One document matched: draft-ietf-intarea-nat-reveal-analysis-01.xml
<?xml version="1.0" encoding="US-ASCII"?>
<!DOCTYPE rfc SYSTEM "rfc2629.dtd">
<?rfc toc="yes"?>
<?rfc tocompact="yes"?>
<?rfc tocdepth="3"?>
<?rfc tocindent="yes"?>
<?rfc symrefs="yes"?>
<?rfc sortrefs="yes"?>
<?rfc comments="yes"?>
<?rfc inline="yes"?>
<?rfc compact="yes"?>
<?rfc subcompact="no"?>
<rfc category="info" docName="draft-ietf-intarea-nat-reveal-analysis-01"
ipr="trust200902">
<front>
<title abbrev="Revealing the origin IP address">Analysis of Solution
Candidates to Reveal a Host Identifier in Shared Address
Deployments</title>
<author fullname="Mohamed Boucadair" initials="M." surname="Boucadair">
<organization>France Telecom</organization>
<address>
<postal>
<street></street>
<city>Rennes</city>
<region></region>
<code>35000</code>
<country>France</country>
</postal>
<email>mohamed.boucadair@orange.com</email>
</address>
</author>
<author fullname="Joe Touch" initials="J." surname="Touch">
<organization>USC/ISI</organization>
<address>
<postal>
<street></street>
<city></city>
<region></region>
<code></code>
<country></country>
</postal>
<email>touch@isi.edu</email>
</address>
</author>
<author fullname="Pierre Levis" initials="P." surname="Levis">
<organization>France Telecom</organization>
<address>
<postal>
<street></street>
<city>Caen</city>
<region></region>
<code>14000</code>
<country>France</country>
</postal>
<email>pierre.levis@orange.com</email>
</address>
</author>
<author fullname="Reinaldo Penno" initials="R." surname="Penno">
<organization>Juniper Networks</organization>
<address>
<postal>
<street>1194 N Mathilda Avenue</street>
<city>Sunnyvale</city>
<region>California</region>
<code>94089</code>
<country>USA</country>
</postal>
<email>rpenno@juniper.net</email>
</address>
</author>
<date day="06" month="March" year="2012" />
<workgroup>INTAREA WG</workgroup>
<keyword>NAT, Host Identifier</keyword>
<abstract>
<t>This document analyzes a set of solution candidates to mitigate some
of the issues encountered when address sharing is used. In particular,
this document focuses on means to reveal a host identifier (HOST_ID)
when a Carrier Grade NAT (CGN) or application proxies are involved in
the path. This host identifier must be unique to each host under the
same shared IP address.</t>
</abstract>
<note title="Requirements Language">
<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">RFC 2119</xref>.</t>
</note>
</front>
<middle>
<section title="Introduction">
<t>As reported in <xref target="RFC6269"></xref>, several issues are
encountered when an IP address is shared among several subscribers.
Examples of such issues are listed below:</t>
<t><?rfc subcompact="yes" ?><list style="symbols">
<t>Implicit identification (Section 13.2 of <xref
target="RFC6269"></xref>)</t>
<t>SPAM (Section 13.3 of <xref target="RFC6269"></xref>)</t>
<t>Blacklisting a mis-behaving user (Section 13.1 of <xref
target="RFC6269"></xref>)</t>
<t>Redirect users with infected machines to a dedicated portal
(Section 5.1 of <xref target="RFC6269"></xref>)</t>
</list></t>
<t>The sole use of the IPv4 address is not sufficient to uniquely
distinguish a host. As a mitigation, it is tempting to investigate means
which would help in disclosing an information to be used by the remote
server as a means to uniquely disambiguate packets of hosts using the
same IPv4 address.</t>
<t>The risk of not mitigating these issues are: OPEX increase for IP
connectivity service providers (costs induced by calls to a hotline),
revenue loss for content providers (loss of users audience), customers
unsatisfaction (low quality of experience, service segregation,
etc.).</t>
<section anchor="problem_space" title="Problem to Be Solved">
<t><list hangIndent="10" style="hanging">
<t hangText="Observation:">Today, some servers use the source IPv4
address as an identifier to treat some incoming connections
differently. Tomorrow, due to the introduction of CGNs (e.g.,
NAT44 <xref target="I-D.ietf-behave-lsn-requirements"></xref>,
NAT64 <xref target="RFC6146"></xref>), that address will be
shared. In particular, when a server receives packets from the
same source address, because this address is shared, the server
does not know which host is the sending host.</t>
<t hangText="Objective:">The server should be able to sort out the
packets by sending host.</t>
<t hangText="Requirement:">The server must have extra information
than the source IP address to differentiate the sending host. We
call HOST_ID this information.</t>
</list></t>
<t>For all solutions analyzed, we provide answers to the following
questions:</t>
<t><?rfc subcompact="no" ?><list hangIndent="5" style="hanging">
<t hangText="What is the HOST_ID?">It must be unique to each host
under the same IP address. It does not need to be globally unique.
