IPv6 Tunnel Broker with the Tunnel Setup Protocol (TSP)
draft-blanchet-v6ops-tunnelbroker-tsp-04

Status of this Memo

By submitting this Internet-Draft, each author represents that any applicable patent or other IPR claims of which he or she is aware have been or will be disclosed, and any of which he or she becomes aware will be disclosed, in accordance with Section 6 of BCP 79.

Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet-Drafts.

Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as “work in progress.”

The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt.

The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html.

This Internet-Draft will expire on November 8, 2008.

Abstract

A tunnel broker with the Tunnel Setup Protocol (TSP) enables the establishment of tunnels of various inner protocols, such as IPv6 or IPv4, inside various outer protocols packets, such as IPv4, IPv6 or UDP over IPv4 for IPv4 NAT traversal. The control protocol (TSP) is used by the tunnel client to negotiate the tunnel with the broker. A mobile node implementing TSP can be connected to both IPv4 and IPv6 networks whether it is on IPv4 only, IPv4 behind a NAT or on IPv6 only. A tunnel broker may terminate the tunnels on remote tunnel servers or on itself. This document describes the TSP protocol within the model of the tunnel broker model.

Table of Contents

1.  Introduction
2.  Description of the TSP framework
    2.1.  NAT Discovery
    2.2.  Any encapsulation
    2.3.  Mobility
3.  Advantages of TSP
4.  Protocol Description
    4.1.  Terminology
    4.2.  Topology
    4.3.  Overview
    4.4.  TSP signaling
        4.4.1.  Signaling transport
        4.4.2.  Authentication phase
        4.4.3.  Command and response phase
    4.5.  Tunnel establishment
        4.5.1.  IPv6-over-IPv4 tunnels
        4.5.2.  IPv6-over-UDP tunnels
    4.6.  Tunnel Keep-alive
    4.7.  XML Messaging
        4.7.1.  Tunnel
        4.7.2.  Client Element
        4.7.3.  Server Element
        4.7.4.  Broker Element
5.  Tunnel request examples
    5.1.  Host tunnel request and reply
    5.2.  Router Tunnel request with a /48 prefix delegation, and reply
    5.3.  IPv4 over IPv6 tunnel request
    5.4.  NAT Traversal tunnel request
6.  Applicability of TSP in Different Networks
    6.1.  Provider Networks with Enterprise Customers
    6.2.  Provider Networks with Home/Small Office Customers
    6.3.  Enterprise Networks
    6.4.  Wireless Networks
    6.5.  Unmanaged networks
    6.6.  Mobile Hosts and Mobile Networks
7.  IANA Considerations
8.  Security Considerations
9.  Conclusion
10.  Acknowledgements
11.  References
    11.1.  Normative References
    11.2.  Informative References
Appendix A.  The TSP DTD
Appendix B.  Error codes
§  Authors' Addresses
§  Intellectual Property and Copyright Statements

1.  Introduction

This document first describes the TSP framework, the protocol details, and the different profiles used. It then describes the applicability of TSP in different environments, some of which were described in the v6ops scenario documents.

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 [RFC2119] (Bradner, S., “Key words for use in RFCs to Indicate Requirement Levels,” March 1997.).

2.  Description of the TSP framework

Tunnel Setup Protocol (TSP) is a signaling protocol to setup tunnel parameters between two tunnel end-points. TSP is implemented as a tiny client code in the requesting tunnel end-point. The other end-point is the server that will setup the tunnel service. TSP uses XML [W3C.REC‑xml‑20040204] (Yergeau, F., Paoli, J., Sperberg-McQueen, C., Bray, T., and E. Maler, “Extensible Markup Language (XML) 1.0 (Third Edition),” February 2004.) basic messaging over TCP or UDP. The use of XML gives extensibility and easy option processing.

TSP negotiates tunnel parameters between the two tunnel end-points. Parameters that are always negociated are:

• authentication of the users, using any kind of authentication mechanism (through SASL [RFC4422] (Melnikov, A. and K. Zeilenga, “Simple Authentication and Security Layer (SASL),” June 2006.)) including anonymous
• Tunnel encapsulation
  • IPv6 over IPv4 tunnels [RFC4213] (Nordmark, E. and R. Gilligan, “Basic Transition Mechanisms for IPv6 Hosts and Routers,” October 2005.)
  • IPv4 over IPv6 tunnels [RFC2473] (Conta, A. and S. Deering, “Generic Packet Tunneling in IPv6 Specification,” December 1998.)
  • IPv6 over UDP-IPv4 tunnels for NAT traversal
• IP address assignment for the tunnel endpoints
• DNS registration of the IP end point address (AAAA)

Other tunnel parameters that may be negotiated are:

• Tunnel keep-alive
• IPv6 prefix assignment when the client is a router
• DNS delegation of the inverse tree, based on the IPv6 prefix assigned
• Routing protocols

The tunnel encapsulation can be explicitly specified by the client, or can be determined during the TSP exchange by the broker. The latter is used to detect the presence of NAT in the path and select IPv6 over UDP-IPv4 encapsulation.

The TSP connection can be established between two nodes, where each node can control a tunnel end-point.

The nodes involved in the framework are:

1. the TSP client
2. client tunnel end-point
3. the TSP server
4. server tunnel end-point

1,3 and 4 form the tunnel broker model [RFC3053] (Durand, A., Fasano, P., Guardini, I., and D. Lento, “IPv6 Tunnel Broker,” January 2001.), where 3 is the tunnel broker and 4 is the tunnel server (Figure 1 (Tunnel Setup Protocol used on Tunnel Broker model)). The tunnel broker may control one or many tunnel servers.

