6to4 Provider Managed Tunnels

Abstract

6to4 Provider Managed Tunnels (6to4-PMT) provide a framework that can help manage 6to4 tunnels operating in an anycast configuration. The 6to4-PMT framework is intended to serve as an option for operators to help improve the experience of 6to4 operation when conditions of the network may provide sub-optimal performance or break normal 6to4 operation. 6to4-PMT supplies a stable provider prefix and forwarding environment by utilizing existing 6to4 relays with an added function of IPv6 Prefix Translation. This operation may be particularly important in NAT444 infrastructures where a customer endpoint may be assigned a non-RFC1918 address, thus breaking the return path for anycast-based 6to4 operation. 6to4-PMT has been successfully used in a production network, implemented as open source code, and implemented by a major routing vendor.

Status of This Memo

This document is not an Internet Standards Track specification; it is published for informational purposes.

This is a contribution to the RFC Series, independently of any other RFC stream. The RFC Editor has chosen to publish this document at its discretion and makes no statement about its value for implementation or deployment. Documents approved for publication by the RFC Editor are not a candidate for any level of Internet Standard; see Section 2 of RFC 5741.

Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at http://www.rfc-editor.org/info/rfc6732.

Copyright Notice

Copyright © 2012 IETF Trust and the persons identified as the document authors. All rights reserved.

This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document.

Table of Contents

   1. Introduction ....................................................3
   2. Motivation ......................................................3
   3. 6to4 Provider Managed Tunnels ...................................5
      3.1. 6to4 Provider Managed Tunnel Model .........................5
      3.2.  Traffic Flow ..............................................5
      3.3.  Prefix Translation ........................................6
      3.4.  Translation State .........................................7
   4. Deployment Considerations and Requirements ......................7
      4.1. Customer Opt-Out ...........................................7
      4.2. Shared CGN Space Considerations ............................8
      4.3. End-to-End Transparency ....................................8
      4.4. Path MTU Discovery Considerations ..........................9
      4.5. Checksum Management ........................................9
      4.6. Application Layer Gateways .................................9
      4.7. Routing Requirements .......................................9
      4.8. Relay Deployments .........................................10
   5. Security Considerations ........................................10
   6. Acknowledgements ...............................................10
   7. References .....................................................11
      7.1. Normative References ......................................11
      7.2. Informative References ....................................11

1. Introduction

6to4 [RFC3056] tunneling, along with the anycast operation described in [RFC3068], is widely deployed in modern Operating Systems and off-the-shelf gateways sold throughout retail and Original Equipment Manufacturer (OEM) channels. Anycast-based 6to4 [RFC3068] allows for tunneled IPv6 connectivity through IPv4 clouds without explicit configuration of a relay address. Since the overall system utilizes anycast forwarding in both directions, flow paths are difficult to determine, tend to follow separate paths in either direction, and often change based on network conditions. The return path is normally uncontrolled by the local operator and can contribute to poor performance for IPv6 and can also act as a breakage point. Many of the challenges with 6to4 are described in [RFC6343]. A specific critical use case for problematic anycast 6to4 operation is related to conditions in which the consumer endpoints are downstream from a northbound Carrier-Grade NAT (CGN) [RFC6264] function when assigned non-RFC1918 IPv4 addresses, which are not routed on interdomain links.

Operators that are actively deploying IPv6 networks and operate legacy IPv4 access environments may want to utilize the existing 6to4 behavior in customer site resident hardware and software as an interim option to reach the IPv6 Internet in advance of being able to offer full native IPv6. Operators may also need to address the brokenness related to 6to4 operation originating from behind a provider NAT function. 6to4-PMT offers an operator the opportunity to utilize IPv6 Prefix Translation to enable deterministic traffic flow and an unbroken path to and from the Internet for IPv6-based traffic sourced originally from these 6to4 customer endpoints.

6to4-PMT translates the prefix portion of the IPv6 address from the 6to4-generated prefix to a provider-assigned prefix that is used to represent the source. This translation will then provide a stable forward and return path for the 6to4 traffic by allowing the existing IPv6 routing and policy environment to control the traffic. 6to4-PMT is primarily intended to be used in a stateless manner to maintain many of the elements inherent in normal 6to4 operation. Alternatively, 6to4-PMT can be used in a stateful translation mode should the operator choose this option.

