Network Working GroupT. Chown
Internet-DraftM. Thompson
Expires: April 19, 2005A. Ford
 University of Southampton, UK
 October 19, 2004

Things to think about when Renumbering an IPv6 network


Status of this Memo

This document is an Internet-Draft and is subject to all provisions of section 3 of RFC 3667. 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 become aware will be disclosed, in accordance with RFC 3668.

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

The list of Internet-Draft Shadow Directories can be accessed at

This Internet-Draft will expire on April 19, 2005.

Copyright Notice

Copyright (C) The Internet Society (2004).


This memo presents a summary of scenarios, issues for consideration and IPv6-specific tools for IPv6 network renumbering, i.e. achieving the transition from the use of an existing network prefix to a new prefix in an IPv6 network. Its focus lies not in the procedure for renumbering, but as a set of "things to think about" when undertaking such a renumbering exercise. The document is not intended to be complete at the -00 phase, and will be enhanced as further operational experience is gathered.

Table of Contents

1.  Introduction
    1.1  Past IPv4 Renumbering studies in the PIER WG
2.  Terminology
3.  Renumbering Event Triggers
    3.1  Change of uplink prefix
        3.1.1  Migration to new provider
        3.1.2  Dial on Demand
        3.1.3  Provider migration and upstream renumbering
        3.1.4  IPv6 transition
    3.2  Change of internal topology
    3.3  Acquisition or merger
    3.4  Network mobility
4.  Renumbering Requirements
    4.1  Minimal disruption
    4.2  Session survivability
        4.2.1  Short-term session survivability
        4.2.2  Medium-term session survivability
        4.2.3  Long-term session survivability
5.  IPv6 Enablers for Renumbering
    5.1  Multi-addressing
        5.1.1  Border filtering
        5.1.2  Duration of overlap
    5.2  Router Advertisement Lifetimes
    5.3  Stateful Configuration with DHCPv6
    5.4  Router Renumbering
    5.5  Prefix Delegation
    5.6  Anycast addressing
    5.7  Multicast
    5.8  Fixed length subnets
    5.9  Multi-homing techniques
        5.9.1  Relevance of multi-homing to renumbering
        5.9.2  Current situation with IPv6 multi-homing
    5.10  Unique Local Addressing
        5.10.1  Private addressing
    5.11  Mobile IPv6
        5.11.1  Visited site renumbers when mobile
        5.11.2  Home site renumbers when mobile
        5.11.3  Home site renumbers when disconnected
6.  Factors affecting the Renumbering Solution
    6.1  With or without a flag day
    6.2  Frequency of renumbering episodes
    6.3  Availability of old prefix
    6.4  Freshness of service data
    6.5  DNS and explicitly specified IP addresses
    6.6  Dual-stack network?
    6.7  Merging networks
    6.8  Embedded prefix data
    6.9  Scalability issues
    6.10  Equipment administrative ownership
    6.11  Support for Mobility?
    6.12  Stateless and Stateful address considerations
    6.13  IPv6 NAT Avoidance
    6.14  Policy and Configuration adaption
        6.14.1  Packet filters, Firewalls and ACLs
        6.14.2  Monitoring tools
7.  Application and service-oriented Issues
    7.1  Shims and sockets
    7.2  Explicitly named IP addresses
    7.3  API dilemna
    7.4  Server Sockets
8.  IETF Call to Arms
9.  IANA Considerations
10.  Security Considerations
11.  Acknowledgements
12.  References
12.1  Normative References
12.2  Informative References
§  Authors' Addresses
§  Intellectual Property and Copyright Statements


1. Introduction

This memo presents a summary of scenarios, issues for consideration and IPv6-specific tools for IPv6 network renumbering, i.e. achieving the transition from the use of an existing network prefix to a new prefix in an IPv6 network. This document does not relate the procedures for IPv6 renumbering; for such a procedure the reader is referred to [1]Baker, F., Lear, E. and R. Droms, Procedures for Renumbering an IPv6 Network without a Flag Day, July 2004.. The authors plan to use this document, together with ongoing operational experience, to refine [1]Baker, F., Lear, E. and R. Droms, Procedures for Renumbering an IPv6 Network without a Flag Day, July 2004. where necessary, to promote that guide from Informational to BCP.

1.1 Past IPv4 Renumbering studies in the PIER WG

A number of years ago (1996-1997), the Procedures for Internet/Enterprise Renumbering (PIER) WG spent time considering the issues for IPv4 renumbering. The WG produced three RFC documents. In RFC1916 [2]Berkowitz, H., Ferguson, P., Leland, W. and P. Nesser, Enterprise Renumbering: Experience and Information Solicitation, February 1996., a "call to arms" for input on renumbering techniques was made. RFC2071 [3]Ferguson, P. and H. Berkowitz, Network Renumbering Overview: Why would I want it and what is it anyway?, January 1997. documents why IPv4 renumbering is required. Interestingly, many, but not all, of the drivers have changed with respect to IPv6. In RFC2072 [4]Berkowitz, H., Router Renumbering Guide, January 1997., a Router Renumbering Guide, some operational procedures are given, much as they are in BakerBaker, F., Lear, E. and R. Droms, Procedures for Renumbering an IPv6 Network without a Flag Day, July 2004.[1] for IPv6.

Reflection on RFC2071 is interesting, witness the quote: "It is also envisioned that Network Address Translation (NAT) devices will be developed to assist in the IPv4 to IPv6 transition, or perhaps supplant the need to renumber the majority of interior networks altogether, but that is beyond the scope of this document." That need however is still very strong, particularly given the lack of Provider Independent (PI) address space in IPv6 (in IPv4, PI address space exists mainly for historical, pre-CIDR reasons).

RFC2072 is more interesting in the context of this document. Some is certainly relevant, though much is not, due to the inherent changes in IPv6. For example, there is no CIDR and address aggregation is given as mandate. Also, IPv6 subnets are in effect fixed length (/64), so local administrators do not need to resize subnets to maximize use of address space as they do in IPv4.

One core message from RFC2072 that holds true today is that of section 4 where the observation is made that renumbering networks whilst remaining the same hierarchy of subnets (i.e. the cardinality of the set of prefixes to renumber remains constant) is the 'easiest' scenario to renumber; when each "old" prefix can be mapped to a single "new" prefix.

A distinction of this work is that, where the PIER working group consider the transition from IPv4 to IPv6 addressing as a renumbering scenario, we strictly consider only the renumbering from IPv6 prefixes to other IPv6 prefixes and leave transition to well documented techniques such as those from the IPv6 Operations (v6ops) working group.

We will analyse RFC2072 in more detail in updates of this document.


2. Terminology

The following terminology is used in this document (to be expanded in future revisions):


3. Renumbering Event Triggers

This section details typical actions that result in the need for a renumbering event, and thus define the scenarios for renumbering.