In its simplest model, one node is the client configured as a tunnel end-point (1 and 2 on same node), and the second node is the server configured as the other tunnel end-point (3 and 4 on same node). This model is shown in Figure 2 (Tunnel Setup Protocol used on Tunnel Server model)

                           _______________
                          | TUNNEL BROKER |--> Databases (DNS)
                          |               |
                          |  TSP          |
                          | SERVER        |
                          |_______________|
                              |     |
         __________           |     |          ________
        |           |         |     |         |        |
        |   TSP     |--[TSP]--      +---------|        |
        |  CLIENT   |                         | TUNNEL |--[NETWORK]--
[HOST]--|           |<==[CONFIGURED TUNNEL]==>| SERVER |
        |___________|                         |        |
                                              |________|

         ___________                           ________
        |           |                         |  TSP   |
        |   TSP     |-----------[TSP]---------| SERVER |
        |  CLIENT   |                         |        |--[NETWORK]--
[HOST]--|           |<==[CONFIGURED TUNNEL]==>| TUNNEL |
        |___________|                         | SERVER |
                                              |________|

From the point of view of an operating system, TSP is implemented as a client application which is able to configure network parameters of the operating system.

2.1.  NAT Discovery

TSP is also used to discover if a NAT is in the path. In this discovery mode, the client sends a TSP message over UDP, containing its tunnel request information (such as its source IPv4 address) to the TSP server. The TSP server compares the IPv4 source address of the packet with the address in the TSP message. If they differ, one or many IPv4 NAT is in the path.

If an IPv4 NAT is discovered, then IPv6 over UDP-IPv4 tunnel encapsulation is selected. Once the TSP signaling is done, the tunnel is established over the same UDP channel used for TSP, so the same NAT address-port mapping is used for both the TSP session and the IPv6 traffic. If no IPv4 NAT is detected in the path by the TSP server, then IPv6 over IPv4 encapsulation is used.

A keep-alive mechanism is also included to keep the NAT mapping active.

The IPv4 NAT discovery builds the most effective tunnel for all cases, including in a dynamic situation where the client moves.

2.2.  Any encapsulation

TSP is used to negotiate IPv6 over IPv4 tunnels, IPv6 over UDP-IPv4 tunnels and IPv4 over IPv6 tunnels. IPv4 over IPv6 tunnels are used in the Dual Stack Transition Mechanism (DSTM) together with TSP [I‑D.bound‑dstm‑exp] (Bound, J., Toutain, L., and JL. Richier, “Dual Stack IPv6 Dominant Transition Mechanism,” October 2005.).

2.3.  Mobility

When a node moves to a different IP network (i.e. change of its IPv4 address when doing IPv6 over IPv4 encapsulation), the TSP client reconnects automatically to the broker to re-establish the tunnel (keep-alive mechanism). On the IPv6 layer, if the client uses user authentication, the same IPv6 address and prefix are kept and re-established, even if the IPv4 address or tunnel encapsulation type changes.

3.  Advantages of TSP

• Tunnels established by TSP are static tunnels, which are more secure than automated tunnels ([RFC3964] (Savola, P. and C. Patel, “Security Considerations for 6to4,” December 2004.)). No 3rd party relay required.
• Stability of the IP address and prefix, enabling applications needing stable address to be deployed and used. For example, when tunneling IPv6, there is no dependency on the underlying IPv4 address.
• Prefix assignment supported. Can use provider address space.
• Signaling protocol flexible and extensible (XML, SASL)
• One solution to many encapsulation techniques: v6 in v4, v4 in v6, v6 over UDP over v4. Can be extended to other encapsulation types, such as v6 in v6.
• Discovery of IPv4 NAT in the path, establishing the most optimized tunnelling technique depending on the discovery.

4.  Protocol Description

4.1.  Terminology

Tunnel Broker (TB):

In a tunnel broker model, the broker is taking charge of all communication between tunnel servers (TS) and tunnel clients (TC). Tunnel clients query brokers for a tunnel and the broker finds a suitable tunnel server, asks the Tunnel server to setup the tunnel and sends the tunnel information to the Tunnel Client.

Tunnel Server (TS):

Tunnel Servers are providing the specific tunnel service to a Tunnel Client. It can receive the tunnel request from a Tunnel Broker (as in the Tunnel Broker model) or directly from the Tunnel Client. The Tunnel Server is the tunnel end-point.

Tunnel Client (TC):

The tunnel client is the entity that needs a tunnel for a particular service or connectivity. A tunnel client can be either a host or a router. The tunnel client is the other tunnel end-point.

v6v4:

IPv6-over-IPv4 tunnel encapsulation

v6udpv4:

IPv6-over-UDP-over-IPv4 tunnel encapsulation

v4v6:

IPv4-over-IPv6 tunnel encapsulation

4.2.  Topology

The following diagrams describe typical TSP scenarios. The goal is to establish a tunnel between Tunnel client and Tunnel server.

4.3.  Overview

The Tunnel Setup Protocol is initiated from a client node to a tunnel broker. The Tunnel Setup Protocol has three phases:

Authentication phase:

The Authentication phase is when the tunnel broker/server advertises its capability to a tunnel client and when a tunnel client authenticate to the broker/server.

Command phase:

The command phase is where the client requests or updates a tunnel.

Response phase:

The response phase is where the tunnel client receives the request response from the tunnel broker/server, and the client accepts or rejects the tunnel offered.

For each command sent by a Tunnel Client there is an expected response by the server.

