One document matched: draft-boucadair-pppext-portrange-option-08.xml


<?xml version="1.0" encoding="US-ASCII"?>
<!DOCTYPE rfc SYSTEM "rfc2629.dtd">
<?rfc toc="yes"?>
<?rfc tocompact="yes"?>
<?rfc tocdepth="4"?>
<?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-boucadair-pppext-portrange-option-08"
     ipr="trust200811">
  <front>
    <title abbrev="Port Range IPCP Options">Huawei Port Range Configuration
    Options for PPP IPCP</title>

    <author fullname="Mohamed Boucadair" initials="M." surname="Boucadair">
      <organization>France Telecom</organization>

      <address>
        <postal>
          <street></street>

          <city>Rennes</city>

          <code>35000</code>

          <country>France</country>
        </postal>

        <email>mohamed.boucadair@orange-ftgroup.com</email>
      </address>
    </author>

    <author fullname="Pierre Levis" initials="P." surname="Levis">
      <organization>France Telecom</organization>

      <address>
        <postal>
          <street></street>

          <city>Caen</city>

          <country>France</country>
        </postal>

        <email>pierre.levis@orange-ftgroup.com</email>
      </address>
    </author>

    <author fullname="Gabor Bajko" initials="G." surname="Bajko">
      <organization>Nokia</organization>

      <address>
        <postal>
          <street></street>
        </postal>

        <email>gabor(dot)bajko(at)nokia(dot)com</email>
      </address>
    </author>

    <author fullname="Teemu Savolainen" initials="T." surname="Savolainen">
      <organization abbrev="Nokia">Nokia</organization>

      <address>
        <email>teemu.savolainen@nokia.com</email>
      </address>
    </author>

    <author fullname="Tina Tsou" initials="T." surname="Tsou">
      <organization>Huawei Technologies (USA)</organization>

      <address>
        <postal>
          <street>2330 Central Expressway</street>

          <city>Santa Clara</city>

          <region>CA</region>

          <code>95050</code>

          <country>USA</country>
        </postal>

        <phone>+1 408 330 4424</phone>

        <email>tina.tsou.zouting@huawei.com</email>
      </address>
    </author>

    <date day="31" month="August" year="2011" />

    <keyword>Port Range, IPv4 Address Exhaustion, IPv4 service continuity,
    IPv6, A+P</keyword>

    <abstract>
      <t>This document defines two Huawei IPCP (IP Configuration Protocol)
      Options used to convey a set of ports. These options can be used in the
      context of port range-based solutions or NAT-based ones for port
      delegation and forwarding purposes.</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>Within the context of IPv4 address depletion, several solutions have
      been investigated to share IPv4 addresses. Two flavors can be
      distinguished: NAT-based solutions (a.k.a., Carrier Grade NAT (CGN,
      <xref target="I-D.ietf-behave-lsn-requirements"></xref>)) or port range
      based ones (e.g., <xref target="RFC6346"></xref> <xref
      target="I-D.boucadair-port-range"></xref><xref
      target="I-D.despres-sam"></xref>). Port range-based solutions do not
      require an additional NAT level in the service provider's domain.
      Several means may be used to convey Port Range information.</t>

      <t>This document defines the notion of Port Mask which is generic and
      flexible. Several allocation schemes may be implemented when using a
      Port Mask. It proposes a basic mechanism that allows the allocation of a
      unique port range to a requesting client. This document defines Huawei
      IPCP options to be used to carry Port Range information.</t>

      <t>IPv4 address exhaustion is only provided as an example of the usage
      of the PPP IPCP Options defined in this document. In particular, Port
      Range Options may be used independently of the presence of IP-Address
      IPCP Option.</t>

      <t>This document adheres to the consideration defined in <xref
      target="RFC2153"></xref>.</t>

      <t>This document is not a product of pppext working group.</t>

      <t>Note that IPR disclosures apply to this document (see
      https://datatracker.ietf.org/ipr/).</t>