In many instances, in particular those where no "flag day" is involved, the process of renumbering will inevitably lead to a scenario where hosts are multi-addressed or multi-homed as part of the renumbering procedure. The relationship between renumbering and multi-homing is discussed later in this document.

In other instances, e.g. a change in the IPv4 address offered by a provider to a site using 6to4 [9]Carpenter, B. and K. Moore, Connection of IPv6 Domains via IPv4 Clouds, February 2001., the change offers no overlap in external connectivity or addressing, and thus there is no multi-homing overlap.

Triggers may be provider-initiated or customer-initiated.

Triggers and scenarios for IPv4 renumbering are discussed in RFC2071, but many of these are no longer relevant, and in IPv6 some new triggers exist, e.g. those related to network mobility or IPv6 transition tools.

3.1 Change of uplink prefix

One of the most common causes for renumbering will be a change in the site's upstream provider. As per RFC3177 [10]IAB and IESG, IAB/IESG Recommendations on IPv6 Address Allocations to Sites, September 2001., the typical allocation for an IPv6 site is a /48 size prefix taken from the globally aggregated address space of the site's provider. With IPv6, sites are highly unlikely to be able to obtain provider independent (PI) address space, as have in some cases been obtained in the past with IPv4. Rather, sites use provider assigned (PA) addressing. As a result, if a site changes provider, it must also change its IPv6 PA prefix.

3.1.1 Migration to new provider

In the simplest case, the customer is triggering the renumbering by choosing to change the site's upstream provider to a new ISP and thus a new PA IPv6 prefix range. This may simply be in the form of selecting a new commercial provider, although there are several other possible scenarios, such as changing from a dial-up to a broadband connection, or moving from a community wireless connection to a fixed broadband connection.

3.1.2 Dial on Demand

A site may connect intermittently to its upstream provider. In such cases the prefix allocated by the provider may change with each connection, as it often does in the case of single IPv4 address allocations to SOHO customers today. Thus the site may receive a prefix still in its provider PA range, but the prefix may vary with each connection, causing a renumbering event.

Dynamically assigned IP addresses are common today with dial-up and ISDN Internet connections, and to a lesser extent some broadband products, particularly cable modems. Usually with dynamically assigned IP addresses on broadband products, the address is only likely to change when the customer reconnects, which could be very infrequently.

This case can be mitigated by encouraging ISPs to offer static IPv6 prefixes to customers. Where /48 prefixes are provided, a large ISP may be forced to require significantly more than the "default" /32 allocation from an RIR to an ISP to be able to service its present and future customer base. With always-on more common in new deployments, provider re-allocation should be less common; however the practice of reallocating IPv4 addresses in SOHO broadband networks is not uncommon in current broadband ISPs.

3.1.3 Provider migration and upstream renumbering

A site's upstream provider may need to renumber, due for example to a change in its network topology or the need to migrate to a different or additional prefix from its Regional Internet Registry (RIR). This will in turn trigger the renumbering of the end site.

Such renumbering events would be expected to be rare, but it should be noted that RIR-assigned IPv6 address space is not owned by an ISP.

3.1.4 IPv6 transition

During transition to IPv6, there are several scenarios where a site may have to renumber. For example, if the site uses 6to4 for access and its IPv4 address is dynamically assigned, an IPv6 renumbering event will be triggered when the site's IPv4 address changes.

Another likely renumbering event would be the change of transition mechanism, such as from 6to4 to a static IPv6-in-IPv4 tunnel, or from any one of those mechanisms to a native IPv6 link. When changing from 6to4 (2002::/16) addresses to native global aggregatable unicast addresses (under 2001::/16), renumbering would be unavoidable. When migrating from a tunnelled to a native connection, renumbering may not be necessary if the same prefix can be routed natively, however this would be provider-dependent.

In addition, there are likely to be many cases of network renumbering occurring when the old 6bone prefix (3FFE::/16) is phased out as per RFC3701 [11]Fink, R. and R. Hinden, 6bone (IPv6 Testing Address Allocation) Phaseout, March 2004., and networks still using it will have to renumber.

Finally, there is at least one transition mechanism, ISATAP [12]Templin, F., Gleeson, T., Talwar, M. and D. Thaler, Intra-Site Automatic Tunnel Addressing Protocol (ISATAP), May 2004., that uses specially crafted host EUI-64 format addresses. Should a site migrate from ISATAP to use either conventional EUI-64 addressing (via stateless address autoconfiguration or perhaps DHCPv6), then renumbering would be required at least in the host part of addresses.

3.2 Change of internal topology

A site may need to renumber all or part of its internal network due to a change of topology, such as creating more or less specific subnets, or acquiring a larger IPv6 address allocation. Motivations for splitting a link into separate subnets may to meet security demands on a particular link (policy for link-based access control rules), or for link load management by shuffling popular services to more appropriate locations in the local topology. Link-merging may be due to department restructuring within the hosting orgnisation, for example.

3.3 Acquisition or merger

Two networks may need to merge to one due to the acquisition or merger of two organisations or companies. Such a reorganisation may require one or more parts of the new network to renumber to the primary PA IPv6 prefix.

3.4 Network mobility

This covers various cases of network mobility, where a static or nomadic network may obtain different uplink connectivity over time, and thus be assigned different IPv6 PA prefixes as the topology changes. One example is the "traditional" NEMO network [13]Ernst, T. and H. Lach, Network Mobility Support Terminology, February 2004., another may be a community wireless network where different sets of nodes gain uplink connectivity - typically to the same provider - at different times.


4. Renumbering Requirements

In this section we enumerate potential specific goals or requirements for sites or users undergoing an IPv6 renumber event.

4.1 Minimal disruption

The renumbering event should cause minimal disruption to the routine operation of the network being renumbered, and the users of that network.

Disruption is a hard term to quantify in a generic way, but it can be expressed by factors such as:

4.2 Session survivability

The concept of session survivability is catered for by [1]Baker, F., Lear, E. and R. Droms, Procedures for Renumbering an IPv6 Network without a Flag Day, July 2004. in that new sessions adopt either old or new prefix based on the state of the renumbering process. However, other approaches to renumbering networks may be appropriate in certain deployments, such as where "flag days" are unavoidable or where two live prefixes are being "swapped". In these cases, further consideration for existing sessions (their longevity, frequency, independence across interactions, etc.) is required.

Some protocols are specifically geared to aid session survivability, e.g. the Stream Control Transmission Protocol (SCTP) [14]Stewart, R., Xie, Q., Morneault, K., Sharp, C., Schwarzbauer, H., Taylor, T., Rytina, I., Kalla, M., Zhang, L. and V. Paxson, Stream Control Transmission Protocol, October 2000., and may prove valuable in mission-critical renumbering scenarios, in particular the extension that enables the dynamic addition and removal of IP addresses from an SCTP endpoint association [15]Stewart, R., Stream Control Transmission Protocol (SCTP) Dynamic Address Reconfiguration, June 2004..