2. Motivation

Many operators endeavor to deploy IPv6 as soon as possible so as to ensure uninterrupted connectivity to all Internet applications and content through the IPv4 to IPv6 transition process. The IPv6 preparations within these organizations are often faced with both financial challenges and timing issues related to deploying IPv6 to the network edge and related transition technologies. Many of the new technologies available for IPv4 to IPv6 transition will require the replacement of the organization's Customer Premises Equipment (CPE) to support technologies like IPv6 Rapid Deployment (6RD) [RFC5969], Dual-Stack Lite [RFC6333], and native dual-stack.

Operators face a number of challenges related to home equipment replacement. Operator-initiated replacement of this equipment will take time due to the nature of mass equipment refresh programs or may require the consumer to replace their own gear. Replacing consumer owned and operated equipment, compounded by the fact that there is also a general unawareness of what IPv6 is, also adds to the challenges faced by operators. It is also important to note that 6to4 is present in much of the equipment found in networks today that do not as of yet, or will not, support 6RD and/or native IPv6.

Operators may still be motivated to provide a form of IPv6 connectivity to customers and would want to mitigate potential issues related to IPv6-only deployments elsewhere on the Internet. Operators also need to mitigate issues related to the fact that 6to4 operation is often on by default, and may be subject to erroneous behavior. The undesired behavior may be related to the use of non-RFC1918 addresses on CPE equipment that operate behind large operator NATs or other conditions as described in a general advisory as laid out in [RFC6343].

6to4-PMT allows an operator to help mitigate such challenges by leveraging the existing 6to4 deployment base, while maintaining operator control of access to the IPv6 Internet. It is intended for use when better options, such as 6RD or native IPv6, are not yet viable. One of the key objectives of 6to4-PMT is to also help reverse the negative impacts of 6to4 in CGN environments. The 6to4-PMT operation can also be used immediately with the default parameters that are often enough to allow it to operate in a 6to4-PMT environment. Once native IPv6 is available to the endpoint, the 6to4-PMT operation is no longer needed and will cease to be used based on correct address selection behaviors in end hosts [RFC6724].

6to4-PMT thus helps operators remove the impact of 6to4 in CGN environments, deals with the fact that 6to4 is often on by default, and allows access to IPv6-only endpoints from IPv4-only addressed equipment. Additionally, it provides relief from many challenges related to mis-configurations in other networks that control return flows via foreign relays. Due to the simple nature of 6to4-PMT, it can also be implemented in a cost-effective and simple manner, allowing operators to concentrate their energy on deploying native IPv6.

3. 6to4 Provider Managed Tunnels

3.1. 6to4 Provider Managed Tunnel Model

The 6to4 managed tunnel model behaves like a standard 6to4 service between the customer IPv6 host or gateway, and the 6to4-PMT Relay (within the provider domain). The 6to4-PMT Relay shares properties with 6RD [RFC5969] by decapsulating and forwarding encapsulated IPv6 flows within an IPv4 packet to the IPv6 Internet. The model provides an additional function that translates the source 6to4 prefix to a provider-assigned prefix that is not found in 6RD [RFC5969] or traditional 6to4 operation.

The 6to4-PMT Relay is intended to provide a stateless (or stateful) mapping of the 6to4 prefix to a provider supplied prefix.

                             | 6to4-PMT Operation  |
          
          +-----+ 6to4 Tunnel +--------+  +------+  IPv6    +----+
          | CPE |-------------|6to4 BR |--| PT66 |--------- |Host|
          +-----+    IPv4     +--------+  +------+ Provider +----+
                    Network                         Prefix
                               Unified or Separate
                                Functions/Platforms

Figure 1: 6to4-PMT Functional Model

This mode of operation is seen as beneficial when compared to broken 6to4 paths and/or environments where 6to4 operation may be functional but highly degraded.

3.2. Traffic Flow

Traffic in the 6to4-PMT model is intended to be controlled by the operator's IPv6 peering operations. Egress traffic is managed through outgoing routing policy, and incoming traffic is influenced by the operator-assigned prefix advertisements using normal interdormain routing functions.