      <section title="Use Cases">
        <t>Port Range Options can be used in port range-based solutions (e.g.,
        <xref target="RFC6346"></xref>) or in a CGN-based solution. These
        options can be used in a CGN context to bypass the NAT (i.e., for
        transparent NAT traversal and avoid involving several NAT levels in
        the path) or to delegate one or a set of ports to the requesting
        client (e.g., avoid ALG (Application Level Gateway) or for port
        forwarding).</t>

        <t>Architecture considerations are out of scope of this document.
        <xref target="RFC6346"></xref> specifies an architecture using the
        options defined in this document.</t>
      </section>

      <section title="Terminology">
        <t>To differentiate between a Port Range containing a contiguous span
        of port numbers and a Port Range with non contiguous and possibly
        random port numbers, the following denominations are used:</t>

        <t><list style="symbols">
            <t>Contiguous Port Range: a set of port values which form a
            contiguous sequence.</t>

            <t>Non Contiguous Port Range: a set of port values which does not
            form a contiguous sequence.</t>

            <t>Random Port Range: a cryptographically random set of port
            values.</t>
          </list>Unless explicitly mentioned, Port Mask refers to the tuple
        (Port Range Value, Port Range Mask).</t>

        <t>In addition, this document makes use of the following terms:</t>

        <t><list style="symbols">
            <t>Delegated port or delegated port range: a port or a range of
            ports belonging to an IP address managed by an upstream device
            (such as NAT), which are delegated to a client for use as source
            address and port when sending packets.</t>

            <t>Forwarded port or forwarder port range: a port or a range of
            ports belonging to an IP address managed by an upstream device
            such as (NAT), which is/are statically mapped to the internal IP
            address of the client and same port number of the client.</t>
          </list>This memo uses the same terminology as per <xref
        target="RFC1661"></xref>.</t>
      </section>
    </section>

    <section anchor="values" title="Port Range Options">
      <t>This section defines the IPCP Option for Port Range delegation. The
      format of vendor-specific options is defined in <xref
      target="RFC2153"></xref>. Below are provided the values to be conveyed
      when the Port Range Option is used:<list style="symbols">
          <t>Organizationally Unique Identifier (OUI): This field is set to
          781DBA (hex).</t>

          <t>Kind: This field is set to F0 (hex).</t>

          <t>Value: The content of this field is specified in <xref
          target="value_1"></xref> and <xref target="value_2"></xref>.</t>
        </list></t>

      <section anchor="value_1"
               title="Description of Port Range Value and Port Range Mask">
        <t>The Port Range Value and Port Range Mask are used to specify one
        range of ports (contiguous or not contiguous) pertaining to a given IP
        address. Concretely, Port Range Mask and Port Range Value are used to
        notify a remote peer about the Port Mask to be applied when selecting
        a port value as a source port. The Port Range Value is used to infer a
        set of allowed port values. A Port Range Mask defines a set of ports
        that all have in common a subset of pre-positioned bits. This set of
        ports is also called Port Range.</t>

        <t>Two port numbers are said to belong to the same Port Range if and
        only if, they have the same Port Range Mask.</t>

        <t>A Port Mask is composed of a Port Range Value and a Port Range
        Mask:</t>

        <t><list style="symbols">
            <t>The Port Range Value indicates the value of the significant
            bits of the Port Mask. The Port Range Value is coded as follows:
            <list style="symbols">
                <t>The significant bits may take a value of 0 or 1.</t>

                <t>All the other bits (a.k.a., non significant ones) are set
                to 0.</t>
              </list></t>

            <t>The Port Range Mask indicates, by the bit(s) set to 1, the
            position of the significant bits of the Port Range Value.</t>
          </list></t>