Sessions may be administratively maintained, such as NFS mounts for user filestore, or they may be user-driven, e.g. long-running ssh sessions.

In general, it is important to consider how TCP and the applications above it handle the connection failures that may result from a change in address.

There are different classes of session duration, as described in the following sections.

4.2.1 Short-term session survivability

A typical short-term session would involve a request-response protocol, such as HTTP, where a new network connection is initiated per transaction, or at worst for a small transaction set. In such cases the migration to a new network prefix is transparent: the client can use the new prefix in new transactions without consequence. Some applications, however, may be skewed by such a shift in connection source for the same entity 'user', for example applications that use recent connection history as a cue to identity (e.g. POP-before-SMTP as used by many dial-on-demand ISP customers, or for applications that care about connection statistics (the same user web-browsing "session" may be split into two where a renumbering event occurs in-between client transactions).

4.2.2 Medium-term session survivability

A medium-term session is typified by an application or service that may persist for perhaps a period of a few minutes up to a period of a day or so. This might involve a TCP-based application that is left running during a working day, such as an interactive shell (SSH) or a large file download.

4.2.3 Long-term session survivability

Long term sessions may typically run for several days, if not weeks or months. These might typically include TCP-based NFS mounts, or long-running TCP applications. Sessions in this context may also include those applications that, once started, do not re-resolve names and so repeatedly open new connections or send new datagrams to the same (as bound at time of initialisation) address throughout their execution lifetime. Even if at API-level applications do attempt to re-resolve the symbol to which they desire to connect, the behaviour of the resolvers is unclear as to whether mappings are refreshed from the naming service, and as such even if the renumbering site does update it's DNS (or NIS, LDAP database etc.), the local result may indeed be cached without any indication passed back up to the application as to how 'old' said binding information is.


5. IPv6 Enablers for Renumbering

This section documents features of IPv6, or IP-related technologies not previously (or widely) available for IPv4, that assist in renumbering.

5.1 Multi-addressing

As per RFC2373 [16]Hinden, R. and S. Deering, IP Version 6 Addressing Architecture, July 1998., IPv6 hosts may be multi-addressed. This means that multiple unicast addresses can be assigned and active on the same interface. These addresses can have different reachabilities ('scopes' such as link-local or global), different statuses including 'preferred' and 'deprecated', and may be ephemeral in nature (such as care-of addresses when attached to a foreign network [17]Johnson, D., Perkins, C. and J. Arkko, Mobility Support in IPv6, June 2004.. RFC3484 address selection semantics [5]Draves, R., Default Address Selection for Internet Protocol version 6 (IPv6), February 2003. determine which of the MxN address pairs to use for communication in the general case.

During a renumbering episode, the addition of an extra address for an endpoint does not add significantly to the complexity, and the (source and destination) address selection mechanisms specified by RFC3484 hold (but are currently at varying stages of implementation in operating systems). RFC3484 also specifies policy hooks to allow administrative override of the default address selection behaviour, for example to specifically lock-down a source prefix to select for a set of particular destinations. Where this policy-based address selection was designed for transition from IPv4 and early IPv6 multi-homing considerations, it is perceived likely to be of benefit to bespoke renumbering tool development.

Experiments to validate the correctness of common IPv6 implementations with regards RFC3484 and other aspects in the context of renumbering are on-going and will be reported in future versions of this memo.

5.1.1 Border filtering

One caveat that arises when assigning multiple globally reachable addresses to node interfaces is that of site ingress filtering: not only is it the norm for sites to filter at their border router traffic that is not destined to local subnets, but it is also increasingly common for sites to filter on egress to prevent administratively local addresses (such as the, now deprecated, site-local prefix) 'leaking' traffic or for malconfigured hosts (e.g. visitors with manually configured stacks without Mobile IPv6) from sourcing traffic that cannot be routed back.

5.1.2 Duration of overlap

A key operational decision when renumbering is enforced due to a change in connectivity provider is how long to sustain the overlap of two live prefixes. The trade-off to be made is the cost of maintaining two contracts with separate providers against the 'smoothness' of the transition to the new prefix as regards local administration overheads, service migration, etc. Where larger corporations can likely suffer the increased financial costs, SMEs and SOHOs might consider as little as one month's overlap too expensive, and so Baker's State 5 (Stable use of either prefix) [1]Baker, F., Lear, E. and R. Droms, Procedures for Renumbering an IPv6 Network without a Flag Day, July 2004. unattainable

In some cases, there may be technical reasons for the overlap to not be feasible, such as with xDSL provision where the new service is a drop-in replacement for the old and the two cannot co-exist (for example, because the provision of the service requires the whole circuit resource from exchange to customer).

5.2 Router Advertisement Lifetimes

RFC2462 (IPv6 Stateless Autoconfiguration) [6]Thomson, S. and T. Narten, IPv6 Stateless Address Autoconfiguration, December 1998. specifies the technique for expiring assigned prefixes and then invalidating them, such that a network has opportunity to gracefully withdraw a prefix from service whilst not terminally disrupting on-going applications that use addresses under it. Section 5.5.4 of RFC2462 in particular details the procedure for deprecation and subsequent invalidation.

By mandating as a node requirement the ability to phase out addresses assigned to an interface, network renumbering is readily facilitated: subnet routers update the pre-existing prefix and mark them as 'deprecated' with a scheduled time for expiration and then advertise (when appropriate) the new prefix that should be chosen for all outgoing communications.

5.3 Stateful Configuration with DHCPv6

As opposed to stateless autoconfiguration, IPv6 stateful or managed configuration can be achieved through the deployment of DHCPv6. Factors concerning the impact of stateful and stateless configuration are considered in Section 6.12Stateless and Stateful address considerations.

Section 18.1.8 of [18]Droms, R., Bound, J., Volz, B., Lemon, T., Perkins, C. and M. Carney, Dynamic Host Configuration Protocol for IPv6 (DHCPv6), July 2003. details how a node should respond to the receipt of stateful configuration data from a DHCPv6 server where the lifetime indicated has expired (is zero). Section 19.4.1 details how clients should respond to being instructed by DHCPv6 servers to reconfigure (potentially forceful renumbering). Section 22.6 details how prefix validity time is conveyed (c.f. the equivalent data in SLAAC's Router Advertisement).

5.4 Router Renumbering

RFC2894 [7]Crawford, M., Router Renumbering for IPv6, August 2000. defines a mechanism for renumbering IPv6 routers throughout a network using a bespoke ICMP message type for manipulating the set of prefixes deployed throughout subnets. Through the use of prefix matching and a rudimentary algebra for bit-wise manipulation of prefix data bound to router interfaces, the mechanism enables administrators to affect every router within a scope from a single administration workstation.

