chunk is a counted array of segments. In XDR form, the count would appear as the first 32-bit word, followed by an HLOO for each element of the array. For instance, a Write chunk with three elements would look like <figure title="" suppress-title="false" align="left" alt="" width="" height=""><artwork xml:space="preserve" name="" type="" align="left" alt="" width="" height="">
3 HLOO HLOO HLOO
</artwork></figure> </t>
<t>The Write List is a list of counted arrays. In XDR form, this is a combination of optional-data and counted arrays. To represent a Write List containing a Write chunk with three segments and a Write chunk with two segments, XDR would encode <figure title="" suppress-title="false" align="left" alt="" width="" height=""><artwork xml:space="preserve" name="" type="" align="left" alt="" width="" height="">
1 3 HLOO HLOO HLOO 1 2 HLOO HLOO 0
</artwork></figure> An empty Write List is encoded as a single 32-bit zero. </t>
<t>The Reply chunk is a Write chunk. Since it is an optional-data field, however, there is a 32-bit field in front of it that contains a one if the Reply chunk is present, or a zero if it is not. After encoding, a Reply chunk with 2 segments would look like <figure title="" suppress-title="false" align="left" alt="" width="" height=""><artwork xml:space="preserve" name="" type="" align="left" alt="" width="" height="">
1 2 HLOO HLOO
</artwork></figure> </t>
<t>Frequently a requester does not provide any chunks. In that case, after the four fixed fields in the RPC-over-RDMA header, there are simply three 32-bit fields that contain zero. </t>
</section>
</section>
<section title="RPC Bind Parameters" anchor="rpc-bind-parameters" toc="default">
<t>In setting up a new RDMA connection, the first action by a requester is to obtain a transport address for the responder. The mechanism used to obtain this address, and to open an RDMA connection is dependent on the type of RDMA transport, and is the responsibility of each RPC protocol binding and its local implementation. </t>
<t>RPC services normally register with a portmap or rpcbind <xref target="RFC1833" pageno="false" format="default"/> service, which associates an RPC Program number with a service address. (In the case of UDP or TCP, the service address for NFS is normally port 2049.) This policy is no different with RDMA transports, although it may require the allocation of port numbers appropriate to each Upper Layer Protocol that uses the RPC framing defined here. </t>
<t>When mapped atop the iWARP transport <xref target="RFC5040" pageno="false" format="default"/> <xref target="RFC5041" pageno="false" format="default"/>, which uses IP port addressing due to its layering on TCP and/or SCTP, port mapping is trivial and consists merely of issuing the port in the connection process. The NFS/RDMA protocol service address has been assigned port 20049 by IANA, for both iWARP/TCP and iWARP/SCTP. </t>
<t>When mapped atop InfiniBand <xref target="IB" pageno="false" format="default"/>, which uses a Group Identifier (GID)-based service endpoint naming scheme, a translation MUST be employed. One such translation is defined in the InfiniBand Port Addressing Annex <xref target="IBPORT" pageno="false" format="default"/>, which is appropriate for translating IP port addressing to the InfiniBand network. Therefore, in this case, IP port addressing may be readily employed by the upper layer. </t>
<t>When a mapping standard or convention exists for IP ports on an RDMA interconnect, there are several possibilities for each upper layer to consider: <list style="symbols"><t>One possibility is to have responder register its mapped IP port with the rpcbind service, under the netid (or netid's) defined here. An RPC-over-RDMA-aware requester can then resolve its desired service to a mappable port, and proceed to connect. This is the most flexible and compatible approach, for those upper layers that are defined to use the rpcbind service. </t><t>A second possibility is to have the responder's portmapper register itself on the RDMA interconnect at a "well known" service address (on UDP or TCP, this corresponds to port 111). A requester could connect to this service address and use the portmap protocol to obtain a service address in response to a program number, e.g., an iWARP port number, or an InfiniBand GID. </t><t>Alternatively, the requester could simply connect to the mapped well-known port for the service itself, if it is appropriately defined. By convention, the NFS/RDMA service, when operating atop such an InfiniBand fabric, will use the same 20049 assignment as for iWARP. </t></list> </t>
<t>Historically, different RPC protocols have taken different approaches to their port assignment; therefore, the specific method is left to each RPC-over-RDMA-enabled Upper Layer binding, and not addressed here. </t>
<t>In <xref target="iana-considerations" pageno="false" format="default"/>, this specification defines two new "netid" values, to be used for registration of upper layers atop iWARP <xref target="RFC5040" pageno="false" format="default"/> <xref target="RFC5041" pageno="false" format="default"/> and (when a suitable port translation service is available) InfiniBand <xref target="IB" pageno="false" format="default"/>. Additional RDMA-capable networks MAY define their own netids, or if they provide a port translation, MAY share the one defined here. </t>
</section>
<section title="Bi-Directional RPC-Over-RDMA" toc="default">
<section title="RPC Direction" toc="default">
<section title="Forward Direction" toc="default">
<t>A traditional ONC RPC client is always a requester. A traditional ONC RPC service is always a responder. This traditional form of ONC RPC message passing is referred to as operation in the "forward direction." </t>
<t>During forward direction operation, the ONC RPC client is responsible for establishing transport connections. </t>
</section>
<section title="Backward Direction" toc="default">
<t>The ONC RPC standard does not forbid passing messages in the other direction. An ONC RPC service endpoint can act as a requester, in which case an ONC RPC client endpoint acts as a responder. This form of message passing is referred to as operation in the "backward direction." </t>
<t>During backward direction operation, the ONC RPC client is responsible for establishing transport connections, even though ONC RPC Calls come from the ONC RPC server. </t>
</section>
<section title="Bi-direction" toc="default">
<t>A pair of endpoints may choose to use only forward or only backward direction operations on a particular transport. Or, the endpoints may send operations in both directions concurrently on the same transport. </t>
<t>Bi-directional operation occurs when both transport endpoints act as a requester and a responder at the same time. As above, the ONC RPC client is responsible for establishing transport connections. </t>
</section>
<section title="XIDs with Bi-direction" toc="default">
<t>During bi-directional operation, the forward and backward directions use independent xid spaces. </t>
<t>In other words, a forward direction requester MAY use the same xid value at the same time as a backward direction requester on the same transport connection, but such concurrent requests represent distinct ONC RPC transactions. </t>
</section>
</section>
<section title="Backward Direction Flow Control" toc="default">
<section title="Backward RPC-over-RDMA Credits" toc="default">
<t>Credits work the same way in the backward direction as they do in the forward direction. However, forward direction credits and backward direction credits are accounted separately. </t>
<t>In other words, the forward direction credit value is the same whether or not there are backward direction resources associated with an RPC-over-RDMA transport connection. The backward direction credit value MAY be different than the forward direction credit value. The rdma_credit field in a backward direction RPC-over-RDMA message MUST NOT contain the value zero. </t>
<t>A backward direction requester (an RPC-over-RDMA service endpoint) requests credits from the responder (an RPC-over-RDMA client endpoint). The responder reports how many credits it can grant. This is the number of backward direction Calls the responder is prepared to handle at once. </t>
<t>When an RPC-over-RDMA server endpoint is operating correctly, it sends no more outstanding requests at a time than the client endpoint's advertised backward direction credit value. </t>