After the response phase is completed, a tunnel is established as requested by the client. If requested, periodic keep-alive packets can be sent from the client to the server.

        tunnel                              tunnel
        client                              broker
          +|         Send version              +
          ||---------------------------------> ||
          ||         Send capabilities         ||
          ||<--------------------------------- +| Authentication
          ||         SASL authentication       || phase
          ||<--------------------------------> ||
 TSP      ||         Authentication OK         ||
 signaling||<--------------------------------- +
          ||         Tunnel request            || Command
          ||---------------------------------> || phase
          ||         Tunnel response           +
          ||<--------------------------------- || Response
          ||         Tunnel acknowledge        || phase
          ||---------------------------------> +
          +|                                   |
          ||         Tunnel established        |
 Data     ||===================================|
 phase    ||                                   |
          +|           (keep-alive)            |

4.4.  TSP signaling

The following sections describes in detail the TSP protocol and the different phases in the TSP signaling.

4.4.1.  Signaling transport

TSP signaling can be transported over TCP or UDP, and over IPv4 or IPv6. The tunnel client selects the transport according to the tunnel encapsulation to be requested. Figure 4 (TSP signaling transport) shows the transport used for TSP signaling with possible tunnel encapsulation requested.

TSP signaling over UDP/v4 MUST be used if a v6 over UDP over IPv4 (v6udpv4) tunnel is to be requested (e.g., for NAT traversal).

    Tunnel
    Encapsulation   Valid       Valid
    Requested       Transport   Address family
    ------------------------------------------
    v6anyv4         TCP UDP     IPv4
    v6v4            TCP UDP     IPv4
    v6udpv4             UDP     IPv4
    v4v6            TCP UDP     IPv6

Note that the TSP framework allows for other type of encapsulation to be defined, such as IPv6 over GRE or IPv6 over IPv6.

4.4.1.1.  TSP signaling over TCP

TSP over TCP is sent over port number 3653 (IANA assigned). TSP data used during signaling is detailed in the next sections.

                   +------+-----------+----------+
                   |  IP  | TCP       | TSP data |
                   |      | port 3653 |          |
                   +------+-----------+----------+
                   where IP is IPv4 or IPv6

4.4.1.2.  TSP signaling over UDP/v4

While TCP provides the connection-oriented and reliable data delivery features required during the TSP signaling session, UDP does not offer any reliability. This reliability is added inside the TSP session as an extra header at the beginning of the UDP payload.

                +------+-----------+------------+----------+
                | IPv4 | UDP       | TSP header | TSP data |
                |      | port 3653 |            |          |
                +------+-----------+------------+----------+

The algorithm used to add reliability to TSP packets sent over UDP is described in section 22.5 in [UNP] (Stevens, R., Fenner, B., and A. Rudoff, “Unix Network Programming, 3rd edition,” 2004.).

   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
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |  0xF  |                 Sequence Number                       |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                            Timestamp                          |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                            TSP data                           |
  ...

The four bit field (0-3) is set to 0xF. This marker is used by the tunnel broker to identify a TSP signaling packets that is sent after an IPv6 over UDP is established. This is explained in section Section 4.5.2 (IPv6-over-UDP tunnels)

Sequence Number:

28 bit field. Set by the tunnel client. Value is increased by one for every new packet sent to the tunnel broker. The return packet from the broker contains the unaltered sequence number.

Timestamp:

32 bit field. Set by the tunnel client. Generated from the client local time value. The return packet from the broker contains the unaltered timestamp.

TSP data:

Same as in the TCP/v4 case. Content described in latter sections.

The TSP client builds its UDP packet as described above and sends it to the tunnel broker. When the tunnel broker responds, the same values for the sequence number and timestamp MUST be sent back to the client. The TSP client can use the timestamp to determine the retransmission timeout (current time minus the packet timestamp). The client SHOULD retransmit the packet when the retransmission timeout is reached. The retransmitted packet MUST use the same sequence number as the original packet so that the server can detect duplicate packets. The client SHOULD use exponential backoff when retransmitting packets to avoid network congestion.

4.4.2.  Authentication phase

The authentication phase has 3 steps :

• Client's protocol version identification
• Server's capability advertisement
• Client authentication

When a TCP or UDP session is established to a tunnel broker, the tunnel client sends the current protocol version it is supporting. The version number syntax is:

VERSION=2.0.0 CR LF

Version 2.0.0 is the version number of this specification. Version 1.0.0 was defined in earlier drafts.

If the server doesn't support the protocol version it sends an error message and closes the session. The server can optionally send a server list that may support the protocol version of the client.

Example of an unsupported client version (without a server list)

      -- Successful TCP Connection --
      C:VERSION=0.1 CR LF
      S:302 Unsupported client version CR LF
      -- Connection closed --

Example of a version not supported (with a server list)

      -- Successful TCP Connection --
      C:VERSION=1.1 CR LF
      S:1302 Unsupported client version CR LF
        <tunnel action="list" type="broker">
           <broker>
              <address type="ipv4">1.2.3.4</address>
           </broker>
           <broker>
              <address type="dn">ts1.isp1.com</address>
           </broker>
        </tunnel>
      -- Connection closed --

If the server supports the version sent by the client, then the server sends a list of the capabilities supported for authentication and tunnels.

CAPABILITY TUNNEL=V6V4 TUNNEL=V6UDPV4 AUTH=ANONYMOUS AUTH=PLAIN AUTH=DIGEST-MD5 CR LF

Tunnel types must be registered with IANA and their profiles are defined in Section 7 (IANA Considerations). Authentication is done using SASL (Melnikov, A. and K. Zeilenga, “Simple Authentication and Security Layer (SASL),” June 2006.) [RFC4422]. Each authentication mechanism should be a registered SASL mechanism. Description of such mechanisms is not in the scope of this document.