        <t>This IPCP Configuration Option provides a way to negotiate the Port
        Range to be used on the local end of the link. It allows the sender of
        the Configure-Request message to state which Port Range associated
        with a given IP address is desired, or to request the peer to provide
        the configuration. The peer can provide this information by NAKing the
        option, and returning a valid Port Range (i.e., (Port Range Value,
        Port Range Mask)).</t>

        <t>When a peer issues a request enclosing IPCP Port Range Option, and
        if the server does not support this option, the Port Range Option is
        rejected by the server.</t>

        <t>The set of ports conveyed in an IPCP Port Range Option belong to
        all transport protocols.</t>

        <t>The Port Range IPCP option adheres to the format defined in Section
        2.1 of <xref target="RFC2153"></xref>. The "value" field of the option
        defined in <xref target="RFC2153"></xref> when conveying Port Range
        IPCP Option is provided in <xref target="Figure_1"></xref>.</t>

        <t><figure anchor="Figure_1"
            title="Format of the Port Range IPCP Option">
            <preamble></preamble>

            <artwork><![CDATA[    0                   1                   2                   3
   0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |M|          Reserved           |      Port Range Value         |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |      Port Range Mask          |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

]]></artwork>

            <postamble></postamble>
          </figure></t>

        <t><list style="symbols">
            <t>M: mode bit. It indicates the mode the port range is allocated
            for. A value of zero indicates the port ranges are delegated,
            while a value of 1 indicates the port ranges are port
            forwarded.</t>

            <t>Port Range Value (PRV): PRV indicates the value of the
            significant bits of the Port Mask. By default, no PRV is assigned.
            </t>

            <t>Port Range Mask (PRM): Port Range Mask indicates the position
            of the bits which are used to build the Port Range Value. By
            default, no PRM value is assigned. The 1 values in the Port Range
            Mask indicate by their position the significant bits of the Port
            Range Value.</t>
          </list></t>

        <t><xref target="Figure_3"></xref> provides an example of the
        resulting Port Range:</t>

        <t>- Port Range Mask is set to 0001010000000000 (5120) and</t>

        <t>- Port Range Value is set to 0000010000000000 (1024).<figure
            anchor="Figure_3"
            title="Example of Port Range Mask and Port Range Value">
            <preamble></preamble>

            <artwork><![CDATA[
0                   1                                  
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5  
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0| Port Range Mask
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
       |   | 
       |   | (two significant bits)
       v   v
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0| Port Range Value
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+  

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|x x x 0 x 1 x x x x x x x x x x| Usable ports (x may be set to 0 or 1)
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

]]></artwork>

            <postamble></postamble>
          </figure>Port values belonging to this Port Range must have the 4th
        bit (resp. the sixth one), from the left, set to 0 (resp. 1). Only
        these port values will be used by the peer when enforcing the
        configuration conveyed by PPP IPCP.</t>
      </section>

      <section title="Description of Cryptographically Random Port Range option">
        <t>A cryptographically random Port Range Option may be used as a
        mitigation tool against blind attacks described in <xref
        target="RFC6056"></xref>.</t>

        <section anchor="function" title="Random Port Delegation Function">
          <t>Delegating random ports can be achieved by defining a function
          which takes as input a key 'k' and an integer 'x' within the range
          (1024, 65535) and produces an output 'y' also within the port range
          (1024, 65535).</t>

          <t>The cryptographical mechanism ensures that the entire 64k port
          range can be efficiently distributed to multiple nodes in a way that
          when nodes calculate the ports, the results will never overlap with
          ports other nodes have calculated (property of permutation), and
          ports in the reserved range (smaller than 1024) are not used. As the
          randomization is done cryptographically, an attacker seeing a node
          using some port X cannot determine which other ports the node may be
          using (as the attacker does not know the key). Calculation of the
          random port list is done as follows:</t>

          <t>The cryptographic mechanism uses an encryption function y =
          E(K,x) that takes as input a key K (for example, 128 bits) and an
          integer x (the plaintext) in range (1024, 65535), and produces an
          output y (the ciphertext), also an integer in range (1024, 65535).
          This section describes one such encryption function, but others are
          also possible.</t>