Of course, the combination of the (public) IP source address and
the identifier (i.e., HOST_ID) ends up being relatively unique. As
unique as today's 32-bit IPv4 addresses which, today, can change
when a host re-connects.</t>
<t
hangText="Where is the HOST_ID? (which protocol, which field):">If
the HOST_ID is put at the IP level, all packets will have to bear
the identifier. If it is put at a higher connection-oriented
level, the identifier is only needed once in the session
establishment phase (for instance TCP three-way-handshake), then,
all packets received in this session will be attributed to the
HOST_ID designated during the session opening.</t>
<t hangText="Who puts the HOST_ID?">For almost all the analyzed
solutions, the address sharing function injects the HOST_ID. When
there are several address sharing functions in the data path, we
describe to what extent the proposed solution is efficient.
Another option to avoid potential performance degradation is to
let the host inject its HOST_ID but the address sharing function
will check its content (just like an IP anti-spoofing
function).</t>
<t hangText="What are the security considerations?">Security
considerations are common to all analyzed solutions (see <xref
target="Security"></xref>). Privacy-related aspect are discussed
in <xref target="privacy"></xref>.</t>
</list></t>
</section>
<section title="IPv6 May Also Be Concerned">
<t>Some of the issues mentioned in <xref
target="problem_space"></xref> are independent of IPv4 vs. IPv6. Even
in IPv6, address sharing can be used for a variety of reasons (e.g.,
to hide network topology, to defeat hosts from offering network
services directly, etc.).</t>
<t>A solution to reveal HOST_ID is also needed in IPv6 deployment.</t>
</section>
<section title="Purpose and Scope">
<t>The purpose of this document is not to argue in favor of mandating
the use of a HOST_ID but to identify encountered issues, proposed
solutions and their limitations.</t>
<t>The purpose of this document is to analyze a set of solution
candidates and to assess to what extent they solve the problem (see
<xref target="problem_space"></xref>). Below are listed the solutions
analyzed in the document:</t>
<t><?rfc subcompact="yes" ?><list style="symbols">
<t>Use the Identification field of IP header (denoted as IP-ID,
<xref target="ip_id"></xref>).</t>
<t>Define a new IP option (<xref target="ip_option"></xref>).</t>
<t>Assign port sets (<xref target="server"></xref>).</t>
<t>Use ICMP (<xref target="server"></xref>).</t>
<t>Define a new TCP Option (<xref target="tcp"></xref>).</t>
<t>Enable Proxy Protocol <xref target="PROXY">(</xref>).</t>
<t>Activate HIP (<xref target="hip"></xref>).</t>
<t>Inject application headers (<xref target="xff"></xref>).</t>
</list></t>
<t><?rfc subcompact="no" ?></t>
</section>
</section>
<section title="Synthesis">
<t>The following Table 1 summarizes the approaches analyzed in this
document.</t>
<t><?rfc subcompact="yes" ?></t>
<t><list style="symbols">
<t>"Success ratio" indicates the ratio of successful communications
when the option is used. Provided figures are inspired from the
results documented in <xref target="Options"></xref>.</t>
<t>"Deployable today" indicates if the solution can be generalized
without any constraint on current architectures and practices.</t>
<t>"Possible Perf Impact" indicates the level of expected
performance degradation. The rationale behind the indicated
potential performance degradation is whether the injection requires
some treatment at the IP level or not.</t>
<t>"OS TCP/IP Modif" indicates whether a modification of the OS
TCP/IP stack is required at the server side.</t>
</list></t>
<t><?rfc subcompact="no" ?></t>
<figure align="center" anchor="table"
title="Table 1: Summary of analyzed solutions.">
<artwork><![CDATA[ +------+------+-------+-------+-------+------+-----+------+
| IP | TCP | IP-ID | HTTP | Proxy | Port | HIP | ICMP |
|Option|Option| | Header| | Set | | |
| | | | (XFF) | | | | |
----------+------+------+-------+-------+-------+------+-----+------+