The tunnel client can then choose to close the session if none of the capabilities fits its needs. If the tunnel client chooses to continue, it authenticates to the server using one of the advertised mechanism using SASL. If the authentication fails, the server sends an error message and closes the session.

The example in Figure 10 (Example of failed authentication) shows a failed authentication where the tunnel client requests an anonymous authentication which is not supported by the server.

Note that linebreaks and indentation within a "C:" or "S:" are editorial and not part of the protocol.

-- Successful TCP Connection --
C:VERSION=2.0.0 CR LF
S:CAPABILITY TUNNEL=V6V4 AUTH=DIGEST-MD5 CR LF
C:AUTHENTICATE ANONYMOUS CR LF
S:300 Authentication failed CR LF

Figure 11 (Successful anonymous authentication) shows a successful anonymous authentication.

-- Successful TCP Connection --
C:VERSION=2.0.0 CR LF
S:CAPABILITY TUNNEL=V6V4 TUNNEL=V6UDPV4 AUTH=ANONYMOUS AUTH=PLAIN
  AUTH=DIGEST-MD5 CR LF
C:AUTHENTICATE ANONYMOUS CR LF
S:200 Success CR LF

Digest-MD5 authentication with SASL follows [RFC2831] (Leach, P. and C. Newman, “Using Digest Authentication as a SASL Mechanism,” May 2000.). Figure 12 (Successful Digest-MD5 authentication) shows a successgul digest-md5 SASL authentication.

-- Successful TCP Connection --
C:VERSION=2.0.0 CR LF
S:CAPABILITY TUNNEL=V6V4 TUNNEL=V6UDPV4 AUTH=ANONYMOUS AUTH=PLAIN
  AUTH=DIGEST-MD5 CR LF
C:AUTHENTICATE DIGEST-MD5 CR LF
S:cmVhbG09aGV4b3Msbm9uY2U9MTExMzkwODk2OCxxb3A9YXV0aCxhbGdvcml0aG09bWQ
  1LXNlc3MsY2hhcnNldD11dGY4
C:Y2hhcnNldD11dGY4LHVzZXJuYW1lPSJ1c2VybmFtZTEiLHJlYWxtPSJoZXhvcyIsbm9
  uY2U9IjExMTM5MDg5NjgiLG5jPTAwMDAwMDAxLGNub25jZT0iMTExMzkyMzMxMSIsZG
  lnZXN0LXVyaT0idHNwL2hleG9zIixyZXNwb25zZT1mOGU0MmIzYzUwYzU5NzcxODUzZ
  jYyNzRmY2ZmZDFjYSxxb3A9YXV0aA==
S:cnNwYXV0aD03MGQ1Y2FiYzkyMzU1NjhiZTM4MGJhMmM5MDczODFmZQ==
S:200 Success CR LF

The base64-decoded version of the SASL exchange is:

S:realm="hexos",nonce="1113908968",qop="auth",algorithm=md5-sess,
  charset=utf8
C:charset=utf8,username="username1",realm="hexos",nonce="1113908968",
  nc=00000001,cnonce="1113923311",digest-uri="tsp/hexos",
  response=f8e42b3c50c59771853f6274fcffd1ca,qop=auth
S:rspauth=70d5cabc9235568be380ba2c907381fe

Once the authentication succeeds, the server sends a success return code and the protocol enters the Command phase.

4.4.3.  Command and response phase

The Command phase is where the tunnel client send a tunnel request or a tunnel update to the server. In this phase, commands are sent as XML messages. The first line is a "Content-length" directive that indicates the size of the following XML message. When the server sends a response, the first line is the "Content-length" directive, the second is the return code and third one is the XML message if any. The "Content-length" is calculated from the first character of the return code line to the last character of the XML message, inclusively.

Spaces can be inserted freely.

      -- UDP session established --
      C:VERSION=2.0.0 CR LF
      S:CAPABILITY TUNNEL=V6V4 TUNNEL=V6UDPV4 AUTH=ANONYMOUS
        AUTH=PLAIN AUTH=DIGEST-MD5 CR LF
      C:AUTHENTICATE ANONYMOUS CR LF
      S:200 Success CR LF

      C:Content-length: 205 CR LF
      <tunnel action="create" type="v6udpv4">
       <client>
        <address type="ipv4">192.0.2.135</address>
      <keepalive interval="30"></keepalive>
      </client>
      </tunnel> CR LF

      S:Content-length: 501 CR LF
      200 Success CR LF
      <tunnel action="info" type="v6udpv4" lifetime="604800">
        <server>
          <address type="ipv4">192.0.2.115</address>
          <address type="ipv6">
          2001:db8:8000:0000:0000:0000:0000:38b2
          </address>
        </server>
        <client>
          <address type="ipv4">192.0.2.135</address>
          <address type="ipv6">
          2001:db8:8000:0000:0000:0000:0000:38b3
          </address>
          <keepalive interval="30">
            <address type="ipv6">
            2001:db8:8000:0000:0000:0000:0000:38b2
            </address>
          </keepalive>
        </client>
      </tunnel> CR LF

      C:Content-length: 35 CR LF
      <tunnel action="accept"></tunnel> CR LF

The example in Figure 13 (Example of a command/response sequence) shows a client requesting an anonymous v6udpv4 tunnel, indicating that a keep-alive packet will be sent every 30 seconds. The tunnel broker responds with the tunnel parameters and indicates its acceptance of the keepalive period (Section 4.6 (Tunnel Keep-alive)). Finally, the client sends an accept message to the server.

Once the accept message has been sent, the server and client configure their tunnel endpoint based on the negotiated tunnel parameters.