The routing model is as predictable as native IPv6 traffic and legacy IPv4-based traffic. Figure 2 provides a view of the routing topology needed to support this relay environment. The diagram references PrefixA as 2002::/16 and PrefixB as the example 2001:db8::/32.

        |  6to4 IPv4 Path     |       Native IPv6 Path            |
               -----------       -----------      -------------
              /  IPv4 Net \     /  IPv6 Net  \  / IPv6 Internet \
        +------+         +--------+         +-------+    +---------+
        | CPE  | PrefixA |6to4-PMT| PrefixB |Peering|    |IPv6 HOST|
        +------+         +--------+         +-------+    +---------+
              \           /     \            /  \               /
               ----------        ------------     --------------
        
                IPv4 6to4       IPv6 Provider       IPv6 Prefix
                 Anycast           Prefix           Propagation

Figure 2: 6to4-PMT Flow Model

Traffic between two 6to4-enabled devices would use the IPv4 path for communication according to [RFC3056] unless the local host still prefers traffic via a relay. 6to4-PMT is intended to be deployed in conjunction with the 6to4 relay function in an attempt to help simplify its deployment. The model can also provide the ability for an operator to forward both 6to4-PMT (translated) and normal 6to4 flows (untranslated) simultaneously based on configured policy.

3.3. Prefix Translation

IPv6 Prefix Translation is a key part of the system as a whole. The 6to4-PMT framework is a combination of two concepts: 6to4 [RFC3056] and IPv6 Prefix Translation. IPv6 Prefix Translation, as used in 6to4-PMT, has some similarities to concepts discussed in [RFC6296]. 6to4-PMT would provide prefix translation based on specific rules configured on the translator that maps the 6to4 2002::/16 prefix to an appropriate provider assigned prefix. In most cases, a ::/32 prefix would work best in 6to4-PMT that matches common Regional Internet Registry (RIR) prefix assignments to operators.

The provider can use any prefix mapping strategy they so choose, but the simpler the better. Simple direct bitmapping can be used, or more advanced forms of translation should the operator want to achieve higher address compression. More advanced forms of translation may require the use of stateful translation.

Figure 3 shows a 6to4 Prefix with a Subnet-ID of "0000" mapped to a provider-assigned, globally unique prefix (2001:db8::/32). With this simple form of translation, there is support for only one Subnet-ID per provider-assigned prefix. In characterization of deployed OSs and gateways, a Subnet-ID of "0000" is the most common default case followed by Subnet-ID "0001". Use of the Subnet-ID can be referenced in [RFC4291]. It should be noted that in normal 6to4 operation, the endpoint (network) has access to 65,536 (16-bits) Subnet IDs. In the

6to4-PMT case as described above using the mapping in Figure 3, all but the one Subnet-ID used for 6to4-PMT would still operate under normal 6to4 operation.

      Pre-Relayed Packet [Provider Access Network Side]
      
      0     16      32     48     64    80     96     112    128 Bits
      | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- |
        2002 : 0C98 : 2C01 : 0000 : xxxx : xxxx : xxxx : xxxx
      | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- |
                 |       |            |      |      |      |
                  ----    ----        |      |      |      |
                      |       |       |      |      |      |
      | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- |
        2001 : 0db8 : 0c98 : 2c01 : xxxx : xxxx : xxxx : xxxx
      | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- |

Post-Relayed Packet [Internet Side]

Figure 3: 6to4-PMT Prefix Mapping

3.4. Translation State

It is preferred that the overall system use deterministic prefix translation mappings such that stateless operation can be implemented. This allows the provider to place N number of relays within the network without the need to manage translation state. Deterministic translation also allows a customer to employ inward services using the translated (provider prefix) address.

If stateful operation is chosen, the operator would need to validate state and routing requirements particular to that type of deployment. The full body of considerations for this type of deployment is not within this scope of this document.

4. Deployment Considerations and Requirements

4.1. Customer Opt-Out

A provider enabling this function should offer a method to allow customers to opt-out of such a service should the customer choose to maintain normal 6to4 operation irrespective of degraded performance. In cases where the customer is behind a CGN device, the customer would not be advised to opt-out and can be assisted in turning off 6to4.

Since the 6to4-PMT system is targeted at customers who are relatively unaware of IPv6 and IPv4, and normally run network equipment with a default configuration, an opt-out strategy is recommended. This method provides 6to4-PMT operation for non-IPv6 savvy customers whose equipment may turn on 6to4 automatically and allows savvy customers to easily configure their way around the 6to4-PMT function.