The approach utilises multicast communication to the all-routers address, FF05::2, scoped to the entire 'site' as determined by router filter policy to distribute configuration updates to all (compliant) routers. The mechanism also works with more specific addressing modalities, such as link-local multicast (FF02::2) to reach all routers on a specific link, or directed unicast to affect a specific router instance.

Example use cases cited are for deploying global routing prefixes across a hierarchical network where site-locals already exist (presumably updated now to Unique Local Addresses), and for renumbering from an existing prefix to another in a similar manner to that proposed by Baker (i.e. the deployment of a new prefix alongside the existing one, which is deprecated and subsequently expired and removed - using the same mechanism described.

Note that RFC2894 was developed before the shift in recommendation away from the [TNS]LA address allocations of RFC2373, although the techniques documented for renumbering the global routing prefix and subnet ID components in the updated address allocation recommendations (RFC3513) are not affected by the architectural change.

As with other prefix assignment techniques, it is the responsibility of the node to correctly deprecate and then expire the use of a previously assigned prefix as defined by the IPv6 Neighbour Discovery protocol, RFC2461 [8]Narten, T., Nordmark, E. and W. Simpson, Neighbor Discovery for IP Version 6 (IPv6), December 1998., section 4.6.2 describing the Prefix Information option in particular.

5.5 Prefix Delegation

Where stateless autoconfiguration enables hosts to request prefixes from link-attached routers, prefix delegation enables routers to request a prefix for advertising from superior routers, i.e. routers closer to the top of the prefix hierarchy - typically topologically closer, therefore, to the provider. Once the router has been delegated prefix(es), it can begin advertising it to the connected subnet (perhaps even multi-link) with indicators for hosts to use stateful (DHCPv6) or stateless address autoconfiguration as per RFC2461.

There have been two principal approaches to prefix delegation proposed: HPD (Hierarchical Prefix Delegation for IPv6), which proposed the use of bespoke ICMPv6 messages for prefix delegation, and IPv6 Prefix Options for Dynamic Host Configuration Protocol [19]Troan, O. and R. Droms, IPv6 Prefix Options for Dynamic Host Configuration Protocol (DHCP) version 6, December 2003., which defines a DHCPv6 option type. Of the two approaches, the DHCPv6-based approach has received wide support and is on the standards track.

5.6 Anycast addressing

Syntactically indistinguishable from unicast addresses, 'anycast' offers nodes a mean to route traffic toward the topologically nearest instance of a service (as represented by an IP address), relying on the routing infrastructure to deliver appropriately. RFC2526 [20]Johnson, D. and S. Deering, Reserved IPv6 Subnet Anycast Addresses, March 1999. defines a set of reserved subnet anycast addresses within the highest 128 values of the 64 bit IID space. Of that space, currently only three are used, of which one is actively used and is for discovery of Mobile IPv6 Home-Agents. At the current time there are no 'global' well-known unicast addresses assigned by IANA.

In order to participate using anycast, nodes need to be configured as routers (to comply with RFC2373Hinden, R. and S. Deering, IP Version 6 Addressing Architecture, July 1998.[16]) and exchange routing information about the reachability of the specific anycast address. This extra level of administration requirement is negligible in the context of services as the services themselves would need configuration anyway.

There have been proposals to define globally well-known anycast addresses for core services, such as the DNS [21]Jeong, J., IPv6 Host Configuration of DNS Server Information Approaches, September 2004.. Anycast scales with regard subnet-anycast in the sense that the global routing prefix used to direct packets to an anycast node within a site is no different from any other host, and therefore nothing 'special' in the global routing architecture is required - only locally within the site does the multi-node nature of anycast need to be considered. However, for global well-known anycast addresses to be defined, host-specific routes will need to be advertised and distributed throughout the entire Internet. As acknowledged by section 2.6 of [22]Hinden, R. and S. Deering, Internet Protocol Version 6 (IPv6) Addressing Architecture, April 2003., this presents a severe scaling limit and it is expected that support for global anycast sets may be unavailable or very restricted.

5.7 Multicast

IPv6 supports an enriched model of multicast compared to IPv4 in that there are well-defined scopes for multicast communication that are readily expressed in the protocol's addressing architecture. Multicast features much more prominently in the core specification, for example it is the enabling technology for the Neighbour Discovery protocol (a much more efficient approach to layer 2 address discovery than compared to ARP with IPv4).

Where multicast is used to discover the availability of core services (e.g. all DHCPv6 servers in a site will join FF02::1:3), the effect of renumbering the unicast address of those services will mean that the services are still readily discoverable without resorting to a (bespoke or otherwise) service location protocol to continue to function - particularly if ULAs are not deployed locally as per Section 5.10Unique Local Addressing.

5.8 Fixed length subnets

The IAB/IESG recommendations for IPv6 address allocations [10]IAB and IESG, IAB/IESG Recommendations on IPv6 Address Allocations to Sites, September 2001. details some of the motivations behind the change in the addressing architecture of IPv6 since its inception, and asserts the current state of a 64-bit 'network' part (the prefix) and a 64-bit 'host' part (the interface identifier). Fixing the lower 64 bits to be exclusive of routing topology significantly reduces the administrative load associated with renumbering and re-subnetting as experienced with IPv4 networks previously, for example, to get better address utilisation efficiency as networks evolve and provider address allocations changed.

The recommendations also discuss what length of network prefix should be allocated to sites, typically provisioning for 16-bits of subnet space in which sites can build their topology. Having such a large address space for sites to divide up at their discretion alleviates many of the drivers for renumbering discussed during the PIER working group's lifetime [3]Ferguson, P. and H. Berkowitz, Network Renumbering Overview: Why would I want it and what is it anyway?, January 1997..

5.9 Multi-homing techniques

A multi-homed site is a site which has multiple upstream providers. A site may be multi-homed for various reasons, however the most common are to provide redundancy in case of failure, to increase bandwidth, and to provide more varied, optimal routes for certain destinations.

In renumbering, multi-homing will either be a temporary state, during the transition, or be a permanent feature of the network configuration, which may be being altered during the renumbering.

5.9.1 Relevance of multi-homing to renumbering

As discussed in section 2, and in particular section 2.5, of [1]Baker, F., Lear, E. and R. Droms, Procedures for Renumbering an IPv6 Network without a Flag Day, July 2004., during the renumbering procedure there will be a period where both the old and the new prefixes are stable and valid for the network. During such a period, the network will therefore be multi-homed, and as such many of the issues relating to multi-homing in IPv6 are also relevant, albeit in a small capacity, to the renumbering procedure. A stable multi-homed situation must therefore be a requirement for renumbering without a 'flag day'.

In such a situation, however, the multi-homed state will not be permanent, and will only exist for the duration for which it is required, i.e. for the period during the renumbering procedure when both prefixes should be valid.

Renumbering can also occur, however, in a network that is already multi-homed, for example with redundant links to multiple providers. Such a site may wish to renumber for any of the situations given in the earlier section, as well as renumbering because of changes in the number of upstream providers. Until the best practice for such a situation is defined, however, its effect on renumbering is not a focus of this document.