4.5.  Tunnel establishment

4.5.1.  IPv6-over-IPv4 tunnels

Once the TSP signaling is completed, a tunnel can be established on the tunnel server and client node. If a v6v4 tunnel has been negotiated, then an IPv6-over-IPv4 tunnel [RFC4213] (Nordmark, E. and R. Gilligan, “Basic Transition Mechanisms for IPv6 Hosts and Routers,” October 2005.) is established using the operating system tunneling interface. On the client node, this is accomplished by the TSP client calling the appropriate OS commands or system calls.

4.5.2.  IPv6-over-UDP tunnels

If a v6udpv4 tunnel is configured, the same source/destination address and port used during the TSP signaling are used to configure the v6udpv4 tunnel. If a NAT is in the path between the TSP client and tunnel broker, the TSP signaling session will have created a UDP state in the NAT. By reusing the same UDP socket parameters to transport IPv6, the traffic will flow across the NAT using the same state.

                +------+-----------+--------+
                | IPv4 | UDP       |  IPv6  |
                | hdr. | port 3653 |        |
                +------+-----------+--------+

At any time, a client may re-establish a TSP signaling session. The client disconnects the current tunnel and starts a new TSP signaling session as described in Section 4.4.1.2 (TSP signaling over UDP/v4). If a NAT is present and the new TSP session uses the same UDP mapping in the NAT as for the tunnel, the tunnel broker will need to disconnect the client tunnel before the client can establish a new TSP session.

4.6.  Tunnel Keep-alive

A TSP client may select to send periodic keep-alive messages to the server in order to maintain its tunnel connectivity. This allows the client to detect network changes and enable automatic tunnel re-establishment. In the case of IPv6-over-UDP tunnels, periodic keep-alive can help refresh the connection state in a NAT if such device is in the tunnel path.

For IPv6-over-IPv4 and IPv6-over-UDP tunnels, the keep-alive message is an ICMPv6 echo request [RFC4443] (Conta, A., Deering, S., and M. Gupta, “Internet Control Message Protocol (ICMPv6) for the Internet Protocol Version 6 (IPv6) Specification,” March 2006.) sent from the client to the tunnel server. The IPv6 destination address of the echo message MUST be the address from the 'keepalive' element sent in the tunnel response during the TSP signaling (Section 4.4.3 (Command and response phase)). The echo message is sent over the configured tunnel.

The tunnel server responds to the ICMPv6 echo requests and can keep track of which tunnel is active. Any client traffic can also be used to verify if the tunnel is active. This can be used by the broker to disconnect tunnels that are no longer in use.

The broker can send a different keep-alive interval from the value specified in the client request. The client MUST conform to the broker specified keep-alive interval. The client SHOULD apply a random "jitter" value to avoid synchronization of keep-alive messages from many clients to the server [FJ93] (Floyd, S. and V. Jacobson, “The Synchronization of Periodic Routing Messages,” September 1993.). This is achieved by using an interval value in the range of [0.75T - T], where T is the keep-alive interval specified by the server.

4.7.  XML Messaging

This section describes the XML messaging used in the TSP signaling during the command and response phase. The XML elements and attributes are listed in the DTD (Appendix A (The TSP DTD)).

4.7.1.  Tunnel

The client and server use the tunnel token with an action attribute. Valid actions for this profile are : 'create', 'delete', 'info', 'accept' and 'reject'.

create:

action used to request a new tunnel or update an existing tunnel. Sent by the tunnel client.

delete:

action used to remove an existing tunnel from the server. Sent by the tunnel client.

info:

action used to request current properties of an existing tunnel. This action is also used by the tunnel broker to send tunnel parameters following a client 'create' action.

accept:

action used by the client to acknowledge the server that the tunnel parameters are accepted. The client will establish a tunnel.

reject:

action used by the client to signal the server that the tunnel parameters offered are rejected and no tunnel will be established.

The tunnel 'lifetime' attribute is set by the tunnel broker and specifies the lifetime of the tunnel in minutes. The lifetime is an administratively set value. When a tunnel lifetime is expired, it is disconnected on the tunnel server.

The 'tunnel' message contains three elements:

<client>:

Client's information

<server>:

Server's information

<broker>:

List of other server's

4.7.2.  Client Element

The client element contains 3 sub-elements: 'address', 'router' and 'keepalive'. These elements are used to describe the client request and will be used by the server to create the appropriate tunnel. The client element is the only element sent by a client.

The 'address' element is used to identify the client IP endpoint of the tunnel. When tunneling over IPv4, the client MUST send only its IPv4 address to the server. When tunneling over IPv6, the client MUST only send its IPv6 address to the server.

The broker then returns the assigned IPv6 or IPv4 address endpoint and domain name inside the 'client' element when the tunnel is created or updated. If supported by the broker, the 'client' element MAY contain the registered DNS name for the address endpoint assigned to the client.

Optionally a client MAY send a 'router' element to ask for a prefix delegation.

Optionally, a client MAY send a 'keepalive' element which contains the keep-alive time interval requested by the client.

4.7.3.  Server Element

The 'server' element contains 2 elements: 'address' and 'router'. These elements are used to describe the server's tunnel endpoint. The 'address' element is used to provide both IPv4 and IPv6 addresses of the server's tunnel endpoint, while the 'router' element provides information for the routing method chosen by the client.

4.7.4.  Broker Element

The 'broker' element is used by a tunnel broker to provide a alternate list of brokers to a client in the case where the server is not able to provide the requested tunnel.

The 'broker' element contains a series of 'address' element(s).

5.  Tunnel request examples

This section presents multiple examples of requests.

5.1.  Host tunnel request and reply

A simple tunnel request consist of a 'tunnel' element which contains only an 'address' element. The tunnel action is 'create', specifying a 'v6v4' tunnel encapsulation type. The response sent by the tunnel broker is an 'info' action. Note that the registered FQDN of the assigned client IPv6 address is also returned to the tunnel client.