Capable customers can also disable anycast-based 6to4 entirely and use traditional 6to4 or other tunneling mechanisms if they are so inclined. This is not considered the normal case, and most endpoints with auto-6to4 functions will be subject to 6to4-PMT operation since most users are unaware of its existence. 6to4-PMT is targeted as an option for stable IPv6 connectivity for average consumers.

4.2. Shared CGN Space Considerations

6to4-PMT operation can also be used to mitigate a known problem with 6to4 occurring when shared address space [RFC6598] or Global Unicast Addresses (GUA) are used behind a CGN and not routed on the Internet. Non-RFC1918, yet unrouted (on interdomain links) address space would cause many deployed OSs and network equipment to potentially auto-enable 6to4 operation even without a valid return path (such as behind a CGN function). The operator's desire to use non-RFC1918 addresses, such as shared address space [RFC6598], is considered highly likely based on real world deployments.

Such hosts, in normal cases, would send 6to4 traffic to the IPv6 Internet via the anycast relay, which would in fact provide broken IPv6 connectivity, since the return path flow is built using an IPv4 address that is not routed or assigned to the source network. The use of 6to4-PMT would help reverse these effects by translating the 6to4 prefix to a provider-assigned prefix, masking this automatic and undesired behavior.

4.3. End-to-End Transparency

The 6to4-PMT mode of operation removes the traditional end-to-end transparency of 6to4. Remote hosts would connect to a 6to4-PMT- serviced host using a translated IPv6 address versus the original 6to4 address based on the 2002::/16 well-known prefix. This can be seen as a disadvantage of the 6to4-PMT system. This lack of transparency should also be contrasted with the normal operating state of 6to4 that provides connectivity that is uncontrolled and often prone to high latency. The lack of transparency is, however, a better form of operation when extreme poor performance, broken IPv6 connectivity, or no IPv6 connectivity is considered as the alternative.

4.4. Path MTU Discovery Considerations

The MTU will be subject to a reduced value due to standard 6to4 tunneling operation. Under normal 6to4 operation, the 6to4 service agent would send an ICMP Packet Too Big Message as part of Path MTU discovery as described in [RFC4443] and [RFC1981], respectively. In 6to4-PMT operation, the PMT Service agent should be aware of the reduced 6to4 MTU and send ICMP messages using the translated address accordingly.

It is also possible to pre-constrain the MTU at the upstream router from the 6to4-PMT service agents that would then have the upstream router send the appropriate ICMP Packet Too Big Messages.

4.5. Checksum Management

Checksum management for 6to4-PMT can be implemented in one of two ways. The first deployment model is based on the stateless 6to4-PMT operational mode. In this case, checksum modifications are made using the method described in [RFC3022], Section 4.2. The checksum is modified to match the parameters of the translated address of the source 6to4-PMT host. In the second deployment model in which stateful 6to4-PMT translation is used, the vendor can implement checksum-neutral mappings as defined in [RFC6296].

4.6. Application Layer Gateways

Vendors can choose to deploy Application Layer Gateways (ALGs) on their platforms that perform 6to4-PMT if they so choose. No ALGs were deployed as part of the open source and vendor product deployments of 6to4-PMT. In the vendor deployment case, the same rules were used as with their NPTv6 [RFC6296] base code.

4.7. Routing Requirements

The provider would need to advertise the well-known IP address range used for normal anycast 6to4 [RFC3068] operation within the local IPv4 routing environment. This advertisement would attract the 6to4 upstream traffic to a local relay. To control this environment and make sure all northbound traffic lands on a provider-controlled relay, the operator may filter the anycast range from being advertised from customer endpoints toward the local network (upstream propagation).

The provider would not be able to control route advertisements inside the customer domain, but that use case is not in scope for this document. In that case, it is likely that the end network/customer understands 6to4 and is maintaining their own relay environment and therefore would not be subject to the operators 6to4 and/or PMT operation.

Within their own network, the provider would also likely want to advertise the 2002::/16 range to help bridge traditional 6to4 traffic within the network (native IPv6 to 6to4-PMT-based endpoint). It would also be advised that the local 6to4-PMT operator not leak the well-known 6to4 anycast IPv4 prefix to neighboring Autonomous Systems to prevent PMT operation for neighboring networks. Policy configuration on the local 6to4-PMT Relay can also be used to disallow PMT operation should the local provider service downstream customer networks.