          <t>The server will select the key K. When the server wants to
          allocate e.g. 2048 random ports, it selects a starting point 'a'
          (1024 <= a <= 65536-2048) in a way that the port range (a,
          a+2048) does not overlap with any other active client, and
          calculates the values E(K,a), E(K,a+1), E(K,a+2), ..., E(K,a+2046),
          E(K,a+2047). These are the port numbers allocated for this node.
          Instead of sending the port numbers individually, the server just
          sends the values 'K', ' a', and '2048'. The client will then repeat
          the same calculation.</t>

          <t>The server SHOULD use different K for each IPv4 address it
          allocates to make attacks as difficult as possible. This way,
          learning the K used in IPv4 address IP1 would not help in attacking
          IPv4 address IP2 that is allocated by the same server to different
          nodes.</t>

          <t>With typical encryption functions (such as AES and DES), the
          input (plaintext) and output (ciphertext) are blocks of some fixed
          size; for example, 128 bits for AES, and 64 bits for DES. For port
          randomization, we need an encryption function whose input and output
          is an integer in range (1024, 65535).</t>

          <t>One possible way to do this is to use the 'Generalized-Feistel
          Cipher' <xref target="CIPHERS"></xref> construction by Black and
          Rogaway, with AES as the underlying round function.</t>

          <t>This would look as follows (using pseudo-code):</t>

          <figure>
            <artwork><![CDATA[        def E(k, x):
            y = Feistel16(k, x)
            if y >= 1024:
                  return y
            else:
                  return E(k, y)]]></artwork>
          </figure>

          <t>Note that although E(k,x) is recursive, it is guaranteed to
          terminate. The average number of iterations is just slightly over
          1.</t>

          <t>Feistel16 is a 16-bit block cipher:</t>

          <t><figure>
              <artwork><![CDATA[        def Feistel16(k, x):
            left = x & 0xff
            right = x >> 8
            for round = 1 to 3:
                temp = left ^ FeistelRound(k, round, right))
                left = right
                right = temp
            return (right << 8) | left]]></artwork>
            </figure>The Feistel round function uses:</t>

          <t><figure>
              <artwork><![CDATA[        def FeistelRound(k, round, x):
            msg[0] = round
            msg[1] = x
            msg[2...15] = 0
            return AES(k, msg)[0]]]></artwork>
            </figure>Performance: To generate list of 2048 port numbers, about
          6000 calls to AES are required (i.e., encrypting 96 kilobytes).
          Thus, it will not be a problem for any device that can do, for
          example, HTTPS (web browsing over SSL/TLS).</t>
        </section>

        <section anchor="value_2"
                 title="Description of Cryptographically Random Port Range Option">
          <t></t>

          <t><figure anchor="Figure_8"
              title="Format of the cryptographically Random Port Range option">
              <preamble>The cryptographically Random Port Range IPCP Option
              adheres to the format defined in Section 2.1 of <xref
              target="RFC2153"></xref>. The "value" field of the option
              defined in <xref target="RFC2153"></xref> when conveying
              cryptographically Random Port Range IPCP Option is illustrated
              in <xref target="Figure_8"></xref></preamble>

              <artwork><![CDATA[
  0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |M|          Reserved           |          function             |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |        starting point         |   number of delegated ports   |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                             key K               ...
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  ...                                                           ...
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  ...                                                           ...
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  ...                                                             |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

]]></artwork>

              <postamble></postamble>
            </figure></t>

          <t><list style="symbols">
              <t>M: mode bit. It indicates the mode the port range is
              allocated for. A value of zero indicates the port ranges are
              delegated, while a value of 1 indicates the port ranges are port
              forwarded.</t>

              <t>Function: A 16 bit field whose value is associated with
              predefined encryption functions. This specification associates
              value 1 with the predefined function described in <xref
              target="function"></xref>.</t>