      -- Successful TCP Connection --
      C:VERSION=2.0.0 CR LF
      S:CAPABILITY TUNNEL=V6V4 AUTH=ANONYMOUS CR LF
      C:AUTHENTICATE ANONYMOUS CR LF
      S:200 Authentication successful CR LF
      C:Content-length: 123 CR LF
        <tunnel action="create" type="v6v4">
           <client>
               <address type="ipv4">1.1.1.1</address>
           </client>
        </tunnel> CR LF
      S: Content-length: 234 CR LF
         200 OK CR LF
         <tunnel action="info" type="v6v4" lifetime="1440">
           <server>
              <address type="ipv4">192.0.2.114</address>
              <address type="ipv6">
              2001:db8:c18:ffff:0000:0000:0000:0000
              </address>
           </server>
           <client>
              <address type="ipv4">1.1.1.1</address>
              <address type="ipv6">
              2001:db8:c18:ffff::0000:0000:0000:0001
              </address>
              <address type="dn">userid.domain</address>
           </client>
         </tunnel> CR LF
      C: Content-length: 35 CR LF
         <tunnel action="accept"></tunnel> CR LF

5.2.  Router Tunnel request with a /48 prefix delegation, and reply

A tunnel request with prefix consist of a 'tunnel' element which contains 'address' element and a 'router' element. The 'router' element also contains the 'dns_server' element which is used to request DNS delegation of the assigned IPv6 prefix. The 'dns_server' element lists the IP address of the DNS servers to be registered for the reverse-mapping zone.

Tunnel request with prefix and static routes.

C: Content-length: 234 CR LF
   <tunnel action="create" type="v6v4">
    <client>
     <address type="ipv4">192.0.2.9</address>
     <router>
      <prefix length="48"/>
      <dns_server>
       <address type="ipv4">192.0.2.5</address>
       <address type="ipv4">192.0.2.4</address>
       <address type="ipv6">2001:db8::1</address>
      </dns_server>
     </router>
    </client>
   </tunnel> CR LF
S: Content-length: 234 CR LF
   200 OK CR LF
   <tunnel action="info" type="v6v4" lifetime="1440">
    <server>
     <address type="ipv4">192.0.2.114</address>
     <address type="ipv6">
     2001:db8:c18:ffff:0000:0000:0000:0000
     </address>
    </server>
    <client>
     <address type="ipv4">192.0.2.9</address>
     <address type="ipv6">
     2001:db8:c18:ffff::0000:0000:0000:0001
     </address>
     <address type="dn">userid.domain</address>
     <router>
      <prefix length="48">2001:db8:c18:1234::</prefix>
      <dns_server>
       <address type="ipv4">192.0.2.5</address>
       <address type="ipv4">192.0.2.4</address>
       <address type="ipv6">2001:db8::1</address>
      </dns_server>
     </router>
    </client>
   </tunnel> CR LF
C: Content-length: 35 CR LF
   <tunnel action="accept"></tunnel> CR LF

5.3.  IPv4 over IPv6 tunnel request

This is similar to the previous 'create' action, but with the tunnel type is set to 'v4v6'.

          -- Successful TCP Connection --
          C:VERSION=1.0 CR LF
          S:CAPABILITY TUNNEL=V4V6 AUTH=DIGEST-MD5 AUTH=ANONYMOUS
            CR LF
          C:AUTHENTICATE ANONYMOUS CR LF
          S:OK Authentication successful CR LF
          C:Content-length: 228 CR LF
            <tunnel action="create" type="v4v6">
               <client>
                   <address type="ipv6">
                   2001:db8:0c18:ffff:0000:0000:0000:0001
                   </address>
               </client>
            </tunnel> CR LF

If the allocation request is accepted, the broker will acknowledge the allocation to the client by sending a 'tunnel' element with the attribute 'action' set to 'info', 'type' set to 'v4v6' and the 'lifetime' attribute set to the period of validity or lease time of the allocation. The 'tunnel' element contains 'server' and 'client' elements.

          S: Content-length: 370 CR LF
             200 OK CR LF
             <tunnel action="info" type="v4v6" lifetime="1440">
               <server>
                  <address type="ipv4" length="30">
                  192.0.2.2
                  </address>
                  <address type="ipv6">
                  2001:db8:c18:ffff:0000:0000:0000:0002
                  </address>
               </server>
               <client>
                  <address type="ipv4" length="30">
                  192.0.2.1
                  </address>
                  <address type="ipv6">
                  2001:db8:c18:ffff::0000:0000:0000:0001
                  </address>
               </client>
             </tunnel> CR LF

In DSTM [I‑D.bound‑dstm‑exp] (Bound, J., Toutain, L., and JL. Richier, “Dual Stack IPv6 Dominant Transition Mechanism,” October 2005.) terminology, the DSTM server is the TSP broker and the TEP is the tunnel server.

5.4.  NAT Traversal tunnel request

When a client is capable of both IPv6 over IPv4 and IPv6 over UDP over IPv4 encapsulation, it can request the broker, by using the "v6anyv4" tunnel mode, to determine if it is behind a NAT and to send the appropriate tunnel encapsulation mode as part of the response. The client can also explicitly request an IPv6 over UDP over IPv4 tunnel by specifying "v6udpv4" in its request.

In the following example, the client informs the broker that it requests to send keep-alives every 30 seconds. In its response, the broker accepted the client suggested keep-alive interval, and the IPv6 destination address for the keep-alive packets is specified.