</section>
<section title="Receive Buffer Management" toc="default">
<t>An RPC-over-RDMA transport endpoint must pre-post receive buffers before it can receive and process incoming RPC-over-RDMA messages. If a sender transmits a message for a receiver which has no posted receive buffer, the RDMA provider MAY drop the RDMA connection. </t>
<section title="Client Receive Buffers" toc="default">
<t>Typically an RPC-over-RDMA caller posts only as many receive buffers as there are outstanding RPC Calls. A client endpoint without backward direction support might therefore at times have no pre-posted receive buffers. </t>
<t>To receive incoming backward direction Calls, an RPC-over-RDMA client endpoint must pre-post enough additional receive buffers to match its advertised backward direction credit value. Each outstanding forward direction RPC requires an additional receive buffer above this minimum. </t>
<t>When an RDMA transport connection is lost, all active receive buffers are flushed and are no longer available to receive incoming messages. When a fresh transport connection is established, a client endpoint must re-post a receive buffer to handle the Reply for each retransmitted forward direction Call, and a full set of receive buffers to handle backward direction Calls. </t>
</section>
<section title="Server Receive Buffers" toc="default">
<t>A forward direction RPC-over-RDMA service endpoint posts as many receive buffers as it expects incoming forward direction Calls. That is, it posts no fewer buffers than the number of RPC-over-RDMA credits it advertises in the rdma_credit field of forward direction RPC replies. </t>
<t>To receive incoming backward direction replies, an RPC-over-RDMA server endpoint must pre-post a receive buffer for each backward direction Call it sends. </t>
<t>When the existing transport connection is lost, all active receive buffers are flushed and are no longer available to receive incoming messages. When a fresh transport connection is established, a server endpoint must re-post a receive buffer to handle the Reply for each retransmitted backward direction Call, and a full set of receive buffers for receiving forward direction Calls. </t>
</section>
</section>
</section>
<section title="Conventions For Backward Operation" toc="default">
<section title="In the Absense of Backward Direction Support" toc="default">
<t>An RPC-over-RDMA transport endpoint might not support backward direction operation. There might be no mechanism in the transport implementation to do so, or the Upper Layer Protocol consumer might not yet have configured the transport to handle backward direction traffic. </t>
<t>A loss of the RDMA connection may result if the receiver is not prepared to receive an incoming message. Thus a denial-of-service could result if a sender continues to send backchannel messages after every transport reconnect to an endpoint that is not prepared to receive them. </t>
<t>For RPC-over-RDMA Version One transports, the Upper Layer Protocol is responsible for informing its peer when it has established a backward direction capability. Otherwise even a simple backward direction NULL probe from a peer would result in a lost connection. </t>
<t>An Upper Layer Protocol consumer MUST NOT perform backward direction ONC RPC operations unless the peer consumer has indicated it is prepared to handle them. A description of Upper Layer Protocol mechanisms used for this indication is outside the scope of this document. </t>
</section>
<section title="Backward Direction Retransmission" toc="default">
<t>In rare cases, an ONC RPC transaction cannot be completed within a certain time. This can be because the transport connection was lost, the Call or Reply message was dropped, or because the Upper Layer consumer delayed or dropped the ONC RPC request. Typically, the requester sends the transaction again, reusing the same RPC XID. This is known as an "RPC retransmission". </t>
<t>In the forward direction, the Caller is the ONC RPC client. The client is always responsible for establishing a transport connection before sending again. </t>
<t>In the backward direction, the Caller is the ONC RPC server. Because an ONC RPC server does not establish transport connections with clients, it cannot send a retransmission if there is no transport connection. It must wait for the ONC RPC client to re-establish the transport connection before it can retransmit ONC RPC transactions in the backward direction. </t>
<t>If an ONC RPC client has no work to do, it may be some time before it re-establishes a transport connection. Backward direction Callers must be prepared to wait indefinitely before a connection is established before a pending backward direction ONC RPC Call can be retransmitted. </t>
</section>
<section title="Backward Direction Message Size" toc="default">
<t>RPC-over-RDMA backward direction messages are transmitted and received using the same buffers as messages in the forward direction. Therefore they are constrained to be no larger than receive buffers posted for forward messages. </t>
<t>It is expected that the Upper Layer Protocol consumer establishes an appropriate payload size limit for backward direction operations, either by advertising that size limit to its peers, or by convention. If that is done, backward direction messages do not exceed the size of receive buffers at either endpoint. </t>
<t>If a sender transmits a backward direction message that is larger than the receiver is prepared for, the RDMA provider drops the message and the RDMA connection. </t>
</section>
<section title="Sending A Backward Direction Call" toc="default">
<t>To form a backward direction RPC-over-RDMA Call message on an RPC-over-RDMA Version One transport, an ONC RPC service endpoint constructs an RPC-over-RDMA header containing a fresh RPC XID in the rdma_xid field. </t>
<t>The rdma_vers field MUST contain the value one. The number of requested credits is placed in the rdma_credit field. </t>
<t>The rdma_proc field in the RPC-over-RDMA header MUST contain the value RDMA_MSG. All three chunk lists MUST be empty. </t>
<t>The ONC RPC Call header MUST follow immediately, starting with the same XID value that is present in the RPC-over-RDMA header. The Call header's msg_type field MUST contain the value CALL. </t>
</section>
<section title="Sending A Backward Direction Reply" toc="default">
<t>To form a backward direction RPC-over-RDMA Reply message on an RPC-over-RDMA Version One transport, an ONC RPC client endpoint constructs an RPC-over-RDMA header containing a copy of the matching ONC RPC Call's RPC XID in the rdma_xid field. </t>
<t>The rdma_vers field MUST contain the value one. The number of granted credits is placed in the rdma_credit field. </t>
<t>The rdma_proc field in the RPC-over-RDMA header MUST contain the value RDMA_MSG. All three chunk lists MUST be empty. </t>
<t>The ONC RPC Reply header MUST follow immediately, starting with the same XID value that is present in the RPC-over-RDMA header. The Reply header's msg_type field MUST contain the value REPLY. </t>
</section>
</section>
<section title="Backward Direction Upper Layer Binding" toc="default">
<t>RPC programs that operate on RPC-over-RDMA Version One only in the backward direction do not require an Upper Layer Binding specification. Because RPC-over-RDMA Version One operation in the backward direction does not allow reduction, there can be no DDP-eligible data items in such a program. Backward direction operation occurs on an already-established connection, thus there is no need to specify RPC bind parameters. </t>
</section>
</section>
<section title="Upper Layer Binding Specifications" anchor="upper-layer-binding-specifications" toc="default">
<t>An Upper Layer Protocol is typically defined independently of any particular RPC transport. An Upper Layer Binding specification (ULB) provides guidance that helps the Upper Layer Protocol interoperate correctly and efficiently over a particular transport. For RPC-over-RDMA Version One, a ULB provides: <list style="symbols"><t>A taxonomy of XDR data items that are eligible for Direct Data Placement </t><t>A method for determining the maximum size of the reply Payload stream for all procedures in the Upper Layer Protocol </t><t>An rpcbind port assignment for operation of the RPC Program and Version on an RPC-over-RDMA transport </t></list> </t>