5.9.2 Current situation with IPv6 multi-homing

Unlike IPv4 multi-homing, where PI address space is relatively easy to obtain and thus a site can broadcast its own routing information, most IPv6 addresses will be PA addresses and thus the site will have no control over routing information. Multi-homing in IPv6 therefore does not necessarily exist in the same way as in IPv4 and the multi6 working group exists to try to find a solution. Current solutions [23]Huston, G., Architectural Approaches to Multi-Homing for IPv6, October 2004. examine the potential for using identifiers within IP to identify a host, as opposed to an IP address, so that connections can continue unhindered across renumbering events. Such solutions are, however, very much in their infancy and as yet do not provide a stable solution to this problem.

5.10 Unique Local Addressing

Section 5 of [24]Hinden, R. and B. Haberman, Unique Local IPv6 Unicast Addresses, September 2004. suggests that the use of Local IPv6 addresses in a site results in making communication using these addresses independent of renumbering a site's provider based global addresses. It also points out that a renumbering episode is not triggered when merging multiple sites that have deployed centrally assigned unique local addresses[25]Hinden, R. and B. Haberman, Centrally Assigned Unique Local IPv6 Unicast Addresses, June 2004. because the FC00::/8 ULA prefix assures global uniqueness.

When merging two sites that have both deployed FD00::/8 locally assigned ULA prefixes, the chance of collision is inherently small given the pseudo-random global-ID determination algorithm of [24]Hinden, R. and B. Haberman, Unique Local IPv6 Unicast Addresses, September 2004.. Consideration of possible collisions may be prudent however unlikely the occurrence may be.

With reference to section 2 of [1]Baker, F., Lear, E. and R. Droms, Procedures for Renumbering an IPv6 Network without a Flag Day, July 2004., the adoption of ULA to assist in network renumbering can be considered a 'seasoning' of Baker's renumbering procedure: where interaction between local nodes and their services cannot suffer the inherent issues observed when migrating to a new aggregatable global unicast prefix, the use of FC00::/7 unique local addresses may offer an appropriately stable and reliable solution. The use of ULAs in single-site networks (e.g. SOHO) appears straightforward and of immediate benefit with regards renumbering episodes triggered by uplink connectivity changes. How their use scales to multi-site (e.g. Enterprise or use in ISP or transit networks) is not so evident.

The use of ULAs may not necessarily be accompanied by PA addresses. If addresses under a PA global routing prefix are not used, some form of IPv6 NAT or application layer gateway deployment will be required for ULA-only nodes internal to the network to communicate with external nodes that are not part of the same ULA topology, that is destination nodes that are not part of the same administrative domain from which the ULA allocation of the local node is made, nor part of a predetermined routing agreement between two organisations utilising different ULAs for nodes within their own sites. ULAs are not intended to be routed globally.

If addresses under a global routing prefix are also deployed, then nodes will need to cater for being multi-addressed, e.g. follow the principles laid out in RFC3484Draves, R., Default Address Selection for Internet Protocol version 6 (IPv6), February 2003.[5]. The administrator should ideally be able to set local policy such that nodes use ULAs for intranet communications and global addresses for extranet communications. The use of ULAs internally would in principle mitigate against global address renumbering of nodes.

ULAs appear to lend themselves particularly well for long-lived sessions (from the categorisation Section 4.2.3Long-term session survivability) whose nature is intra-site, for example local filestore mounts over TCP-mounted NFS: With clients using ULA source addresses to mount filestore using the ULA of an NFS server, both client and server can have their global routing prefix renumbered without consequence to ongoing local connections.

5.10.1 Private addressing

Whilst not a recommended standards-compliant deployment, sites may choose to deploy - or 'keep' deployed in the case of renumbered prefixes - otherwise-valid global routing prefixes that they have not been allocated by the registries. Providing that these addresses are never routed off-site, such behaviour does not impinge on other sites at all except that site that is later allocated the prefix (or sub-prefix therefrom) being mis-used. This is somewhat akin to site-local addressing and suffers the very same issues that have resulted in that particular architecture of addresses being officially deprecated from use.

5.11 Mobile IPv6

Mobile IPv6 (MIPv6) [17]Johnson, D., Perkins, C. and J. Arkko, Mobility Support in IPv6, June 2004. specifies routing support to permit an IPv6 host to continue using its "permanent" home address as it moves around the Internet. Mobile IPv6 supports transparency above the IP layer, including maintenance of active TCP connections and UDP port bindings. There are a number of issues to take into account when renumbering episodes occur where Mobile IPv6 is deployed:

5.11.1 Visited site renumbers when mobile

When a node is mobile and attached to a foreign network it, like any other node on the link, is subject to prefix renumbering at that site. Detecting a new prefix through the receipt of router advertisements, the mobile node can then re-bind with its home agent informing it of its care-of address - just as if it had detached from the foreign network and migrated elsewhere. Where the node receives forewarning of the renumbering episode, the Mobility specification suggests that the node explicitly solicits an update of the prefix information on its home network

5.11.2 Home site renumbers when mobile

When mobile, a host can still be contacted at its original (home) address. Should the home network renumber whilst the node is away but active (i.e. having bound to the home agent and registered a live care-of address), then it can be informed of the new global routing prefix used at the home site through the Mobile Prefix Solicitation and Mobile Prefix Advertisement ICMPv6 messages (sections 6.7 and 6.8 of RFC3775 respectively).

5.11.3 Home site renumbers when disconnected

Finally, if a mobile node is detached (i.e. no binding with the home agent exists with the node present on a foreign network) and the home network renumbers, the recommended procedure - documented as an appendix to the mobility specification and therefore not necessarily proven - is to fall back to alternative methods of 'rediscovering' its home network, using the DNS to find the new global routing prefix for the home network and therefore the Home Agent's subnet anycast address, 'guessing' at what the node's new home address would be on the basis of a 64 bit prefix and 64 bit interface identifier, and then attempting to perform registration to bind its new location.


6. Factors affecting the Renumbering Solution

There are many aspects of the renumbering procedure that may benefit from the development of bespoke renumbering tools (scripts, etc.), likely to result from the study of the experimental scenarios associated with this work. This section investigates the various factors that should be considered for developing such tools.

6.1 With or without a flag day

A network may be renumbered with or without a flag day. In the context of this document we are focusing on without a flag day, although many of the issues will still apply when renumbering is effected with a flag day.

Despite the similarities, because there is an outage of services when renumbering with a flag day, it is not necessary to ensure continuity of network connections, and almost all reconfiguration can be done during the outage, thus greatly simplifying the task of renumbering.