  C:VERSION=2.0.0 CR LF
  S:CAPABILITY TUNNEL=V6V4 TUNNEL=V6UDPV4 AUTH=DIGEST-MD5 CR LF
  C:AUTHENTICATE ... CR LF
  S:200 Authentication successful CR LF
  C:Content-length: ... CR LF
    <tunnel action="create" type="v6anyv4">
       <client>
           <address type="ipv4">10.1.1.1</address>
           <keepalive interval="30"></keepalive>
       </client>
    </tunnel> CR LF
  S: Content-length: ... CR LF
     200 OK CR LF
     <tunnel action="info" type="v6udpv4" lifetime="1440">
       <server>
          <address type="ipv4">192.0.2.114</address>
          <address type="ipv6">
          2001:db8:c18:ffff:0000:0000:0000:0002
          </address>
       </server>
       <client>
          <address type="ipv4">10.1.1.1</address>
          <address type="ipv6">
          2001:db8:c18:ffff::0000:0000:0000:0003
          </address>
          <keepalive interval="30">
             <address type="ipv6">
             2001:db8:c18:ffff:0000:0000:0000:0002
             </address>
          </keepalive>
       </client>
     </tunnel> CR LF

6.  Applicability of TSP in Different Networks

This section describes the applicability of TSP in different networks.

6.1.  Provider Networks with Enterprise Customers

In a provider network where IPv4 is dominant, a tunnelled infrastructure can be used to provide IPv6 services to the enterprise customers, before a full IPv6 native infrastructure is built. In order to start deploying in a controlled manner and to give enterprise customers a prefix, the TSP framework is used. The TSP server can be in the core, in the aggregation points or in the PoPs to offer the service to the customers. IPv6 over IPv4 encapsulation can be used. If the customers are behind an IPv4 NAT, then IPv6 over UDP-IPv4 encapsulation can be used. TSP can be used in combination of other techniques.

6.2.  Provider Networks with Home/Small Office Customers

In a provider network where IPv4 is dominant, a tunnelled infrastructure can be used to provider IPv6 services to the home/small office customers, before a full IPv6 native infrastructure is built. The small networks such as Home/Small offices have a non-upgradable gateway with NAT. TSP with NAT traversal is used to offer IPv6 connectivity and a prefix to the internal network.

Automation of the prefix assignment and DNS delegation, done by TSP, is a very important feature for a provider in order to substantially decrease support costs. The provider can use the same AAA database that is used to authenticate the IPv4 broadband users. Customers can deploy home IPv6 networks without any intervention of the provider support people.

With the NAT discovery function of TSP, providers can use the same TSP infrastructure for both NAT and non-NAT parts of the network.

6.3.  Enterprise Networks

In an enterprise network where IPv4 is dominant, a tunnelled infrastructure can be used to provider IPv6 services to the IPv6 islands (hosts or networks) inside the enterprise, before a full IPv6 native infrastructure is built [RFC4057] (Bound, J., “IPv6 Enterprise Network Scenarios,” June 2005.). TSP can be used to give IPv6 connectivity, prefix and routing for the islands. This gives to the enterprise a full control deployment of IPv6 while maintaining automation and permanence of the IPv6 assignments to the islands.

6.4.  Wireless Networks

In a wireless network where IPv4 is dominant, hosts and networks move and change IPv4 address. TSP enables the automatic re-establishment of the tunnel when the IPv4 address change.

In a wireless network where IPv6 is dominant, hosts and networks move. TSP enables the automatic re-establishment of the IPv4 over IPv6 tunnel.

6.5.  Unmanaged networks

An unmanaged network is where no network manager or staff is available to configure network devices [RFC3904] (Huitema, C., Austein, R., Satapati, S., and R. van der Pol, “Evaluation of IPv6 Transition Mechanisms for Unmanaged Networks,” September 2004.). TSP is particularly useful in this context where automation of all necessary information for the IPv6 connectivity is handled by TSP: tunnel end-points parameters, prefix assignment, dns delegation, routing.

An unmanaged network may be behind a NAT, maybe not. With the NAT discovery function, TSP works automatically in both cases.

6.6.  Mobile Hosts and Mobile Networks

Mobile hosts are common and used. Laptops moving from wireless, wired in office, home, ... are examples. They often have IPv4 connectivity, but not necessarily IPv6. TSP framework enables the mobile hosts to have IPv6 connectivity wherever they are, by having the TSP client send updated information of the new environment to the TSP server, when a change occurs. Together with NAT discovery and traversal, the mobile host can be always IPv6 connected wherever it is.

Mobile here means only the change of IPv4 address. Mobile-IP mechanisms and fast hand-off take care of additional constraints in mobile environments.

Mobile networks share the applicability of the mobile hosts. Moreover, in the TSP framework, they also keep their prefix assignment and can control the routing. NAT discovery can also be used.

7.  IANA Considerations

A tunnel type registry should be setup by IANA. The following strings are defined in this document:

• "v6v4" for IPv6 in IPv4 encapsulation (using IPv4 protocol 41)
• "v6udpv4" for IPv6 in UDP in IPv4 encapsulation
• "v6anyv4" for IPv6 in IPv4 or IPv6 in UDP in IPv4 encapsulation
• "v4v6" for IPv4 in IPv6 encapsulation.

Registration of a new tunnel type can be obtained on a first come first served policy [RFC2434] (Narten, T. and H. Alvestrand, “Guidelines for Writing an IANA Considerations Section in RFCs,” October 1998.). A new registration should provide a point of contact, the tunnel type string, and a brief description on the applicability.

IANA assigned 3653 as the TSP port number.

8.  Security Considerations

Authentication of the TSP session uses the SASL [RFC4422] (Melnikov, A. and K. Zeilenga, “Simple Authentication and Security Layer (SASL),” June 2006.) framework, where the authentication mechanism is negotiated between the client and the server. The framework uses the level of authentication needed for securing the session, based on the policies.