<t>Each RPC Program and Version tuple that utilizes RPC-over-RDMA Version One needs to have an Upper Layer Binding specification. Requesters MUST NOT send RPC-over-RDMA messages for Upper Layer Protocols that do not have a Upper Layer Binding. Responders MUST NOT reply to RPC-over-RDMA messages for Upper Layer Protocols that do not have a Upper Layer Binding. </t>
<section title="DDP-Eligibility" toc="default">
<t>An Upper Layer Binding designates some XDR data items as eligible for Direct Data Placement. As an RPC-over-RDMA message is formed, DDP-eligible data items can be removed from the Payload stream and placed directly in the receiver's memory (reduced). </t>
<t>An XDR data item should be considered for DDP-eligibility if there is a clear benefit to moving the contents of the item directly from the sender's memory to the receiver's memory. Criteria for DDP-eligibility include: <list style="symbols"><t>The XDR data item is frequently sent or received, and its size is often much larger than typical inline thresholds. </t><t>Transport-level processing of the XDR data item is not needed. For example, the data item is an opaque byte array, which requires no XDR encoding and decoding of its content. </t><t>The content of the XDR data item is sensitive to address alignment. For example, pullup would be required on the receiver before the content of the item can be used. </t><t>The XDR data item does not contain DDP-eligible data items. </t></list> </t>
<t>Senders MUST NOT reduce data items that are not DDP-eligible. Such data items MAY, however, be moved as part of a Position Zero Read Chunk or a Reply chunk. </t>
<t>The interface by which an Upper Layer implementation indicates the DDP-eligibility of a data item to the RPC transport is not described by this specification. The only requirements are that the receiver can re-assemble the transmitted RPC-over-RDMA message into a valid XDR stream, and that DDP-eligibility rules specified by the Upper Layer Binding are respected. </t>
<t>There is no provision to express DDP-eligibility within the XDR language. The only definitive specification of DDP-eligibility is the Upper Layer Binding itself. </t>
<section title="DDP-Eligibility Violation" toc="default">
<t>A DDP-eligibility violation occurs when a requester forms a Call message with a non-DDP-eligible data item in a Read chunk. A violation occurs when a responder forms a Reply message without reducing a DDP-eligible data item when there is a Write list provided by the requester. </t>
<t>In the first case, a responder MUST NOT process the Call message. </t>
<t>In the second case, as a requester parses a Reply message, it must assume that the responder has correctly reduced a DDP-eligible result data item. If the responder has not done so, it is likely that the requester cannot finish parsing the Payload stream and that an XDR error would result. </t>
<t>Both types of violations MUST be reported as described in <xref target="xdr-errors" pageno="false" format="default"/>. </t>
</section>
</section>
<section title="Maximum Reply Size" toc="default">
<t>A requester provides resources for both a Call message and its matching Reply message. A requester forms the Call message itself, thus can compute the exact resources needed for it. </t>
<t>A requester must allocate resources for the Reply message (an RPC-over-RDMA credit, a Receive buffer, and possibly a Write list and Reply chunk) before the responder has formed the actual reply. To accommodate all possible replies for the procedure in the Call message, a requester must allocate reply resources based on the maximum possible size of the expected Reply message. </t>
<t>If there are procedures in the Upper Layer Protocol for which there is no clear reply size maximum, the Upper Layer Binding needs to specify a dependable means for determining the maximum. </t>
</section>
<section title="Additional Considerations" toc="default">
<t>There may be other details provided in an Upper Layer Binding. <list style="symbols"><t>An Upper Layer Binding may recommend an inline threshold value or other transport-related parameters for RPC-over-RDMA Version One connections bearing that Upper Layer Protocol. </t><t>An Upper Layer Protocol may provide a means to communicate these transport-related parameters between peers. Note that RPC-over-RDMA Version One does not specify any mechanism for changing any transport-related parameter after a connection has been established. </t><t>Multiple Upper Layer Protocols may share a single RPC-over-RDMA Version One connection when their Upper Layer Bindings allow the use of RPC-over-RDMA Version One and the rpcbind port assignments for the Protocols allow connection sharing. In this case, the same transport parameters (such as inline threshold) apply to all Protocols using that connection. </t></list> </t>
<t>Given the above, Upper Layer Bindings and Upper Layer Protocols must be designed to interoperate correctly no matter what connection parameters are in effect on a connection. </t>
</section>
<section title="Upper Layer Protocol Extensions" toc="default">
<t>An RPC Program and Version tuple may be extensible. For instance, there may be a minor versioning scheme that is not reflected in the RPC version number. Or, the Upper Layer Protocol may allow additional features to be specified after the original RPC program specification was ratified. </t>
<t>Upper Layer Bindings are provided for interoperable RPC Programs and Versions by extending existing Upper Layer Bindings to reflect the changes made necessary by each addition to the existing XDR. </t>
</section>
</section>
<section title="Protocol Extensibility" anchor="protocol-extensibility" toc="default">
<t>The RPC-over-RDMA header format is specified using XDR, unlike the message header format of RPC on TCP. Defining the header using XDR allows minor issues with the transport protocol to be addressed and optional features to be introduced by making extensions to the RPC-over-RDMA header XDR. Such changes can be made without a change to the protocol version number. </t>
<t>When more invasive changes to the protocol are to be made, a protocol version number change is required. In either case, any changes to the RPC-over-RDMA protocol can only be effected by publication of a Standards Track document with appropriate review by the nfsv4 Working Group and the IESG. </t>
<t>Although it is possible to make XDR changes which are not limited to the use of compatible extensions in new RPC-over-RDMA versions, such changes should only be done when absolutely necessary, as they limit interoperability with existing implementations. It is appropriate for the nfsv4 Working Group to consider alternatives carefully before using this approach. </t>
<t>Unlike the rest of this document, which defines the base of RPC-over-RDMA Version One, <xref target="protocol-extensibility" pageno="false" format="default"/> (except for <xref target="rpc-over-rdma-version-one-extension-practices" pageno="false" format="default"/>) applies to all versions of RPC-over-RDMA. New versions of RPC-over-RDMA may be published as separate protocols without updating this document, but any change to the extensibility model defined here requires publication of a Standards Track document updating this document. </t>
<section title="Changes To RPC-Over-RDMA Header XDR" anchor="changes-to-rpc-over-rdma-header-xdr" toc="default">
<t>The first four fields in the RPC-over-RDMA header (now in struct rpcrdma_prefix) must remain aligned at the same fixed offsets for all versions of the RPC-over-RDMA protocol. The version number must be in a fixed place in order to enable version mismatch detection. For version mismatches to be reported in a fashion that all future version implementations can reliably decode, the rdma_which field must be in a fixed place, the value of RDMA_ERR_VERS must always remain the same, and the field placement of the RDMA_ERR_VERS arm of the rpcrdma1_error union (now in struct rpcrdma_err_vers) must always remain the same. </t>
<t>Given these constraints, one way to extend RPC-over-RDMA is to add new values to the rdma_proc enumerated type and new components (arms) to the rpcrdma1_body union. New argument and result types may be introduced for each new procedure defined this way. These extensions would be specified by new Internet Drafts with appropriate Working Group and IESG review to ensure continued interoperation with existing implementations. </t>