6.2 Frequency of renumbering episodes

The many different renumbering scenarios, discussed in Section 3Renumbering Event Triggers, can have vastly different frequencies of renumbering events. In the case of a provider offering only dynamically assigned IP addresses, it could be very frequent, for example as frequent as 'per-connection' for dial-on-demand services, or weekly for some broadband services. Such renumbering events usually only occur when a customer reconnects to such services or are explicitly cited in a subscription agreement and as such are often pre-determined.

The renumbering of a site due to upstream renumbering is relevant to all connections from a small dial-up link to a large enterprise. It is of particular interest since the end user has no control over the timing or frequency of the renumbering events. It is expected, however, that such events are likely to be very infrequent.

The other irregular renumbering events are those that occur due to end user migrating, either to a new provider, or to a new address allocation of their choosing. The timing of such an event is therefore often within the control of the end user (within reason), and are also likely to be one-off events, or at the very least, highly infrequent.

6.3 Availability of old prefix

The length of time for which the old prefix remains available has impacts on how long can be allowed for the renumbering procedure, and the maximum time for which existing sessions could continue. If end users have control over the renumbering procedure (such as when changing provider), then they can continue providing the old prefix for as long as required, within reason (such as cost aspects). This heavily mitigates the issues of session survivability, and relaxes the speed at which hosts must be reconfigured.

If the end users do not have such control, such as when the upstream provider forces the renumbering, the availability of the old prefix is determined entirely by the upstream provider's willingness to continue providing it, which is likely to be based on the technicalities of their own renumbering situation. The end user should therefore not rely on retaining the old prefix for a relatively long period of time. In addition, many situations, such as dial-on-demand with dynamic IP addresses, and nomadic networks, will lose their old prefix quickly, if not almost instantaneously.

It would be possible to continue using the old prefix internally, even when the external connectivity for that prefix is no longer active, for example to keep access to core services such as DNS servers while the transition is taking place. This should, however, be considered bad practice in case of route leaking and application confusion, and should only be used as a last resort to ensure internal continuity of service, if the availability of the old prefix is too short to allow a full transition to take place.

6.4 Freshness of service data

One of the largest issues when renumbering a network will be the effect on applications that are already running. In particular, applications that periodically contact a particular host may do an initial hostname lookup, and cache the result for use throughout the lifetime of the program. In such a situation, there is no way for the application to find out that the host in question has been renumbered, and it should stop using its already cached address. It is therefore recommended that applications should regularly request hostname lookups for the desired hosts, leaving the caching to the resolver. It is then up to the resolver to ensure that resource record TTLs are observed, and its cached response is updated as necessary.

Despite this, there is still a serious issue in that there is no method of caching resolvers knowing when a renumbering event is going to take place. If a typical RR's TTL is one day, then that should be reduced not less than a day before the renumbering event, so that resolvers will more frequently check for changed records. This will work successfully for a pre-planned renumbering event, but problems of stale, cached records will exist if the renumbering event is unplanned (e.g. by receiving a new router advertisement from upstream).

There are also cases where the use of a resolver is not practical, such as with packet filter rules. If a packet filter has been configured with explicit hostnames, these are translated to IP addresses for fast packet matching. Such a packet filter is likely to need to be reloaded for the DNS changes to be recognised.

A similar problem exists when a nameserver is renumbered. If the operating system's resolver has cached the nameserver address, it will at some point find it unavailable. To mitigate this problem, it is suggested that at least one off-site nameserver is included in the configuration. In addition, well-known anycast addresses (see Section 5.6Anycast addressing) could be used, so that the client's DNS configuration does not need to be changed at all during the renumbering event.

6.5 DNS and explicitly specified IP addresses

The basic process of renumbering, involving the introduction of a new prefix and the deprecation and eventual removal of the old prefix, could be hypothetically handled by a special tool, with no manual intervention. Such a tool would have to become significantly more complex in order to handle all the cases where IP addresses are explicitly specified (a comprehensive list is given in Section Section 7.2Explicitly named IP addresses). Other particularly notable cases that could be changed with a tool, were it to be developed, include DNS zone files and DHCPv6 configuration.

6.6 Dual-stack network?

There are several issues to consider when renumbering a dual-stacked network. In the simplest case, the IPv4 addresses will be remaining the same while the IPv6 addresses are renumbered. This could, for example, be due to an upstream renumbering, a change of IPv6 transition method (such as a tunnel), or a topology change. In such a case, the IPv4 connectivity remains unchanged, and as such can be used as a fallback during the renumbering to assist with session continuity, DNS services, etc.

The other case is when the IPv4 network is being renumbered along with the IPv6 network. Again this could be due to an upstream change, a network reconfiguration, or because the two are inter-linked - such as with the 6to4 transition mechanism. In this case, it is unlikely that the existence of IPv4 on the network can be used for any advantage, and instead many of the same issues are likely to be found when renumbering the IPv4 network as for the IPv6 network, except for the fact that more of the renumbering must be manually configured, for example by reconfiguring the stateful IPv4 DHCP configuration, or even manually configuring IPv4 addresses.

6.7 Merging networks

Renumbering of all or part of a network due to merging two or more smaller networks has many of the concerns already discussed, but it may not affect the whole network. For example, multiple disparate networks may be merged together as one entirely new subnet, and thus all hosts must be renumbered; but it is also possible that one of the networks in the merger retains its prefix, and the other network(s) merge with it.

When the networks merge, the router advertises itself, and the new prefix if appropriate, to the new hosts, and Duplicate Address Detection (DAD, see Section 5.4 of [6]Thomson, S. and T. Narten, IPv6 Stateless Address Autoconfiguration, December 1998.) must be applied by the new hosts to ensure they are not taking addresses already assigned to the existing hosts. Things become a little more complicated, however, in the case of link-local addresses. Hosts are unlikely to re-run the DAD algorithm on their link-local addresses after a network merge, so there is the possibility of an address conflict. However, as is noted in RFC2462, DAD is not completely reliable, and as such it cannot be assumed that initially after a network merge all link-local addresses will be unique.

6.8 Embedded prefix data

IPv6 global routing prefixes can also appear 'embedded' in other addressing within a site, and consideration needs to be given to how those nodes that use - particularly serve - those addresses transition to the new prefix being deployed.

One such example are those addresses as per RFC3306 [26]Haberman, B. and D. Thaler, Unicast-Prefix-based IPv6 Multicast Addresses, August 2002., which specifies how unicast prefixes can be embedded into multicast addresses for the purposes of enabling network operators to identify their multicast addresses without the need for an inter-domain allocation protocol. By way of example, a site renumbering away from prefix 2001:db8:beef::/48 might have globally-scoped multicast addresses in use under the prefix ff3e:30:2001:db8:beef::/96. How those multicast sources re-source their group addresses requires consideration.

6.9 Scalability issues

During the renumbering transition, there will be a time when two prefixes are valid for use. At this point, there will be a considerable amount of configuration that will have to be (temporarily) duplicated. In particular, routing entries on the hosts will be doubled, and there will, for a short period, be two forward records for every hostname. Security is another key scalability issue. All access control lists, packet filters, etc, will need to be updated to cope with the multiple addresses that each host will have. This could have a noticeable impact on packet filter performance, especially if it lead to, for example, the doubling of several hundred firewall rules.