UDP | Yes | No | Yes | No | No | Yes | | Yes |
----------+------+------+-------+-------+-------+------+-----+------+
TCP | Yes | Yes | Yes | No | Yes | Yes | | Yes |
----------+------+------+-------+-------+-------+------+-----+------+
HTTP | Yes | Yes | Yes | Yes | Yes | Yes | | Yes |
----------+------+------+-------+-------+-------+------+-----+------+
Encrypted | Yes | Yes | Yes | No | Yes | Yes | | Yes |
Traffic | | | | | | | | |
----------+------+------+-------+-------+-------+------+-----+------+
Success | 30% | 99% | 100% | 100% | Low | 100% |Low | ~100%|
Ratio | | | | | | | | (6) |
----------+------+------+-------+-------+-------+------+-----+------+
Possible | High | Med | Low | Med | High | No | N/A | High |
Perf | | to | to | to | | | | |
Impact | | High | Med | High | | | | |
----------+------+------+-------+-------+-------+------+-----+------+
OS TCP/IP | Yes | Yes | Yes | No | No | No | | Yes |
Modif | | | | | | | | |
----------+------+------+-------+-------+-------+------+-----+------+
Deployable| Yes | Yes | Yes | Yes | No | Yes | No | Yes |
Today | | | | | | | | |
----------+------+------+-------+-------+-------+------+-----+------+
Notes | | | (1) | (2) | | (1) | (4) | (7) |
| | | | | | (3) | (5) | |
----------+------+------+-------+-------+-------+------+-----+------+
Notes:
(1) Requires mechanism to advertise NAT is participating in this
scheme (e.g., DNS PTR record).
(2) This solution is widely deployed.
(3) When the port set is not advertised, the solution is less
efficient for third-party services.
(4) Requires the client and the server to be HIP-compliant and HIP
infrastructure to be deployed.
(5) If the client and the server are HIP-enabled, the address
sharing function does not need to insert a host-hint. If the
client is not HIP-enabled, designing the device that performs
address sharing to act as a UDP/TCP-HIP relay is not viable.
(6) Implementation specific.
(7) The solution is inefficient is various scenarios as discussed
in Section 3.
]]></artwork>
</figure>
<t><?rfc subcompact="no" ?>According to the above table and the analysis
elaborated in <xref target="Analysis"></xref>:<list style="symbols">
<t>IP Option, IP-ID and Proxy Protocol proposals are broken;</t>
<t>HIP is not largely deployed;</t>
<t>The use of Port Set may contradict the port randomization <xref
target="RFC6056"></xref> requirement identified in <xref
target="RFC6269"></xref>. This solution can be used by a service
provider for the delivery of its own service offerings relying on
implicit identification.</t>
<t>XFF is de facto standard deployed and supported in operational
networks (e.g., HTTP Severs, Load-Balancers, etc.).</t>
<t>From an application standpoint, the TCP Option is superior to XFF
since it is not restricted to HTTP. Nevertheless XFF is compatible
with the presence of address sharing and load-balancers in the
communication path. To provide a similar functionality, the TCP
Option may be extended to allow conveying a list of IP addresses and
port numbers to not lose the source IP address in the presence of
load-balancers. Another alternative is to combine the usage of both
the HOST_ID TCP Option and XFF. Note that TCP Option requires the
modification of the OS TCP/IP stack of remote servers; which can be
seen as a blocking point.</t>
</list>For all HOST_ID proposals, the following recommendations are
made: <list style="hanging">
<t hangText="Uniqueness of identifiers in HOST_ID:">It is
RECOMMENDED that HOST_IDs be limited to providing local uniqueness
rather than global uniqueness.</t>
<t hangText="Refresh rate of HOST_ID:">Address sharing function
SHOULD NOT use permanent HOST_ID values.</t>
<t hangText="Manipulate HOST_IDs:">Address sharing function SHOULD
be able to strip, re-write and add HOST_ID fields.</t>
<t hangText="Interference between HOST_IDs:">An address sharing
function, able to inject HOST_IDs in several layers, SHOULD reveal
subsets of the same information (e.g., full IP address, lower 16
bits of IP address, etc.).</t>
</list></t>
<t></t>
</section>