<t>XDR extensions may introduce only optional features to an existing RPC-over-RDMA protocol version. To detect when an optional rdma_proc value is supported by a receiver, it is desirable to have a specific value of the rdma_err field, say, RDMA_ERR_PROC, that indicates when the receiver does not recognize an rdma_proc value. </t>
<t>In RPC-over-RDMA Version One, a receiver can indicate that it does not recognize an rdma_proc enum value only by returning an RDMA_ERROR procedure with the rdma_err field set to RDMA_ERR_BADHEADER (see <xref target="xdr-errors" pageno="false" format="default"/>). This is indistinguishable from a situation where the receiver does indeed support the procedure, but the XDR is malformed. </t>
<t>To resolve this problem, an RPC-over-RDMA Version One sender uses the following convention. If the first time the sender uses an optional rdma_proc value the receiver returns an RDMA_ERROR procedure with RDMA_ERR_BADHEADER in the rdma_err field, the sender simply marks that feature as unsupported and does not send it again on the current connection instance. Subsequent to an initial successful result, receiving RDMA_ERR_BADHEADER retains its more relaxed meaning of "generic XDR parsing error." </t>
<t>To ensure backwards compatibility when such an extension mechanism is in place, the value of RDMA_ERR_BADHEADER must remain the same for all versions of the RPC-over-RDMA protocol. </t>
<t>Most changes to the RPC-over-RDMA XDR will take the form of a compatible extension to the existing XDR. Changes which do not update the version number (see <xref target="rpc-over-rdma-version-numbering" pageno="false" format="default"/>) must take this form. </t>
<t>For an XDR description B to be a compatible extension of an XDR description A, the following must be the case: <list style="symbols"><t>All input recognized as description valid by A must be recognized as valid by description B </t><t>Any input recognized as valid by both descriptions must be interpreted as having the same structure according to both descriptions </t><t>Any input recognized as valid by description B but not by description A can be recognizable as part of a supported./unknown extension using description A </t></list> </t>
<t>The following changes can be made compatibly: <list style="symbols"><t>Addition of a new message header type and associated header body </t><t>Addition of new enum values and associated arms to unions that do not include a default case </t><t>Addition of previously undefined flag bits to flag words that are included in existing header bodies </t></list> Each such addition is referred to as a "protocol element." A set of protocol elements defined together such that all must be supported or not supported by a receiver is called a "feature." </t>
<t>Because of the simplicity of the existing protocol and deficiencies in the existing error reporting structure, some of the above techiques are not realizable within RPC-over-RDMA Version One. For a discussion of protocol extension practices within RPC-over-RDMA Version One, including XDR extension, see <xref target="rpc-over-rdma-version-one-extension-practices" pageno="false" format="default"/>. </t>
</section>
<section title="Feature Statuses With RPC-Over-RDMA Versions" toc="default">
<t>Within a given RPC-over-RDMA version, every known feature is either OPTIONAL, REQUIRED, or "not allowed". <list style="symbols"><t>REQUIRED features MUST be supported by all receivers. Senders can depend on them being supported. </t><t>OPTIONAL features MAY be supported by particular receivers. Senders need to be prepared for the absence of support. </t><t>"Not allowed" features are typically those that were formally OPTIONAL or REQUIRED, but for which support has been removed. </t></list> All features defined in this document are REQUIRED in RPC-over-RDMA Version One. OPTIONAL features may be added to Version One as specified in <xref target="rpc-over-rdma-version-one-extension-practices" pageno="false" format="default"/>. </t>
<t>The terms "OPTIONAL" and "REQUIRED" are used as specified in <xref target="RFC2119" pageno="false" format="default"/> as indicated in <xref target="requirements-language" pageno="false" format="default"/>. These status values are assigned by those writing additional specifications (e.g., new RPC-over-RDMA versions or extensions to existing RPC-over-RDMA versions). Their choice in this regard is their guidance to implementers. As used in this document, these terms are only directed to implementers of RPC-over-RDMA Version One. </t>
<t>The status of features may change between RPC-over-RDMA protocol versions. </t>
</section>
<section title="RPC-Over-RDMA Version Numbering" anchor="rpc-over-rdma-version-numbering" toc="default">
<t>RPC-over-RDMA version numbering enables both endpoints to agree on a set of interoperable behaviors and determine which OPTIONAL features are available. </t>
<t>An expected pattern of protocol development is to introduce OPTIONAL features within a given version using XDR extension. Such features often need a significant period of optional general use to ensure they are capable of being implemented broadly. This is especially true for infrastructural features that others will build upon. When it is appropriate for OPTIONAL features to become REQUIRED, that would be an occasion to create a new RPC-over-RDMA protocol version. </t>
<t>The value of the RPC-over-RDMA header's version field has to be updated when the protocol is altered in a way that prevents interoperability with current implementations. A version change is needed whenever: <list style="symbols"><t>The RPC-over-RDMA header XDR definition is changed to add a REQUIRED protocol element, or an existing OPTIONAL feature is made REQUIRED </t><t>A REQUIRED feature is made OPTIONAL </t><t>A REQUIRED or OPTIONAL feature is converted to be "not allowed" </t><t>An XDR change is made that is not a compatible extension as defined in <xref target="changes-to-rpc-over-rdma-header-xdr" pageno="false" format="default"/> </t><t>The use of a previously not used abstract RDMA operation is specified as REQUIRED </t><t>The use of an existing REQUIRED abstract RDMA operation is removed </t></list> </t>
<t>When a version number change is to be made, the nfsv4 Working Group creates a Standards Track document that does one of the following: <list style="numbers"><t>Documents the whole protocol as amended </t><t>Documents changes relative to the previous minor version </t><t>Documents extensions made since the previous minor versions by normatively referencing the documents defining those extensions </t><t>Documents all REQUIRED functionality, and includes OPTIONAL features by normatively referencing the documents defining those extensions </t></list> The Working Group retains all these options, but the last is typically preferred. When an XDR change that is not a compatible extension is made, the first is most desirable. In any case, if there are features whose status has been changed to "not allowed", the document needs to explain that change and how it is intended that existing implementations address the feature removal. </t>
</section>
<section title="RPC-Over-RDMA Version One Extension Practices" anchor="rpc-over-rdma-version-one-extension-practices" toc="default">
<t>This subsection applies primarily to RPC-over-RDMA Version One but remains in effect unless modified by documents defining future RPC-over-RDMA versions. Such documents need not update this document. </t>
<section title="Documentation Requirements" toc="default">
<t>RPC-over-RDMA Version One may be extended by defining a new message header type and XDR description of the corresponding header body. </t>
<t>A set of such new protocol elements may be introduced by a Standards Track document and are together considered an OPTIONAL feature. nfsv4 Working Group and IESG review, together with appropriate testing of prototype implementations, should ensure continued interoperation with existing implementations. </t>
<t>Documents describing extensions to RPC-over-RDMA Version One should contain: <list style="symbols"><t>An explanation of the purpose and use of each new protocol element </t><t>An XDR description and a script to extract it </t><t>A receiver response that a sender can use to determine that support is in fact present </t><t>A description of interactions with existing features (e.g., any requirement that another OPTIONAL or REQUIRED feature needs to be present and supported for the new feature to work) </t></list> Implementers concatenate the XDR description of the new feature with the XDR description of the base protocol, extracted from this document, to produce a combined XDR description for the RPC-over-RDMA Version One protocol with the specified extension. </t>