              <t>Starting Point: A 16 bit value used as an input to the
              specified function</t>

              <t>Number of delegated ports: A 16 bit value specifying the
              number of ports delegated to the client for use as source port
              values.</t>

              <t>Key K: A 128 bit key used as input to the predefined function
              for delegated port calculation.</t>
            </list></t>

          <t>When the option is included in the IPCP Configure-Request 'key
          field' and 'starting point' field SHALL be set to all zeros. The
          requester MAY indicate in the 'function' field which encryption
          function requester prefers, and in the 'number of delegated ports'
          field the number of ports the requester would like to obtain. If
          requester has no preference it SHALL set also the 'function' field
          and/or 'number of delegated ports' field to zero.</t>

          <t>The usage of the option in IPCP message negotiation
          (Request/Reject/Nak/Ack) follows the logic described for Port Mask
          and Port Range options at <xref target="value_1"></xref>.</t>
        </section>
      </section>

      <section title="Illustration Examples">
        <t></t>

        <section title="Overview">
          <t>These flows provide examples of the usage of IPCP to convey the
          Port Range Option. As illustrated in <xref
          target="Figure_4"></xref>, IPCP messages are exchanged between a
          Host and a BRAS (Broadband Access Server).</t>

          <t><list style="numbers">
              <t>The first example illustrates a successful IPCP exchange;</t>

              <t>The second example shows the IPCP exchange that occurs when
              Port Range Option is not supported by the server;</t>

              <t>The third example shows the IPCP exchange that occurs when
              Port Range Option is not supported by the client;</t>

              <t>The fourth example shows the IPCP exchange that occurs when
              Port Range Option is not supported by the client and a non null
              IP (i.e., an address different from 0.0.0.0) address is enclosed
              in the first configuration request issued by the peer.</t>
            </list></t>
        </section>

        <section title="Successful Flow: Port Range Options supported by both the Client and the Server">
          <t>The following message exchange (i.e., <xref
          target="Figure_4"></xref>) provides an example of successful IPCP
          configuration operation when the Port Range IPCP Option is used.</t>

          <t><figure anchor="Figure_4" title="Successful flow">
              <preamble></preamble>

              <artwork><![CDATA[
  +-----+                                          +-----+
  | Host|                                          | BRAS|
  +-----+                                          +-----+
     |                                                |
     |              (1) IPCP Configure-Request        |
     |                IP ADDRESS=0.0.0.0              |
     |                PORT RANGE VALUE=0              |
     |                PORT RANGE MASK=0               |
     |===============================================>|
     |                                                |
     |              (2) IPCP Configure-Nak            |
     |                IP ADDRESS=a.b.c.d              |
     |                PORT RANGE VALUE=80             |
     |                PORT RANGE MASK=496             |
     |<===============================================|
     |                                                |
     |              (3) IPCP Configure-Request        |
     |                IP ADDRESS=a.b.c.d              |
     |                PORT RANGE VALUE=80             |
     |                PORT RANGE MASK=496             |
     |===============================================>|
     |                                                |
     |              (4) IPCP Configure-Ack            |
     |                IP ADDRESS=a.b.c.d              |
     |                PORT RANGE VALUE=80             |
     |                PORT RANGE MASK=496             |
     |<===============================================|
     |                                                |

]]></artwork>

              <postamble></postamble>
            </figure></t>

          <t>The main steps of this flow are listed below:<list style="empty">
              <t>(1) The Host sends a first Configure-Request which includes
              the set of options it desires to negotiate. All these
              Configuration Options are negotiated simultaneously. In this
              example, Configure-Request carries information about IP-address,
              Port Range Value and Port Range Mask. In this example,
              IP-address Option is set to 0.0.0.0, Port Range Value is set to
              0 and Port Range Mask is set to 0.</t>