<section anchor="Analysis" title="Solutions Analysis">
<t></t>
<section anchor="ip_id"
title="Use the Identification Field of IP Header (IP-ID)">
<t></t>
<section title="Description">
<t>IP-ID (Identification field of IP header) can be used to insert
an information which uniquely distinguishes a host among those
sharing the same IPv4 address. An address sharing function can
re-write the IP-ID field to insert a value unique to the host (16
bits are sufficient to uniquely disambiguate hosts sharing the same
IP address). Note that this field is not altered by some NATs; hence
some side effects such as counting hosts behind a NAT as reported in
<xref target="Count"></xref>.</t>
<t>A variant of this approach relies upon the format of certain
packets, such as TCP SYN, where the IP-ID can be modified to contain
a 16 bit HOST_ID. Address sharing devices performing this function
would require to indicate they are performing this function out of
band, possibly using a special DNS record.</t>
</section>
<section title="Analysis">
<t>This usage is not compliant with what is recommended in <xref
target="I-D.ietf-intarea-ipv4-id-update"></xref>.</t>
</section>
</section>
<section anchor="ip_option" title="Define an IP Option">
<t></t>
<section title="Description">
<t>A solution alternative to convey the HOST_ID is to define an IP
option <xref target="RFC0791"></xref>. HOST_ID IP option can be
inserted by the address sharing function to uniquely distinguish a
host among those sharing the same IP address. An example of such
option is documented in <xref
target="I-D.chen-intarea-v4-uid-header-option"></xref>. This IP
option allows to convey an IPv4 address, an IPv6 prefix, a GRE key,
IPv6 Flow Label, etc.</t>
<t>Another way for using IP option has been described in Section 4.6
of <xref target="RFC3022"></xref>.</t>
</section>
<section title="Analysis">
<t>Unlike the solution presented in <xref target="tcp"></xref>, this
proposal can apply for any transport protocol. Nevertheless, it is
widely known that routers (and other middleboxes) filter IP options.
IP packets with IP options can be dropped by some IP nodes. Previous
studies demonstrated that "IP Options are not an option" (Refer to
<xref target="Not_An_Option"></xref>, <xref
target="Options"></xref>).</t>
<t>As a conclusion, using an IP option to convey a host-hint is not
viable.</t>
</section>
</section>
<section anchor="server" title="Assign Port Sets">
<t></t>
<section title="Description">
<t>This solution does not require any action from the address
sharing function to disclose a host identifier. Instead of assuming
all transport ports are associated with one single host, each host
under the same external IP address is assigned a restricted port
set. These port sets are then advertised to remote servers using
off-line means. This announcement is not required for the delivery
of internal services (i.e., offered by the service provider
deploying the address sharing function) relying on implicit
identification.</t>
<t>Port sets assigned to hosts may be static or dynamic.</t>
<t>Port set announcements to remote servers do not require to reveal
the identity of individual hosts but only to advertise the enforced
policy to generate non-overlapping port sets (e.g., the transport
space associated with an IP address is fragmented to contiguous
blocks of 2048 port numbers).</t>
</section>
<section title="Analysis">
<t>The solution does not require defining new fields nor options; it
is policy-based.</t>
<t>The solution may contradict the port randomization as identified
in <xref target="RFC6269"></xref>. A mitigation would be to avoid
assigning static port sets to individual hosts.</t>
<t>The method is convenient for the delivery of services offered by
the service provider offering also the IP connectivity service.</t>
</section>
</section>
<section anchor="icmp" title="Use ICMP">
<t></t>
<section title="Description">
<t>Another alternative is to convey the HOST_ID using a separate
notification channel than the packets issued to invoke the service.