</section>
<section title="Detecting Support For Message Header Types" toc="default">
<t>A sender determines whether a receiver supports an OPTIONAL message header type by issuing a simple test request using that message header type. The receiver sends an affirmative response that indicates the message header type is supported. The response message header type may itself be an extension. The sender ties together the message and response using the rdma_xid field. </t>
<t>The receiver indicates that it does not recognize a particular rdma_which value by returning an RDMA_ERROR message type with the rdma_err field set to RDMA_ERR_BADHEADER and with the rdma_xid field set to a value that matches the test message. </t>
<t>This is indistinguishable from a situation where the receiver does support the procedure but the test message is malformed. However, if the sender always tests for receiver support using a simple instance of the message header type to be tested, such an error at this point indicates the sender and receiver have no prospect of using the new protocol element interoperably. A lack of support for this feature can be reasonably assumed. </t>
<t>A sender should issue OPTIONAL message header types one-at-a-time until it receives indication of the receiver's support status of that message header type. </t>
</section>
</section>
</section>
<section title="Security Considerations" anchor="security-considerations" toc="default">
<section title="Memory Protection" toc="default">
<t>A primary consideration is the protection of the integrity and privacy of local memory by an RPC-over-RDMA transport. The use of RPC-over-RDMA MUST NOT introduce any vulnerabilities to system memory contents, nor to memory owned by user processes. </t>
<t>It is REQUIRED that any RDMA provider used for RPC transport be conformant to the requirements of <xref target="RFC5042" pageno="false" format="default"/> in order to satisfy these protections. These protections are provided by the RDMA layer specifications, and in particular, their security models. </t>
<section title="Protection Domains" toc="default">
<t>The use of Protection Domains to limit the exposure of memory segments to a single connection is critical. Any attempt by an endpoint not participating in that connection to re-use memory handles needs to result in immediate failure of that connection. Because Upper Layer Protocol security mechanisms rely on this aspect of Reliable Connection behavior, strong authentication of remote endpoints is recommended. </t>
</section>
<section title="Handle Predictability" toc="default">
<t>Unpredictable memory handles should be used for any operation requiring advertised memory segments. Advertising a continuously registered memory region allows a remote host to read or write to that region even when an RPC involving that memory is not under way. Therefore implementations should avoid advertising persistently registered memory. </t>
</section>
<section title="Memory Fencing" toc="default">
<t>Advertised memory segments should be invalidated as soon as related RPC operations are complete. Invalidation and DMA unmapping of segments should be complete before the Upper Layer is allowed to continue execution and use or alter the contents of a memory region. </t>
</section>
</section>
<section title="RPC Message Security" toc="default">
<t>ONC RPC provides cryptographic security via the RPCSEC_GSS framework <xref target="RFC2203" pageno="false" format="default"/>. RPCSEC_GSS implements message authentication, per-message integrity checking, and per-message confidentiality. However, integrity and privacy services require significant movement of data on each endpoint host. Some performance benefits enabled by RDMA transports can be lost. Note that some performance loss is expected when RPCSEC_GSS integrity or privacy is in use on any RPC transport. </t>
<section title="RPC-Over-RDMA Link-Level Protection" anchor="rpc-over-rdma-link-level-protection" toc="default">
<t>Link-level protection is a more appropriate security mechanism for RDMA transports. Certain configurations of IPsec can be co-located in RDMA hardware, for example, without any change to RDMA consumers or loss of data movement efficiency. </t>
<t>The use of link-level protection MAY be negotiated through the use of the RPCSEC_GSS security flavor defined in <xref target="RFC5403" pageno="false" format="default"/> in conjunction with the Channel Binding mechanism <xref target="RFC5056" pageno="false" format="default"/> and IPsec Channel Connection Latching <xref target="RFC5660" pageno="false" format="default"/>. Use of such mechanisms is REQUIRED where integrity and/or privacy is desired and where efficiency is required. </t>
</section>
<section title="RPCSEC_GSS On RPC-Over-RDMA Transports" toc="default">
<t>RPCSEC_GSS <xref target="RFC5403" pageno="false" format="default"/> extends the ONC RPC protocol <xref target="RFC5531" pageno="false" format="default"/> without changing the format of RPC messages. By observing the conventions described in this section, an RPC-over-RDMA implementation can support RPCSEC_GSS in a way that interoperates successfully with other implementations. </t>
<t>As part of the ONC RPC protocol, protocol elements of RPCSEC_GSS that appear in the Payload stream of an RPC-over-RDMA message (such as control messages exchanged as part of establishing or destroying a security context, or data items that are part of RPCSEC_GSS authentication material) MUST NOT be reduced. </t>
<section title="RPCSEC_GSS Context Negotiation" toc="default">
<t>Some NFS client implementations use a separate connection to establish a GSS context for NFS operation. These clients use TCP and the standard NFS port (2049) for context establishment, but there is no guarantee that an NFS/RDMA server provides a TCP-based NFS server on port 2049. </t>
</section>
<section title="RPC-Over-RDMA With RPCSEC_GSS Authentication" toc="default">
<t>The RPCSEC_GSS authentication service has no impact on the DDP-eligibity of data items in an Upper Layer Protocol. </t>
<t>However, RPCSEC_GSS authentication material appearing in an RPC message header is often larger than material associated with, say, the AUTH_SYS security flavor. In particular, when an RPCSEC_GSS pseudoflavor is in use, a requester needs to accommodate a larger RPC credential when marshaling Call messages, and to provide for a maximum size RPCSEC_GSS verifier when allocating reply buffers and Reply chunks. </t>
<t>RPC messages, and thus Payload streams, are made larger as a result. Upper Layer Protocol operations that fit in a Short Message when a simpler form of authentication is in use might need to be reduced or conveyed via a Long Message when RPCSEC_GSS authentication is in use. This can impact efficiency when RPCSEC_GSS authentication is use. </t>
<t>Because average RPC message size is larger when RPCSEC_GSS authentication is in use, it is more likely that a requester will provide both a Read list and a Reply chunk in the same RPC-over-RDMA header to convey a Long call and provision a receptacle for a Long reply. </t>
</section>
<section title="RPC-Over-RDMA With RPCSEC_GSS Integrity Or Privacy" toc="default">
<t>The RPCSEC_GSS integrity service enables endpoints to detect modification of RPC messages in flight. The RPCSEC_GSS privacy service prevents all but the intended recipient from viewing the cleartext content of RPC messages. RPCSEC_GSS integrity and privacy are end-to-end; that is, they protect RPC arguments and results from application to server endpoint, and back. </t>
<t>The RPCSEC_GSS integrity and encryption services operate on whole RPC messages after they have been XDR encoded for transmit, and before they have been XDR decoded after receipt. Both the sender and the receiver endpoints use intermediate buffers to prevent exposure of encrypted data or unverified cleartext data to RPC consumers. After verification, encryption, and message wrapping has been performed, the transport layer can use RDMA data transfer between these intermediate buffers. </t>