The scalability issues created by the increase in configuration to cope with the temporary existence of multiple addresses per host adds a complexity in management, but how much so is up to the end-users themselves. A user may choose to do direct transitions of some services (such as web servers) from one IP address to another, without going through a stage where the service is available on either address. While that is not strictly providing a fully seamless transition, it could significantly reduce the management complexity, without a significant impact on service, especially if the DNS updates are rapid.

It should also be noted that during a renumbering event, since the DNS resource record TTLs are significantly shorter, the primary DNS servers for the domains will receive significantly more queries, as resolvers do not cache the responses for so long, and regularly check back with the master.

6.10 Equipment administrative ownership

The question of who owns and administers (also, who is authorised to administer) the site's access router is an issue in some renumbering situations. In the enterprise scenarios, the liaison between the end users and remote admins is likely to be relatively easy; this is less likely to be the case for a SOHO scenario. This is not likely to be a major issue, however, since SOHO renumbering is likely to only be required if the remote admins deem it necessary, or if the end user is sufficiently technically competent and decides to renumber their own network.

6.11 Support for Mobility?

Renumbering a network which has mobile IPv6 active is a potentially complex issue to think about. In particular, can changed router advertisements correctly reach the mobile nodes, and can they be correctly renumbered, like a node on the local network? In addition, an even more complex issue is what happens when the home agent renumbers? Is it possible for the mobile nodes to be informed and correctly renumber and continue, or will the link be irretrievably broken?

6.12 Stateless and Stateful address considerations

Most IPv6 networks are likely to be configured using StateLess Address AutoConfigurationThomson, S. and T. Narten, IPv6 Stateless Address Autoconfiguration, December 1998.[6], and in order to work through the multi-staged process as documented by Baker [1]Baker, F., Lear, E. and R. Droms, Procedures for Renumbering an IPv6 Network without a Flag Day, July 2004., the new prefix is introduced via router advertisements, and then the old prefix is deprecated, and finally removed.

Initially the router advertisements will contain only the prefix of the old network, then for a time they will contain both the old and the new, but with a shorter lifetime on the old prefix to indicate that it is deprecated. Finally the router advertisements will contain only the new prefix.

Some IPv6 networks will be configured, at least in part, by Stateful Address AutoConfiguration, using DHCPv6Droms, R., Bound, J., Volz, B., Lemon, T., Perkins, C. and M. Carney, Dynamic Host Configuration Protocol for IPv6 (DHCPv6), July 2003.[18]. Here, clients will query the DHCPv6 server and be assigned IPv6 addresses with a given lifetime.

The key difference between SLAAC and SAAC, at least from the renumbering point-of-view, is that SLAAC is both a 'push' and 'pull' protocol (the server can broadcast router advertisements, and the clients can request them), where as DHCPv6 is only a 'pull' protocol (the server only responds when it is queried by the client). This makes renumbering more complex under a DHCPv6 system, since it should be planned in advance, as the lifetimes of the DHCPv6 address assignments must be reduced before the event, so that the clients can respond in a timely manner and acquire addresses from the new prefix.

Sometimes, DHCPv6 will be used alongside SLAAC. SLAAC will provide the address assignment, and DHCPv6 will provide additional host configuration options, such as DNS servers. If any of the DHCPv6 options are directly related to the IPv6 addresses being renumbered, then the configuration must be changed at the appropriate time during the renumbering event, even though it itself does not handle the address assignments. Clients of the configuration protocol should poll the service to obtain potentially updated ancillary data, such as suggested by [27]Venaas, S. and T. Chown, Lifetime Option for DHCPv6, September 2004..

Where DHCPv6 has been employed, careful consideration about the configuration of the service is required such that administrators can be confident that clients will re-contact the service to refresh their configuration data. As alluded to in sections 22.4 and 22.5 of [18]Droms, R., Bound, J., Volz, B., Lemon, T., Perkins, C. and M. Carney, Dynamic Host Configuration Protocol for IPv6 (DHCPv6), July 2003., the configurable timers that offer servers the ability to control when clients recontacts the server about its configuration can be set such that clients rarely (if ever!) connect to validate their configuration set.

6.13 IPv6 NAT Avoidance

RFC2072 stated: "Network address translation (NAT) is a valuable technique for renumbering, or even for avoiding the need to renumber significant parts of an enterprise." That is, by 'hiding' the subnet topology and making independent of any connectivity provider the addressing model used within a site, NATs enable renumbering of entire networks because the only device that is renumbered when global addressing changes is the outside edge of the NAT devices.

However, NAT is strongly discouraged in IPv6, not least because they obscure identity - the basis for permission, authorisation, verification and validation - and thus should not be considered as being available as a solution. A significant reason to deploy IPv6 is to simplify network and application operation by NAT removal, for example to provide true end-to-end connectivity, to make simple the gateway between site and Internet, to encourage 'considered' policy as regards secure access rather than the weak and dangerous defense of hiding behind a NAT. A more detailed discussion of the motivations for 'protecting' the network architecture from NATs can be found in [28]Velde, G., IPv6 Network Architecture Protection, October 2004..

6.14 Policy and Configuration adaption

Section 3.1 of Baker [1]Baker, F., Lear, E. and R. Droms, Procedures for Renumbering an IPv6 Network without a Flag Day, July 2004. is aptly titled "Find all the places", and serves as a gentle reminder to application developers that embedding addresses is bad at best. Where common UNIX tools such as grep allow administrators to crawl the file systems of servers for places where address information is hard-coded, the proliferation of technologies such as NetInfo and other directory- or hive-based configuration schemes makes the job of finding all the places that addresses are hard-coded intractable.

Beyond the call to arms for application and services developers made by Baker, and specific to the challenges of renumbering, the following security and policy-related services that initial research has flagged as particularly troublesome:

6.14.1 Packet filters, Firewalls and ACLs

Throughout the transition from the old address set to the new, all packet filters and firewalls will need to adapt to map policy to both sets of addresses - perhaps even selectively as the old addresses become deprecated. Whilst technologies such as Router Renumbering and Neighbour Discovery automate to a large extent the transition of router and node configurations, and dynamic DNS update for the re-mapping of resource records to reflect the new addresses [29]Vixie, P., Thomson, S., Rekhter, Y. and J. Bound, Dynamic Updates in the Domain Name System (DNS UPDATE), April 1997., no such mechanism exists at present for mechanising the adaption of security policy.

Particularly troublesome policies to administer include egress filtering, where packet filters discard outbound packets that have source addresses that should not exist within the site, and filtering inbound site-local addresses in cases where two organisations are renumbering as a step toward merging their networks together (although the use of site-local addressing is now deprecated).