</t>
<t>An implementation example is defined in <xref
target="I-D.yourtchenko-nat-reveal-ping"></xref>. This solution
relies on a mechanism where the address sharing function
encapsulates the necessary differentiating information into an ICMP
Echo Request packet that it sends in parallel with the initial
session creation (e.g., SYN). The information included in the ICMP
Request Data portion describes the five-tuples as seen on both of
the sides of the address sharing function. </t>
</section>
<section title="Analysis">
<t><?rfc subcompact="yes" ?><list style="symbols">
<t>This ICMP proposal is valid for both UDP and TCP. Address
sharing function may be configurable with the transport protocol
which is allowed to trigger those ICMP messages.</t>
<t>A hint should be provided to the ultimate server (or
intermediate nodes) an ICMP Echo Request conveys a HOST_ID. This
may be implemented using magic numbers.</t>
<t>Even if ICMP packets are blocked in the communication path,
the user connection does not have to be impacted.</t>
<t>Some implementations requiring to delay the establishment of
a session until receiving the companion ICMP Echo Request, may
lead to some user experience degradation. </t>
<t>Because of the presence of load-balancers in the path, the
ultimate server receiving the SYN packet may not be the one
which may receive the ICMP message conveying the HOST_ID. </t>
<t>Because of the presence of load-balancers in the path, the
port number assigned by address sharing may be lost. Therefore
the mapping information conveyed in the ICMP may not be
sufficient to associate a SYN packet with a received ICMP.</t>
<t>The proposal is not compatible with the presence of cascaded
NAT.</t>
<t>The ICMP proposal will add a traffic overhead for both the
server and the address sharing device. </t>
<t>The ICMP proposal is similar to other mechanisms (e.g.,
syslog, netflow) for reporting dynamic mappings to a mediation
platform (mainly for legal traceability purposes). Performance
degradation are likely to be experienced by address sharing
functions because ICMP messages are to be sent in particular for
each new instantiated mapping (and also even if the mapping
exists). </t>
<t>In some scenarios (e.g., Fixed-Mobile Convergence, Open WiFi,
etc.), HOST_ID should be interpreted by intermediate devices
which embed Policy Enforcement Points (PEP, <xref
target="RFC2753"></xref>) responsible for granting access to
some services. These PEPs need to inspect all received packets
in order to find the companion (traffic) messages to be
correlated with ICMP messages conveying HOST_IDs. This induces
more complexity to these intermediate devices.</t>
</list></t>
<t><?rfc subcompact="no" ?></t>
</section>
</section>
<section anchor="tcp" title="Define a TCP Option">
<t></t>
<section title="Description">
<t>HOST_ID may be conveyed in a dedicated TCP Option. An example is
specified in <xref target="I-D.wing-nat-reveal-option"></xref> which
defines a new TCP Option called USER_HINT. This option encloses the
TCP client's identifier (e.g., the lower 16 bits of their IPv4
address, their VLAN ID, VRF ID, subscriber ID). The address sharing
device inserts this TCP Option into the TCP SYN packet.</t>
</section>
<section title="Analysis">
<t>Using a new TCP Option to convey the HOST_ID does not require any
modification to the applications but it is applicable only for
TCP-based applications. Applications relying on other transport
protocols are therefore left unsolved.</t>
<t><xref target="I-D.wing-nat-reveal-option"></xref> discusses the
interference with other TCP Options.</t>
<t>The risk related to handling a new TCP Option is low as measured
in <xref target="Options"></xref>. <xref
target="I-D.abdo-hostid-tcpopt-implementation"></xref> provides a
detailed implementation and experimentation report of HOST_ID TCP
Option. <xref target="I-D.abdo-hostid-tcpopt-implementation"></xref>
investigated in depth the impact of activation HOST_ID in host,
address sharing function and the enforcement of policies at the
server side. <xref
target="I-D.abdo-hostid-tcpopt-implementation"></xref> reports a
failure ratio of 0,103% among top 100000 websites.</t>
<t>Some downsides have been raised against defining a TCP Option to
reveal a host identity:</t>
<t><list style="symbols">
<t>Conveying an IP address in a TCP Option may be seen as a
violation of OSI layers but since IP addresses are already used
for the checksum computation, this is not seen as a blocking
point. Moreover, updated version of <xref
target="I-D.wing-nat-reveal-option"></xref> does not allow
anymore to convey an IP address (the HOST_ID is encoded in
16bits).</t>
<t>TCP Option space is limited, and might be consumed by the TCP
client. Earlier versions of <xref