              <t>(2) BRAS sends back a Configure-Nak and sets the enclosed
              options to its preferred values. In this example: IP-Address
              Option is set to a.b.c.d, Port Range Value is set to 80 and Port
              Range Mask is set to 496.</t>

              <t>(3) The Host re-sends a Configure-Request requesting
              IP-address Option to be set to a.b.c.d, Port Range Value to be
              set to 80 and Port Range Mask to be set to 496.</t>

              <t>(4) BRAS sends a Configure-Ack message</t>
            </list>As a result of this exchange, Host is configured to use as
          local IP address a.b.c.d and the following 128 contiguous Port
          Ranges resulting of the Port Mask (Port Range Value == 0, Port Range
          Mask == 496):</t>

          <t><list style="empty">
              <t>- from 80 to 95</t>

              <t>- from 592 to 607</t>

              <t>- ...</t>

              <t>- from 65104 to 65119</t>
            </list></t>
        </section>

        <section title="Port Range Option Not Supported by the Server">
          <t>This example (<xref target="Figure_5"></xref>) depicts an
          exchange of messages when the BRAS does not support IPCP Port Range
          Option.</t>

          <t><figure anchor="Figure_5"
              title="Failed flow: Port Range Option not supported by the server">
              <preamble></preamble>

              <artwork><![CDATA[
  +-----+                                          +-----+
  | Host|                                          | BRAS|
  +-----+                                          +-----+
     |                                                |
     |              (1) IPCP Configure-Request        |
     |                IP ADDRESS=0.0.0.0              |
     |                PORT RANGE VALUE=0              |
     |                PORT RANGE MASK=0               |
     |===============================================>|
     |                                                |
     |              (2) IPCP Configure-Reject         |
     |                PORT RANGE VALUE=0              |
     |                PORT RANGE MASK=0               |
     |<===============================================|
     |                                                |
     |              (3) IPCP Configure-Request        |
     |                IP ADDRESS=0.0.0.0              |
     |===============================================>| 
     |                                                |
     |              (4) IPCP Configure-Nak            |
     |                IP ADDRESS=a.b.c.d              |
     |<===============================================|
     |                                                |
     |              (5) IPCP Configure-Request        |
     |                IP ADDRESS=a.b.c.d              |
     |===============================================>|
     |                                                |
     |              (6) IPCP Configure-Ack            |
     |                IP ADDRESS=a.b.c.d              |
     |<===============================================|
     |                                                |

]]></artwork>

              <postamble></postamble>
            </figure></t>

          <t>The main steps of this flow are listed hereafter:<list
              style="empty">
              <t>(1) The Host sends a first Configure-Request which includes
              the set of options it desires to negotiate. All these
              Configuration Options are negotiated simultaneously. In this
              example, Configure-Request carries the codes of IP-address, Port
              Range Value and Port Range Mask options. In this example,
              IP-address Option is set to 0.0.0.0, Port Range Value is set to
              0 and Port Range Mask is set to 0.</t>

              <t>(2) BRAS sends back a Configure-Reject to decline Port Range
              option.</t>

              <t>(3) The Host sends a Configure-Request which includes only
              the codes of IP-Address option. In this example, IP-Address
              Option is set to 0.0.0.0.</t>

              <t>(4) BRAS sends back a Configure-Nak and sets the enclosed
              option to its preferred value. In this example: IP-Address
              Option is set to a.b.c.d.</t>

              <t>(5) The Host re-sends a Configure-Request requesting
              IP-Address Option to be set to a.b.c.d.</t>

              <t>(6) BRAS sends a Configure-Ack message.</t>
            </list>As a result of this exchange, Host is configured to use as
          local IP address a.b.c.d. This IP address is not a shared IP
          address.</t>
        </section>

        <section title="Port Range Option not Supported by the Client">
          <t>This example (<xref target="Figure_6"></xref>) depicts exchanges
          when only shared IP addresses are assigned to end-user's devices.
          The server is configured to assign only shared IP addresses. If Port
          Range Options are not enclosed in the configuration request, the
          request is rejected and the requesting peer will be unable to access
          the service as depicted in <xref target="Figure_6"></xref>.</t>