<t>The process of reducing a DDP-eligible data item removes the data item and its XDR padding from the encoded XDR stream. XDR padding of a reduced data item is not transferred in an RPC-over-RDMA message. After reduction, the Payload stream contains fewer octets then the whole XDR stream did beforehand. XDR padding octets are often zero bytes, but they don't have to be. Thus reducing DDP-eligible items affects the result of message integrity verification or encryption. </t>
<t>Therefore a sender MUST NOT reduce a Payload stream when RPCSEC_GSS integrity or encryption services are in use. Effectively, no data item is DDP-eligible in this situation, and Chunked Messages cannot be used. In this mode, an RPC-over-RDMA transport operates in the same manner as a transport that does not support direct data placement. </t>
<t>When RPCSEC_GSS integrity or privacy is in use, a requester provides both a Read list and a Reply chunk in the same RPC-over-RDMA header to convey a Long call and provision a receptacle for a Long reply. </t>
</section>
<section title="RPC-Over-RDMA Header Exposure" toc="default">
<t>Like the base fields in an ONC RPC message (XID, call direction, and so on), the contents of an RPC-over-RDMA message's Transport stream are not protected by RPCSEC_GSS. This exposes XIDs, connection credit limits, and chunk lists (but not the content of the data items they refer to) to malicious behavior, which could redirect data that is transferred by the RPC-over-RDMA message, result in spurious retransmits, or trigger connection loss. </t>
<t>Encryption at the link layer, as described in <xref target="rpc-over-rdma-link-level-protection" pageno="false" format="default"/>, protects the content of the Transport stream. </t>
</section>
</section>
</section>
</section>
<section title="IANA Considerations" anchor="iana-considerations" toc="default">
<t>Three assignments are specified by this document. These are unchanged from <xref target="RFC5666" pageno="false" format="default"/>: <list style="symbols"><t>A set of RPC "netids" for resolving RPC-over-RDMA services </t><t>Optional service port assignments for Upper Layer Bindings </t><t>An RPC program number assignment for the configuration protocol </t></list> </t>
<t>These assignments have been established, as below. </t>
<t>The new RPC transport has been assigned an RPC "netid", which is an rpcbind <xref target="RFC1833" pageno="false" format="default"/> string used to describe the underlying protocol in order for RPC to select the appropriate transport framing, as well as the format of the service addresses and ports. </t>
<t>The following "Netid" registry strings are defined for this purpose: <figure title="" suppress-title="false" align="left" alt="" width="" height=""><artwork xml:space="preserve" name="" type="" align="left" alt="" width="" height="">
NC_RDMA "rdma"
NC_RDMA6 "rdma6"
</artwork></figure> </t>
<t>These netids MAY be used for any RDMA network satisfying the requirements of <xref target="rdma-transport-requirements" pageno="false" format="default"/>, and able to identify service endpoints using IP port addressing, possibly through use of a translation service as described above in <xref target="rpc-bind-parameters" pageno="false" format="default"/>. The "rdma" netid is to be used when IPv4 addressing is employed by the underlying transport, and "rdma6" for IPv6 addressing. </t>
<t>The netid assignment policy and registry are defined in <xref target="RFC5665" pageno="false" format="default"/>. </t>
<t>As a new RPC transport, this protocol has no effect on RPC Program numbers or existing registered port numbers. However, new port numbers MAY be registered for use by RPC-over-RDMA-enabled services, as appropriate to the new networks over which the services will operate. </t>
<t>For example, the NFS/RDMA service defined in <xref target="RFC5667" pageno="false" format="default"/> has been assigned the port 20049, in the IANA registry: <figure title="" suppress-title="false" align="left" alt="" width="" height=""><artwork xml:space="preserve" name="" type="" align="left" alt="" width="" height="">
nfsrdma 20049/tcp Network File System (NFS) over RDMA
nfsrdma 20049/udp Network File System (NFS) over RDMA
nfsrdma 20049/sctp Network File System (NFS) over RDMA
</artwork></figure> </t>
<t>The RPC program number assignment policy and registry are defined in <xref target="RFC5531" pageno="false" format="default"/>. </t>
</section>
<section title="Acknowledgments" toc="default">
<t>The editor gratefully acknowledges the work of Brent Callaghan and Tom Talpey on the original RPC-over-RDMA Version One specification <xref target="RFC5666" pageno="false" format="default"/>. </t>
<t>Dave Noveck provided excellent review, constructive suggestions, and consistent navigational guidance throughout the process of drafting this document. Dave also contributed much of the organization and content of <xref target="protocol-extensibility" pageno="false" format="default"/> and helped the authors understand the complexities of XDR extensibility. </t>
<t>The comments and contributions of Karen Deitke, Dai Ngo, Chunli Zhang, Dominique Martinet, and Mahesh Siddheshwar are accepted with great thanks. The editor also wishes to thank Bill Baker for his support of this work. </t>
<t>The extract.sh shell script and formatting conventions were first described by the authors of the NFSv4.1 XDR specification <xref target="RFC5662" pageno="false" format="default"/>. </t>
<t>Special thanks go to nfsv4 Working Group Chair Spencer Shepler and nfsv4 Working Group Secretary Thomas Haynes for their support. </t>
</section>
</middle>
<back>
<references title="Normative References">
<reference anchor="RFC1833" target="http://www.rfc-editor.org/info/rfc1833">
<front>
<title>Binding Protocols for ONC RPC Version 2</title>
<author initials="R." surname="Srinivasan" fullname="R. Srinivasan">
<organization/>
</author>
<date year="1995" month="August"/>
<abstract>
<t>This document describes the binding protocols used in conjunction with the ONC Remote Procedure Call (ONC RPC Version 2) protocols. [STANDARDS-TRACK]</t>
</abstract>
</front>
<seriesInfo name="RFC" value="1833"/>
<seriesInfo name="DOI" value="10.17487/RFC1833"/>
</reference>
<reference anchor="RFC2119" target="http://www.rfc-editor.org/info/rfc2119">
<front>
<title>Key words for use in RFCs to Indicate Requirement Levels</title>
<author initials="S." surname="Bradner" fullname="S. Bradner">
<organization/>
</author>
<date year="1997" month="March"/>
<abstract>
<t>In many standards track documents several words are used to signify the requirements in the specification. These words are often capitalized. This document defines these words as they should be interpreted in IETF documents. This document specifies an Internet Best Current Practices for the Internet Community, and requests discussion and suggestions for improvements.</t>
</abstract>
</front>
<seriesInfo name="BCP" value="14"/>
<seriesInfo name="RFC" value="2119"/>
<seriesInfo name="DOI" value="10.17487/RFC2119"/>
</reference>
<reference anchor="RFC2203" target="http://www.rfc-editor.org/info/rfc2203">
<front>
<title>RPCSEC_GSS Protocol Specification</title>
<author initials="M." surname="Eisler" fullname="M. Eisler">
<organization/>
</author>
<author initials="A." surname="Chiu" fullname="A. Chiu">
<organization/>
</author>
<author initials="L." surname="Ling" fullname="L. Ling">
<organization/>
</author>
<date year="1997" month="September"/>
<abstract>
<t>This memo describes an ONC/RPC security flavor that allows RPC protocols to access the Generic Security Services Application Programming Interface (referred to henceforth as GSS-API). [STANDARDS-TRACK]</t>
</abstract>
</front>
<seriesInfo name="RFC" value="2203"/>
<seriesInfo name="DOI" value="10.17487/RFC2203"/>
</reference>
<reference anchor="RFC4506" target="http://www.rfc-editor.org/info/rfc4506">
<front>
<title>XDR: External Data Representation Standard</title>
<author initials="M." surname="Eisler" fullname="M. Eisler" role="editor">
<organization/>
</author>
<date year="2006" month="May"/>
<abstract>
<t>This document describes the External Data Representation Standard (XDR) protocol as it is currently deployed and accepted. This document obsoletes RFC 1832. [STANDARDS-TRACK]</t>
</abstract>
</front>
<seriesInfo name="STD" value="67"/>
<seriesInfo name="RFC" value="4506"/>
<seriesInfo name="DOI" value="10.17487/RFC4506"/>
</reference>
<reference anchor="RFC5042" target="http://www.rfc-editor.org/info/rfc5042">