target="I-D.wing-nat-reveal-option"></xref> discuss two
approaches to sending the HOST_ID: sending the HOST_ID in the
TCP SYN (which consumes more bytes in the TCP header of the TCP
SYN) and sending the HOST_ID in a TCP ACK (which consumes only
two bytes in the TCP SYN). Content providers may find it more
desirable to receive the HOST_ID in the TCP SYN, as that more
closely preserves the HOST_ID received in the source IP address
as per current practices. It is more complicated to implement
sending the HOST_ID in a TCP ACK, as it can introduce MTU issues
if the ACK packet also contains TCP data, or a TCP segment is
lost. The latest specification of the HOST_ID TCP Option,
documented at <xref target="I-D.wing-nat-reveal-option"></xref>,
allows only to enclose the HOST_ID in the TCP SYN packet.</t>
<t>When there are several NATs in the path, the original HOST_ID
may be lost. In such case, the procedure may not be
efficient.</t>
<t>Interference with current usages such as X-Forwarded-For (see
<xref target="xff"></xref>) should be elaborated to specify the
behavior of servers when both options are used; in particular
specify which information to use: the content of the TCP Option
or what is conveyed in the application headers.</t>
<t>When load-balancers or proxies are in the path, this option
does not allow to preserve the original source IP address and
source port. Preserving such information is required for logging
purposes for instance (e.g., <xref target="RFC6302"></xref>) .
<xref target="I-D.abdo-hostid-tcpopt-implementation"></xref>
defines a TCP Option which allows to reveal various combinations
of source information (e.g., source port, source port and source
IP address, source IPv6 prefix, etc.).</t>
</list>More discussion about issues raised when extending TCP can
be found at <xref target="ExtendTCP"></xref>.</t>
</section>
</section>
<section anchor="PROXY" title="PROXY Protocol">
<t></t>
<section title="Description">
<t>The solution, referred to as Proxy Protocol <xref
target="Proxy"></xref>, does not require any application-specific
knowledge. The rationale behind this solution is to prepend each
connection with a line reporting the characteristics of the other
side's connection as shown in the example depicted in <xref
target="proxy_eg"></xref>:</t>
<t><figure align="center" anchor="proxy_eg"
title="Example of PROXY conection report">
<artwork><![CDATA[ PROXY TCP4 192.0.2.1 192.0.2.15 56324 443\r\n]]></artwork>
</figure></t>
<t>Upon receipt of a message conveying this line, the server removes
the line. The line is parsed to retrieve the transported protocol.
The content of this line is recorded in logs and used to enforce
policies.</t>
</section>
<section title="Analysis">
<t>This solution can be deployed in a controlled environment but it
can not be deployed to all access services available in the
Internet. If the remote server does not support the Proxy Protocol,
the session will fail. Other complications will raise due to the
presence of firewalls for instance.</t>
<t>As a consequence, this solution is broken and can not be
recommended.</t>
</section>
</section>
<section anchor="hip" title="Host Identity Protocol (HIP)">
<t></t>
<section title="Description">
<t><xref target="RFC5201"></xref> specifies an architecture which
introduces a new namespace to convey an identity information.</t>
</section>
<section title="Analysis">
<t>This solution requires both the client and the server to support
HIP <xref target="RFC5201"></xref>. Additional architectural
considerations are to be taken into account such as the key
exchanges, etc.</t>
<t>If the address sharing function is required to act as a
UDP/TCP-HIP relay, this is not a viable option.</t>
</section>
</section>
<section anchor="xff" title="Inject Application Headers">
<t></t>
<section title="Description">
<t>Another option is to not require any change at the transport nor
the IP levels but to convey at the application payload the required
information which will be used to disambiguate hosts. This format
and the related semantics depend on its application (e.g., HTTP,
SIP, SMTP, etc.).</t>
<t>For HTTP, the X-Forwarded-For (XFF) or Forwarded-For (<xref
target="I-D.ietf-appsawg-http-forwarded"></xref>) headers can be
used to display the original IP address when an address sharing
device is involved. Service Providers operating address sharing
devices can enable the feature of injecting the XFF header which
will enclose the original IPv4 address or the IPv6 prefix part (see
the example shown in <xref target="xff_eg"></xref>). The address
sharing device has to strip all included XFF headers before
injecting their own. Servers may rely on the contents of this field
to enforce some policies such as blacklisting misbehaving users.