4.8. Relay Deployments

The 6to4-PMT function can be deployed onto existing 6to4 relays (if desired) to help minimize network complexity and cost. 6to4-PMT has already been developed on Linux-based platforms that are package add-ons to the traditional 6to4 code. The only additional considerations beyond normal 6to4 relay operation would include the need to route specific IPv6 provider prefix ranges used for 6to4-PMT operation towards peers and transit providers.

5. Security Considerations

6to4-PMT operation would be subject to the same security concerns as normal 6to4 operation as described in [RFC6169]. 6to4-PMT is also not plainly perceptible by external hosts, and local entities appear as native IPv6 hosts to the external hosts.

6. Acknowledgements

Thanks to the following people for their textual contributions and/or guidance on 6to4 deployment considerations: Dan Wing, Wes George, Scott Beuker, JF Tremblay, John Brzozowski, Chris Metz, and Chris Donley.

Additional thanks to the following for assisting with the coding and testing of 6to4-PMT: Marc Blanchet, John Cianfarani, Tom Jefferd, Nik Lavorato, Robert Hutcheon, and Ida Leung.

7. References

7.1. Normative References

   [RFC3056]  Carpenter, B. and K. Moore, "Connection of IPv6 Domains
              via IPv4 Clouds", RFC 3056, February 2001.
   
   [RFC3068]  Huitema, C., "An Anycast Prefix for 6to4 Relay Routers",
              RFC 3068, June 2001.

7.2. Informative References

   [RFC1981]  McCann, J., Deering, S., and J. Mogul, "Path MTU Discovery
              for IP version 6", RFC 1981, August 1996.
   
   [RFC3022]  Srisuresh, P. and K. Egevang, "Traditional IP Network
              Address Translator (Traditional NAT)", RFC 3022, January
              2001.
   
   [RFC4291]  Hinden, R. and S. Deering, "IP Version 6 Addressing
              Architecture", RFC 4291, February 2006.
   
   [RFC4443]  Conta, A., Deering, S., and M. Gupta, Ed., "Internet
              Control Message Protocol (ICMPv6) for the Internet
              Protocol Version 6 (IPv6) Specification", RFC 4443, March
              2006.
   
   [RFC5969]  Townsley, W. and O. Troan, "IPv6 Rapid Deployment on IPv4
              Infrastructures (6rd) -- Protocol Specification", RFC
              5969, August 2010.
   
   [RFC6169]  Krishnan, S., Thaler, D., and J. Hoagland, "Security
              Concerns with IP Tunneling", RFC 6169, April 2011.
   
   [RFC6264]  Jiang, S., Guo, D., and B. Carpenter, "An Incremental
              Carrier-Grade NAT (CGN) for IPv6 Transition", RFC 6264,
              June 2011.
   
   [RFC6296]  Wasserman, M. and F. Baker, "IPv6-to-IPv6 Network Prefix
              Translation", RFC 6296, June 2011.
   
   [RFC6333]  Durand, A., Droms, R., Woodyatt, J., and Y. Lee,
              "Dual-Stack Lite Broadband Deployments Following IPv4
              Exhaustion", RFC 6333, August 2011.
   
   [RFC6343]  Carpenter, B., "Advisory Guidelines for 6to4 Deployment",
              RFC 6343, August 2011.
   
   [RFC6598]  Weil, J., Kuarsingh, V., Donley, C., Liljenstolpe, C., and
              M. Azinger, "IANA-Reserved IPv4 Prefix for Shared Address
              Space", BCP 153, RFC 6598, April 2012.
   
   [RFC6724]  Thaler, D., Ed., Draves, R., Matsumoto, A., and T. Chown,
              "Default Address Selection for Internet Protocol Version 6
              (IPv6)", RFC 6724, September 2012.

Authors' Addresses

Victor Kuarsingh (editor)
Rogers Communications
8200 Dixie Road
Brampton, Ontario L6T 0C1
Canada

   EMail: victor.kuarsingh@gmail.com
   URI:   http://www.rogers.com

Yiu L. Lee
Comcast
One Comcast Center
Philadelphia, PA 19103
U.S.A.

   EMail: yiu_lee@cable.comcast.com
   URI:   http://www.comcast.com

Olivier Vautrin
Juniper Networks
1194 N Mathilda Avenue
Sunnyvale, CA 94089
U.S.A.

   EMail: olivier@juniper.net
   URI:   http://www.juniper.net