<front>
<title>Direct Data Placement Protocol (DDP) / Remote Direct Memory Access Protocol (RDMAP) Security</title>
<author initials="J." surname="Pinkerton" fullname="J. Pinkerton">
<organization/>
</author>
<author initials="E." surname="Deleganes" fullname="E. Deleganes">
<organization/>
</author>
<date year="2007" month="October"/>
<abstract>
<t>This document analyzes security issues around implementation and use of the Direct Data Placement Protocol (DDP) and Remote Direct Memory Access Protocol (RDMAP). It first defines an architectural model for an RDMA Network Interface Card (RNIC), which can implement DDP or RDMAP and DDP. The document reviews various attacks against the resources defined in the architectural model and the countermeasures that can be used to protect the system. Attacks are grouped into those that can be mitigated by using secure communication channels across the network, attacks from Remote Peers, and attacks from Local Peers. Attack categories include spoofing, tampering, information disclosure, denial of service, and elevation of privilege. [STANDARDS-TRACK]</t>
</abstract>
</front>
<seriesInfo name="RFC" value="5042"/>
<seriesInfo name="DOI" value="10.17487/RFC5042"/>
</reference>
<reference anchor="RFC5056" target="http://www.rfc-editor.org/info/rfc5056">
<front>
<title>On the Use of Channel Bindings to Secure Channels</title>
<author initials="N." surname="Williams" fullname="N. Williams">
<organization/>
</author>
<date year="2007" month="November"/>
<abstract>
<t>The concept of channel binding allows applications to establish that the two end-points of a secure channel at one network layer are the same as at a higher layer by binding authentication at the higher layer to the channel at the lower layer. This allows applications to delegate session protection to lower layers, which has various performance benefits.</t>
<t>This document discusses and formalizes the concept of channel binding to secure channels. [STANDARDS-TRACK]</t>
</abstract>
</front>
<seriesInfo name="RFC" value="5056"/>
<seriesInfo name="DOI" value="10.17487/RFC5056"/>
</reference>
<reference anchor="RFC5403" target="http://www.rfc-editor.org/info/rfc5403">
<front>
<title>RPCSEC_GSS Version 2</title>
<author initials="M." surname="Eisler" fullname="M. Eisler">
<organization/>
</author>
<date year="2009" month="February"/>
<abstract>
<t>This document describes version 2 of the RPCSEC_GSS protocol. Version 2 is the same as version 1 (specified in RFC 2203) except that support for channel bindings has been added. RPCSEC_GSS allows remote procedure call (RPC) protocols to access the Generic Security Services Application Programming Interface (GSS-API). [STANDARDS-TRACK]</t>
</abstract>
</front>
<seriesInfo name="RFC" value="5403"/>
<seriesInfo name="DOI" value="10.17487/RFC5403"/>
</reference>
<reference anchor="RFC5531" target="http://www.rfc-editor.org/info/rfc5531">
<front>
<title>RPC: Remote Procedure Call Protocol Specification Version 2</title>
<author initials="R." surname="Thurlow" fullname="R. Thurlow">
<organization/>
</author>
<date year="2009" month="May"/>
<abstract>
<t>This document describes the Open Network Computing (ONC) Remote Procedure Call (RPC) version 2 protocol as it is currently deployed and accepted. This document obsoletes RFC 1831. [STANDARDS-TRACK]</t>
</abstract>
</front>
<seriesInfo name="RFC" value="5531"/>
<seriesInfo name="DOI" value="10.17487/RFC5531"/>
</reference>
<reference anchor="RFC5660" target="http://www.rfc-editor.org/info/rfc5660">
<front>
<title>IPsec Channels: Connection Latching</title>
<author initials="N." surname="Williams" fullname="N. Williams">
<organization/>
</author>
<date year="2009" month="October"/>
<abstract>
<t>This document specifies, abstractly, how to interface applications and transport protocols with IPsec so as to create "channels" by latching "connections" (packet flows) to certain IPsec Security Association (SA) parameters for the lifetime of the connections. Connection latching is layered on top of IPsec and does not modify the underlying IPsec architecture.</t>
<t>Connection latching can be used to protect applications against accidentally exposing live packet flows to unintended peers, whether as the result of a reconfiguration of IPsec or as the result of using weak peer identity to peer address associations. Weak association of peer ID and peer addresses is at the core of Better Than Nothing Security (BTNS); thus, connection latching can add a significant measure of protection to BTNS IPsec nodes.</t>
<t>Finally, the availability of IPsec channels will make it possible to use channel binding to IPsec channels. [STANDARDS-TRACK]</t>
</abstract>
</front>
<seriesInfo name="RFC" value="5660"/>
<seriesInfo name="DOI" value="10.17487/RFC5660"/>
</reference>
<reference anchor="RFC5665" target="http://www.rfc-editor.org/info/rfc5665">
<front>
<title>IANA Considerations for Remote Procedure Call (RPC) Network Identifiers and Universal Address Formats</title>
<author initials="M." surname="Eisler" fullname="M. Eisler">
<organization/>
</author>
<date year="2010" month="January"/>
<abstract>
<t>This document lists IANA Considerations for Remote Procedure Call (RPC) Network Identifiers (netids) and RPC Universal Network Addresses (uaddrs). This document updates, but does not replace, RFC 1833. [STANDARDS-TRACK]</t>
</abstract>
</front>
<seriesInfo name="RFC" value="5665"/>
<seriesInfo name="DOI" value="10.17487/RFC5665"/>
</reference>
</references>
<references title="Informative References">
<reference anchor="RFC0768" target="http://www.rfc-editor.org/info/rfc768">
<front>
<title>User Datagram Protocol</title>
<author initials="J." surname="Postel" fullname="J. Postel">
<organization/>
</author>
<date year="1980" month="August"/>
</front>
<seriesInfo name="STD" value="6"/>
<seriesInfo name="RFC" value="768"/>
<seriesInfo name="DOI" value="10.17487/RFC0768"/>
</reference>
<reference anchor="RFC0793" target="http://www.rfc-editor.org/info/rfc793">
<front>
<title>Transmission Control Protocol</title>
<author initials="J." surname="Postel" fullname="J. Postel">
<organization/>
</author>
<date year="1981" month="September"/>
</front>
<seriesInfo name="STD" value="7"/>
<seriesInfo name="RFC" value="793"/>
<seriesInfo name="DOI" value="10.17487/RFC0793"/>
</reference>
<reference anchor="RFC1094" target="http://www.rfc-editor.org/info/rfc1094">
<front>
<title>NFS: Network File System Protocol specification</title>
<author initials="B." surname="Nowicki" fullname="B. Nowicki">
<organization/>
</author>
<date year="1989" month="March"/>
<abstract>
<t>This RFC describes a protocol that Sun Microsystems, Inc., and others are using. A new version of the protocol is under development, but others may benefit from the descriptions of the current protocol, and discussion of some of the design issues.</t>
</abstract>
</front>
<seriesInfo name="RFC" value="1094"/>
<seriesInfo name="DOI" value="10.17487/RFC1094"/>
</reference>
<reference anchor="RFC1813" target="http://www.rfc-editor.org/info/rfc1813">
<front>
<title>NFS Version 3 Protocol Specification</title>
<author initials="B." surname="Callaghan" fullname="B. Callaghan">
<organization/>
</author>
<author initials="B." surname="Pawlowski" fullname="B. Pawlowski">
<organization/>
</author>
<author initials="P." surname="Staubach" fullname="P. Staubach">
<organization/>
</author>
<date year="1995" month="June"/>
<abstract>
<t>This paper describes the NFS version 3 protocol. This paper is provided so that people can write compatible implementations. This memo provides information for the Internet community. This memo does not specify an Internet standard of any kind.</t>
</abstract>
</front>
<seriesInfo name="RFC" value="1813"/>
<seriesInfo name="DOI" value="10.17487/RFC1813"/>
</reference>
<reference anchor="RFC5040" target="http://www.rfc-editor.org/info/rfc5040">
<front>
<title>A Remote Direct Memory Access Protocol Specification</title>
<author initials="R." surname="Recio" fullname="R. Recio">
<organization/>
</author>
<author initials="B." surname="Metzler" fullname="B. Metzler">
<organization/>
</author>
<author initials="P." surname="Culley" fullname="P. Culley">
<organization/>
</author>
<author initials="J." surname="Hilland" fullname="J. Hilland">
<organization/>
</author>
<author initials="D." surname="Garcia" fullname="D. Garcia">
<organization/>
</author>
<date year="2007" month="October"/>