Note that XFF can also be logged by some servers (this is for
instance supported by Apache).</t>
<t><figure align="center" anchor="xff_eg"
title="Example of Forwarded-For">
<artwork><![CDATA[Forwarded: for=192.0.2.1,for=[2001:db8::1]
Forwarded: proto=https;by=192.0.2.15]]></artwork>
</figure></t>
</section>
<section title="Analysis">
<t>Not all applications impacted by the address sharing can support
the ability to disclose the original IP address. Only a subset of
protocols (e.g., HTTP) can rely on this solution.</t>
<t>For the HTTP case, to prevent users injecting invalid HOST_IDs,
an initiative has been launched to maintain a list of trusted ISPs
using XFF: See for example the list available at: <xref
target="Trusted_ISPs"></xref> of trusted ISPs as maintained by
Wikipedia. If an address sharing device is on the trusted XFF ISPs
list, users editing Wikipedia located behind the address sharing
device will appear to be editing from their "original" IP address
and not from the NATed IP address. If an offending activity is
detected, individual hosts can be blacklisted instead of all hosts
sharing the same IP address.</t>
<t>XFF header injection is a common practice of load balancers. When
a load balancer is in the path, the original content of any included
XFF header should not be stripped. Otherwise the information about
the "origin" IP address will be lost.</t>
<t>When several address sharing devices are crossed, XFF header can
convey the list of IP addresses (e.g., <xref
target="xff_eg"></xref>). The origin HOST_ID can be exposed to the
target server.</t>
<t>XFF also introduces some implementation complexity if the HTTP
packet is at or close to the MTU size.</t>
<t>It has been reported that some "poor" implementation may
encounter some parsing issues when injecting XFF header.</t>
<t>For encrypted HTTP traffic, injecting XFF header may be
broken.</t>
</section>
</section>
</section>
<section anchor="privacy" title="HOST_ID and Privacy">
<t>IP address sharing is motivated by a number of different factors. For
years, many network operators have conserved the use of public IPv4
addresses by making use of Customer Premises Equipment (CPE) that
assigns a single public IPv4 address to all hosts within the customer's
local area network and uses NAT <xref target="RFC3022"></xref> to
translate between locally unique private IPv4 addresses and the CPE's
public address. With the exhaustion of IPv4 address space, address
sharing between customers on a much larger scale is likely to become
much more prevalent. While many individual users are unaware of and
uninvolved in decisions about whether their unique IPv4 addresses get
revealed when they send data via IP, some users realize privacy benefits
associated with IP address sharing, and some may even take steps to
ensure that NAT functionality sits between them and the public Internet.
IP address sharing makes the actions of all users behind the NAT
function unattributable to any single host, creating room for abuse but
also providing some identity protection for non-abusive users who wish
to transmit data with reduced risk of being uniquely identified.</t>
<t>The proposals considered in this document add a measure of uniqueness
back to hosts that share a public IP address. The extent of that
uniqueness depends on which information is included in the HOST_ID.</t>
<t>The volatility of the HOST_ID information is similar to the source IP
address: a distinct HOST_ID may be used by the address sharing function
when the host reboots or gets a new internal IP address. As with
persistent IP addresses, persistent HOST_IDs facilitate user tracking
over time.</t>
<t>As a general matter, the HOST_ID proposals do not seek to make hosts
any more identifiable than they would be if they were using a public,
non-shared IP address. However, depending on the solution proposal, the
addition of HOST_ID information may allow a device to be fingerprinted
more easily than it otherwise would be. Should multiple solutions be
combined (e.g., TCP Option and XFF) that include different pieces of
information in the HOST_ID, fingerprinting may become even easier.</t>
<t>The trust placed in the information conveyed in the HOST_ID is likely
to be the same as for current practices with source IP addresses. In
that sense, a HOST_ID can be spoofed as this is also the case for
spoofing an IP address. Furthermore, users of network-based anonymity
services (like Tor) may be capable of stripping HOST_ID information
before it reaches its destination.</t>
<t>For more discussion about privacy, refer to <xref
target="RFC6462"></xref>.</t>
</section>
<section anchor="IANA" title="IANA Considerations">
<t>This document does not require any action from IANA.</t>
</section>
<section anchor="Security" title="Security Considerations">
<t>The same security concerns apply for the injection of an IP option,
TCP Option and application-related content (e.g., XFF) by the address
sharing device. If the server trusts the content of the HOST_ID field, a
third party user can be impacted by a misbehaving user to reveal a
"faked" original IP address.</t>
</section>
<section title="Acknowledgments">
<t>Many thanks to D. Wing and C. Jacquenet for their review, comments
and inputs.</t>
<t>Thanks also to P. McCann, T. Tsou, Z. Dong, B. Briscoe, T. Taylor, M.