          <t><figure anchor="Figure_6"
              title="Port Range Option not supported by the Client">
              <preamble></preamble>

              <artwork><![CDATA[
  +-----+                                          +-----+
  | Host|                                          | BRAS|
  +-----+                                          +-----+
     |                                                |
     |              (1) IPCP Configure-Request        |
     |                IP ADDRESS=0.0.0.0              |
     |===============================================>|
     |                                                |
     |              (2) IPCP Protocol-Reject          |
     |<===============================================|
     |                                                |

]]></artwork>

              <postamble></postamble>
            </figure>The main steps of this flow are:<list style="empty">
              <t>(1) The Host sends a Configure-Request requesting IP-Address
              Option to be set to 0.0.0.0 and without enclosing the Port Range
              Option.</t>

              <t>(2) BRAS sends a Protocol-Reject message.</t>
            </list>As a result of this exchange, Host is not able to access
          the service.</t>
        </section>
      </section>
    </section>

    <section anchor="IANA" title="IANA Considerations">
      <t>No action is required from IANA since this document adheres to <xref
      target="RFC2153"></xref>.</t>
    </section>

    <section anchor="Security" title="Security Considerations">
      <t>This document does not introduce any security issue in addition to
      those related to PPP. Service providers should use authentication
      mechanisms such as CHAP <xref target="RFC1994"></xref> or PPP link
      encryption <xref target="RFC1968"></xref>.</t>

      <t>The use of small and non-random port range may increase host exposure
      to attacks described in <xref target="RFC6056"></xref>. This risk can be
      reduced by using larger port ranges, by using Random Port Range Option
      or by activating means to improve the robustness of TCP against Blind
      In-Window Attacks <xref target="RFC5961"></xref>.</t>
    </section>

    <section title="Contributors">
      <t>Jean-Luc Grimault and Alain Villefranque contributed to this
      document.</t>
    </section>

    <section anchor="Acknowledgements" title="Acknowledgements">
      <t>The authors would like to thank C. Jacquenet, J. Carlson, B.
      Carpenter, M. Townsley and J. Arkko for their review.</t>
    </section>
  </middle>

  <back>
    <references title="Normative References">
      <?rfc include="reference.RFC.2119"?>

      <?rfc include='reference.RFC.1661'?>

      <?rfc include='reference.RFC.1968'?>

      <?rfc include='reference.RFC.1994'?>

      <?rfc include='reference.RFC.2153'?>

      <?rfc include='reference.RFC.5961'?>
    </references>

    <references title="Informative References">
      <?rfc include='reference.I-D.ietf-behave-lsn-requirements'?>

      <?rfc include='reference.RFC.6346'?>

      <?rfc include='reference.I-D.boucadair-port-range'?>

      <?rfc include='reference.I-D.despres-sam'?>

      <?rfc include='reference.RFC.6056'?>

      <reference anchor="CIPHERS"
                 target=" CT-RSA 2002, Lecture Notes in Computer Science vol. 2271">
        <front>
          <title>Ciphers with Arbitrary Finite Domains Topics in
          Cryptology</title>

          <author fullname="John Black" initials="J." surname="Black">
            <organization>Geoff Huston</organization>
          </author>

          <author fullname="Phillip Rogaway" initials="P." surname="Rogaway">
            <organization></organization>

            <address>
              <postal>
                <street></street>

                <city></city>

                <region></region>

                <code></code>

                <country></country>
              </postal>

              <phone></phone>

              <facsimile></facsimile>

              <email></email>

              <uri></uri>
            </address>
          </author>

          <date month="" year="2002" />
        </front>
      </reference>
    </references>
  </back>
</rfc>

PAFTECH AB 2003-20262026-04-24 05:59:52