<abstract>
<t>This document defines a Remote Direct Memory Access Protocol (RDMAP) that operates over the Direct Data Placement Protocol (DDP protocol). RDMAP provides read and write services directly to applications and enables data to be transferred directly into Upper Layer Protocol (ULP) Buffers without intermediate data copies. It also enables a kernel bypass implementation. [STANDARDS-TRACK]</t>
</abstract>
</front>
<seriesInfo name="RFC" value="5040"/>
<seriesInfo name="DOI" value="10.17487/RFC5040"/>
</reference>
<reference anchor="RFC5041" target="http://www.rfc-editor.org/info/rfc5041">
<front>
<title>Direct Data Placement over Reliable Transports</title>
<author initials="H." surname="Shah" fullname="H. Shah">
<organization/>
</author>
<author initials="J." surname="Pinkerton" fullname="J. Pinkerton">
<organization/>
</author>
<author initials="R." surname="Recio" fullname="R. Recio">
<organization/>
</author>
<author initials="P." surname="Culley" fullname="P. Culley">
<organization/>
</author>
<date year="2007" month="October"/>
<abstract>
<t>The Direct Data Placement protocol provides information to Place the incoming data directly into an upper layer protocol's receive buffer without intermediate buffers. This removes excess CPU and memory utilization associated with transferring data through the intermediate buffers. [STANDARDS-TRACK]</t>
</abstract>
</front>
<seriesInfo name="RFC" value="5041"/>
<seriesInfo name="DOI" value="10.17487/RFC5041"/>
</reference>
<reference anchor="RFC5532" target="http://www.rfc-editor.org/info/rfc5532">
<front>
<title>Network File System (NFS) Remote Direct Memory Access (RDMA) Problem Statement</title>
<author initials="T." surname="Talpey" fullname="T. Talpey">
<organization/>
</author>
<author initials="C." surname="Juszczak" fullname="C. Juszczak">
<organization/>
</author>
<date year="2009" month="May"/>
<abstract>
<t>This document addresses enabling the use of Remote Direct Memory Access (RDMA) by the Network File System (NFS) protocols. NFS implementations historically incur significant overhead due to data copies on end-host systems, as well as other processing overhead. This document explores the potential benefits of RDMA to these implementations and evaluates the reasons why RDMA is especially well-suited to NFS and network file protocols in general. This memo provides information for the Internet community.</t>
</abstract>
</front>
<seriesInfo name="RFC" value="5532"/>
<seriesInfo name="DOI" value="10.17487/RFC5532"/>
</reference>
<reference anchor="RFC5661" target="http://www.rfc-editor.org/info/rfc5661">
<front>
<title>Network File System (NFS) Version 4 Minor Version 1 Protocol</title>
<author initials="S." surname="Shepler" fullname="S. Shepler" role="editor">
<organization/>
</author>
<author initials="M." surname="Eisler" fullname="M. Eisler" role="editor">
<organization/>
</author>
<author initials="D." surname="Noveck" fullname="D. Noveck" role="editor">
<organization/>
</author>
<date year="2010" month="January"/>
<abstract>
<t>This document describes the Network File System (NFS) version 4 minor version 1, including features retained from the base protocol (NFS version 4 minor version 0, which is specified in RFC 3530) and protocol extensions made subsequently. Major extensions introduced in NFS version 4 minor version 1 include Sessions, Directory Delegations, and parallel NFS (pNFS). NFS version 4 minor version 1 has no dependencies on NFS version 4 minor version 0, and it is considered a separate protocol. Thus, this document neither updates nor obsoletes RFC 3530. NFS minor version 1 is deemed superior to NFS minor version 0 with no loss of functionality, and its use is preferred over version 0. Both NFS minor versions 0 and 1 can be used simultaneously on the same network, between the same client and server. [STANDARDS-TRACK]</t>
</abstract>
</front>
<seriesInfo name="RFC" value="5661"/>
<seriesInfo name="DOI" value="10.17487/RFC5661"/>
</reference>
<reference anchor="RFC5662" target="http://www.rfc-editor.org/info/rfc5662">
<front>
<title>Network File System (NFS) Version 4 Minor Version 1 External Data Representation Standard (XDR) Description</title>
<author initials="S." surname="Shepler" fullname="S. Shepler" role="editor">
<organization/>
</author>
<author initials="M." surname="Eisler" fullname="M. Eisler" role="editor">
<organization/>
</author>
<author initials="D." surname="Noveck" fullname="D. Noveck" role="editor">
<organization/>
</author>
<date year="2010" month="January"/>
<abstract>
<t>This document provides the External Data Representation Standard (XDR) description for Network File System version 4 (NFSv4) minor version 1. [STANDARDS-TRACK]</t>
</abstract>
</front>
<seriesInfo name="RFC" value="5662"/>
<seriesInfo name="DOI" value="10.17487/RFC5662"/>
</reference>
<reference anchor="RFC5666" target="http://www.rfc-editor.org/info/rfc5666">
<front>
<title>Remote Direct Memory Access Transport for Remote Procedure Call</title>
<author initials="T." surname="Talpey" fullname="T. Talpey">
<organization/>
</author>
<author initials="B." surname="Callaghan" fullname="B. Callaghan">
<organization/>
</author>
<date year="2010" month="January"/>
<abstract>
<t>This document describes a protocol providing Remote Direct Memory Access (RDMA) as a new transport for Remote Procedure Call (RPC). The RDMA transport binding conveys the benefits of efficient, bulk-data transport over high-speed networks, while providing for minimal change to RPC applications and with no required revision of the application RPC protocol, or the RPC protocol itself. [STANDARDS-TRACK]</t>
</abstract>
</front>
<seriesInfo name="RFC" value="5666"/>
<seriesInfo name="DOI" value="10.17487/RFC5666"/>
</reference>
<reference anchor="RFC5667" target="http://www.rfc-editor.org/info/rfc5667">
<front>
<title>Network File System (NFS) Direct Data Placement</title>
<author initials="T." surname="Talpey" fullname="T. Talpey">
<organization/>
</author>
<author initials="B." surname="Callaghan" fullname="B. Callaghan">
<organization/>
</author>
<date year="2010" month="January"/>
<abstract>
<t>This document defines the bindings of the various Network File System (NFS) versions to the Remote Direct Memory Access (RDMA) operations supported by the RPC/RDMA transport protocol. It describes the use of direct data placement by means of server-initiated RDMA operations into client-supplied buffers for implementations of NFS versions 2, 3, 4, and 4.1 over such an RDMA transport. [STANDARDS-TRACK]</t>
</abstract>
</front>
<seriesInfo name="RFC" value="5667"/>
<seriesInfo name="DOI" value="10.17487/RFC5667"/>
</reference>
<reference anchor="RFC7530" target="http://www.rfc-editor.org/info/rfc7530">
<front>
<title>Network File System (NFS) Version 4 Protocol</title>
<author initials="T." surname="Haynes" fullname="T. Haynes" role="editor">
<organization/>
</author>
<author initials="D." surname="Noveck" fullname="D. Noveck" role="editor">
<organization/>
</author>
<date year="2015" month="March"/>
<abstract>
<t>The Network File System (NFS) version 4 protocol is a distributed file system protocol that builds on the heritage of NFS protocol version 2 (RFC 1094) and version 3 (RFC 1813). Unlike earlier versions, the NFS version 4 protocol supports traditional file access while integrating support for file locking and the MOUNT protocol. In addition, support for strong security (and its negotiation), COMPOUND operations, client caching, and internationalization has been added. Of course, attention has been applied to making NFS version 4 operate well in an Internet environment.</t>
<t>This document, together with the companion External Data Representation (XDR) description document, RFC 7531, obsoletes RFC 3530 as the definition of the NFS version 4 protocol.</t>
</abstract>
</front>
<seriesInfo name="RFC" value="7530"/>
<seriesInfo name="DOI" value="10.17487/RFC7530"/>
</reference>
<reference anchor="IB" target="http://www.infinibandta.org">
<front>
<title>InfiniBand Architecture Specifications</title>
<author>
<organization>InfiniBand Trade Association</organization>
</author>
<date/>
</front>
</reference>
<reference anchor="IBPORT" target="http://www.infinibandta.org">
<front>
<title>IP Addressing Annex</title>
<author>
<organization>InfiniBand Trade Association</organization>
</author>
<date/>
</front>
</reference>
</references>
</back>
</rfc>
| PAFTECH AB 2003-2026 | 2026-04-20 04:58:21 |