Blanchet, D. Wing and A. Yourtchenko for the discussions in Prague.</t>
<t>Some of the issues related to defining a new TCP Option have been
raised by L. Eggert.</t>
<t>Privacy text is provided by A. Cooper.</t>
</section>
</middle>
<back>
<references title="Normative References">
<?rfc include="reference.RFC.2119"?>
<?rfc include='reference.RFC.0791'?>
<?rfc include='reference.RFC.6056'?>
<?rfc include='reference.RFC.3022'?>
</references>
<references title="Informative References">
<?rfc include='reference.I-D.wing-nat-reveal-option'?>
<?rfc include='reference.I-D.ietf-appsawg-http-forwarded'?>
<?rfc include='reference.RFC.5201'?>
<?rfc include='reference.RFC.6462'?>
<?rfc include='reference.RFC.2753'?>
<?rfc include='reference.RFC.6302'?>
<?rfc include='reference.I-D.yourtchenko-nat-reveal-ping'?>
<?rfc include='reference.I-D.chen-intarea-v4-uid-header-option'?>
<?rfc include='reference.RFC.6146'?>
<?rfc include='reference.I-D.abdo-hostid-tcpopt-implementation'?>
<?rfc include='reference.I-D.ietf-behave-lsn-requirements'?>
<?rfc include='reference.RFC.6269'?>
<?rfc include='reference.I-D.ietf-intarea-ipv4-id-update'?>
<reference anchor="Not_An_Option"
target="http://www.eecs.berkeley.edu/Pubs/TechRpts/2005/EECS-2005-24.html">
<front>
<title>IP options are not an option</title>
<author fullname="" initials="" surname="">
<organization>R. Fonseca, G. Porter, R. Katz, S. Shenker, and I.
Stoica,</organization>
</author>
<date year="2005" />
</front>
</reference>
<reference anchor="Options"
target="http://conferences.sigcomm.org/imc/2004/papers/p336-medina.pdf">
<front>
<title>Measuring Interactions Between Transport Protocols and
Middleboxes</title>
<author fullname="" initials="" surname="">
<organization>Alberto Medina, Mark Allman, Sally
Floyd</organization>
</author>
<date year="2005" />
</front>
</reference>
<reference anchor="ExtendTCP"
target="http://nrg.cs.ucl.ac.uk/mjh/tmp/mboxes.pdf">
<front>
<title>Is it still possible to extend TCP?</title>
<author fullname="" initials="" surname="">
<organization>Honda, M., Nishida, Y., Raiciu, C., Greenhalgh, A.,
Handley, M. and H. Tokuda,</organization>
</author>
<date month="November" year="2011" />
</front>
</reference>
<reference anchor="Trusted_ISPs"
target="http://meta.wikimedia.org/wiki/XFF_project#Trusted_XFF_list">
<front>
<title>Trusted XFF list</title>
<author>
<organization></organization>
</author>
<date />
</front>
</reference>
<reference anchor="Count"
target="http://www.cs.columbia.edu/~smb/papers/fnat.pdf">
<front>
<title>A technique for counting NATted hosts</title>
<author fullname="Steve Belloven">
<organization></organization>
</author>
<date />
</front>
</reference>
<reference anchor="Proxy"
target="http://haproxy.1wt.eu/download/1.5/doc/proxy-protocol.txt">
<front>
<title>The PROXY protocol</title>
<author fullname="Willy Tarreau" initials="W." surname="Tarreau">
<organization></organization>
</author>
<date month="November" year="2010" />
</front>
</reference>
</references>
</back>
</rfc>
| PAFTECH AB 2003-2026 | 2026-04-24 01:40:17 |