Static tunnels are created when the TSP negotiation is terminated. Static tunnels are not open gateways and exhibit less security issues than automated tunnels. Static IPv6 in IPv4 tunnels security considerations are described in [RFC4213] (Nordmark, E. and R. Gilligan, “Basic Transition Mechanisms for IPv6 Hosts and Routers,” October 2005.).

In order to help ensure that the traffic is traceable to its correct source network, a tunnel server implementation should allow ingress filtering on the user tunnel [RFC3704] (Baker, F. and P. Savola, “Ingress Filtering for Multihomed Networks,” March 2004.).

A customer A behind a NAT can use a large number of (private) IPv4 addresses and/or source ports and request multiple v6udpv4 tunnels. That would quickly saturate the tunnel server capacity. The tunnel broker implementation should offer a way to throttle and limit the number of tunnel established to the same IPv4 address.

9.  Conclusion

The Tunnel Setup Protocol (TSP) is applicable in many environments, such as: providers, enterprises, wireless, unmanaged networks, mobile hosts and networks. TSP gives the two tunnel end-points the ability to negotiate tunnel parameters, as well as prefix assignment, dns delegation and routing in an authenticated session. It also provides IPv4 NAT discovery function by using the most effective encapsulation. It also supports the IPv4 mobility of the nodes.

10.  Acknowledgements

This draft is the merge of many previous drafts about TSP. Octavio Medina has contributed to an earlier draft (IPv4 in IPv6). Thanks to the following people for comments on improving and clarifying this document: Pekka Savola, Alan Ford, Jeroen Massar and Jean-Francois Tremblay.

11.  References

11.1. Normative References

11.2. Informative References

Appendix A.  The TSP DTD

<?xml version="1.0"?>
<!DOCTYPE tunnel  [
<!ELEMENT tunnel (server?,client?,broker?)>
  <!ATTLIST tunnel action
               (create|delete|info|accept|reject) #REQUIRED >
  <!ATTLIST tunnel type
               (v6v4|v4v6|v6anyv4|v6udpv4) #REQUIRED >
  <!ATTLIST tunnel lifetime CDATA "1440"    >

<!ELEMENT server        (address+,router?)>

<!ELEMENT client        (address+,router?)>

<!ELEMENT broker        (address+)>

<!ELEMENT router        (prefix?,dns_server?)>

<!ELEMENT dns_server    (address+)>

<!ELEMENT prefix        (#PCDATA)>
  <!ATTLIST prefix length CDATA #REQUIRED>

<!ELEMENT address       (#PCDATA)>
  <!ATTLIST address type (ipv4|ipv6|dn) #REQUIRED>
  <!ATTLIST address length CDATA "">

<!ELEMENT keepalive (address?)>
  <!ATTLIST keepalive interval CDATA #REQUIRED>
]>

Appendix B.  Error codes

Error codes are sent as a numeric value followed by a text message describing the code, similar to SMTP. The codes are sent from the broker to the client. The currently defined error codes are showned below. Upon receiving an error, the client will display the appropriate message to the user.

New error messages may be defined in the future. For interoperability purpose, the error code range to use should be from 300 to 599.

The reply code 200 is used to inform the client that an action successfully completed. For example, this reply code is used in response to an authentication request and a tunnel creation request.

The server may redirect the client to another broker. The details on how these brokers are knowned or discovered is beyond the scope of this document. When a list of tunnel brokers follows the error code as a referal service, then 1000 is added to the error code.

The predefined values are :

200 Success:

Successful operation

300 Authentication failed:

Invalid userid, password or authentication mechanism.

301 No more tunnels available:

The server has reached its capacity limit.

302 Unsupported client version:

The client version is not supported by the server.

303 Unsupported tunnel type:

The server does not provide the requested tunnel type.

310 Server side error:

Undefined server error.

500 Invalid request format or specified length:

Received request has invalid syntax or truncated

501 Invalid IPv4 address:

IPv4 address specified by the client is invalid

502 Invalid IPv6 address:

IPv6 address specified by the client is invalid

506 IPv4 address already used for existing tunnel

A IPv6-over-IPv4 tunnel already exists using the same IPv4 address endpoints.

507 Requested prefix length cannot be assigned

The requested prefix length cannot be allocated on the server

521 Request already in progress

The client tunnel request is being processed by the server. Temporary error.

530 Server too busy

Request cannot be process, insufficient resources. Temporary error.

Authors' Addresses

Full Copyright Statement

Copyright © The IETF Trust (2008).

This document is subject to the rights, licenses and restrictions contained in BCP 78, and except as set forth therein, the authors retain all their rights.

This document and the information contained herein are provided on an “AS IS” basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE IETF TRUST AND THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.

Intellectual Property

The IETF takes no position regarding the validity or scope of any Intellectual Property Rights or other rights that might be claimed to pertain to the implementation or use of the technology described in this document or the extent to which any license under such rights might or might not be available; nor does it represent that it has made any independent effort to identify any such rights. Information on the procedures with respect to rights in RFC documents can be found in BCP 78 and BCP 79.

Copies of IPR disclosures made to the IETF Secretariat and any assurances of licenses to be made available, or the result of an attempt made to obtain a general license or permission for the use of such proprietary rights by implementers or users of this specification can be obtained from the IETF on-line IPR repository at http://www.ietf.org/ipr.

The IETF invites any interested party to bring to its attention any copyrights, patents or patent applications, or other proprietary rights that may cover technology that may be required to implement this standard. Please address the information to the IETF at ietf-ipr@ietf.org.