Where renumbering is due to a 'clean break' from previous connectivity provider, another consideration is for the ingress filtering performed by the provider. For instance, the new provider may refuse to receive into their routing topology those packets whose source address is under the old prefix, and likewise for the old provider and new prefix. Whilst it is not the business of the IETF to mandate business practice, it is likely that the provision of out-of-allocation prefix routing as part of a multi-homing service contract would be a chargeable service and not one that an enterprise trying to make a clean break away would likely be willing to pay just for the duration of transition to their new prefix.

Beyond the immediate up-stream provider, there are other policy-based considerations to take into account when renumbering. Some rudimentary authenticated access mechanisms rely on access queries coming from a particular IP network, for example, and so those application service providers will need to update their access control lists. Likewise all the internal applications (possibly meant for 'internal' eyes only) will have to have their access controls updated to reflect the change. The use of symbolic access controls (i.e. DNS domain names) rather than embedded addresses may serve to mitigate much of the distributed administrative load here, especially during the mid-renumbering states where both sets of addresses are still live and valid.

6.14.2 Monitoring tools

Network monitoring and supervisory utilities such as RMON probes, etc., are often deployed to monitor network status based on IP traffic. During a renumbering episode, the addresses for which the probes should monitoring and the addresses of logging services to which the probes report (e.g. in the case of remote SNMP logging) need to be tracked.

"Helpdesk ops" service liveness monitoring software also poses a particular problem where liveness is determined, for example, by a null transaction (e.g. for POP3 mail server, authenticating and performing a NOOP) made against a named service instance, if the name is by IP then two instances of the liveness test will be required: one on the old address to cater for those remote parties that are not yet aware of the new address, and one test against the new.


7. Application and service-oriented Issues

In this section we highlight issues and common approaches to software development that 'disrupt' protocol layering to the extent that applications become aware of renumbering episodes, even if catastrophic and without knowing how to recover without failing.

NOTE: This section, like section 6 before it, will evolve as experience grows researching the various renumbering strategies in controlled experiments - particularly in light of Section 8IETF Call to Arms.

7.1 Shims and sockets

As discussed in Section 6.14Policy and Configuration adaption, Baker's draft calls for application developers to consider the effects of renumbering whilst applications are 'live', particularly as regards caching the results of symbol resolution. Where applications maintain open connections to services over a sustained period of time (as opposed to the ephemeral nature of protocol interactions such as with HTTP), any change in either end's addressing may intrude on the application's execution - particularly if the change is abrupt or the session longer than the expiry and withdrawal time of the old addresses.

Various options may be available to minimise the risk of application disruption in this instance. A HIP-like 'shim' [30]Moskowitz, R., Host Identity Protocol Architecture, June 2004., as is being developed as a candidate solution to the general multi-homing problem, removes the tight coupling between a connection and a service's topological location: as the renumbering event takes place, the locator is updated to reflect the new address topology and the application blissfully unaware - a form of layer 3.5 mobility.

Alternatively, should the old address space be available such that a single (or subnet of) Mobile IPv6 Home Agents be deployed in the routing path of the to-be-otherwise-interrupted connection, then the endpoint being renumbered could utilise layer 3 mobility once the old prefix removed from its link, i.e. register with the Home Agent in the old prefix topology - presumably in the provider's network, formerly upstream from the site - and rely on Mobile IPv6 route optimisation to make good the additional overhead imposed by the reverse tunneling to the new prefix.

Applications that employ SCTP as opposed TCP or UDP for communication avoid all of the issues highlighted in this sub-section due to the provision of dynamic endpoint reconfiguration in the protocol (see Section 4.2Session survivability).

7.2 Explicitly named IP addresses

There are many places in the network where IP addresses are embedded as opposed to symbolic names, and finding them all to be updated during a renumbering episode is not a trivial task. This section details an evolving list of such places as surveyed as common.

Addresses may be hard-coded in software configuration files or services, in software source-code itself (which is particularly cumbersome if no source is available, e.g. a bespoke utility built to order), in firmware (for example, an access-controlling hardware dongle), or even in hardware, e.g. fixed by DIP switches.

A non-exhaustive list of instances of such addresses includes:

Some hard-coded IP address information will be held in remote locations, e.g. remote firewalls, DNS glue, etc. adding to the complexity of the search for all instances of the old prefix. Should symbols be used rather than addresses, administrative ownership of DNS - with due consideration for the TTL of resource records - and other naming services ease this particularly problematic issue of data ownership and validity.

7.3 API dilemna

In light of Section 6.4Freshness of service data, there is an open question as to whether we need an extension to the sockets API that would allow applications resolving addresses to be able to determine the freshness of the resolved data. A straw poll of networking applications demonstrated that common programming practise is to 'resolve once, bind many' during the lifetime of an application, caching the initial lookup result and assuming that it is still valid throughout. Whilst this is a perfectly valid approach for short-lived applications, where the chance of renumbering - site or the single node - increases with regards the longevity of the application, the likelihood of the resolved data being intrusively inaccurate also increases.

7.4 Server Sockets

Certain services create a server socket instance on which they intend to receive client connections throughout their execution lifetime, never re-binding that socket unless explicitly shut-down and restarted. An example would be a webserver, which may in fact bind to multiple different IP addresses to serve content for different domains where the particular business case is for customers to be allocated their 'own' IP address (e.g. for reverse DNS to reflect their branded domain name). Address space usage inefficiencies aside, the class of service that creates a server socket that persists on the initially-bound address is problematic during renumbering.

A typical work-around would be to schedule a restart of all such services having first identified whether they can operate on both address prefixes (to satisfy the middle states of Baker [1]Baker, F., Lear, E. and R. Droms, Procedures for Renumbering an IPv6 Network without a Flag Day, July 2004.), or at least to schedule their migration to the new address configuration in light of the DNS name bindings (considering caches and TTL), and the nature of existing clients that may still be bound to the old service (consider graceful migration).


8. IETF Call to Arms

In the above considerations, a number of actions would be most helpful in advancing the understanding of the practical implications and robustness of IPv6 renumbering. These include:

Effort is on-going researching the issues discussed and validating and extending [1]Baker, F., Lear, E. and R. Droms, Procedures for Renumbering an IPv6 Network without a Flag Day, July 2004. and associated activities and the results of that work will be reflected in revisions to this memo as well as fedback to the appropriate draft authors.


9. IANA Considerations

This document makes no request of IANA.


10. Security Considerations

The security considerations as outlined in [1]Baker, F., Lear, E. and R. Droms, Procedures for Renumbering an IPv6 Network without a Flag Day, July 2004. still hold, with the following supporting comments... (tbd)


11. Acknowledgements

The authors gratefully acknowledge the many helpful discussions and suggestions of their colleagues from the 6NET consortium, particularly Fred Baker, Graca Carvalho, Ralph Droms, Eliot Lear, Christian Schild, Andre Stolzé, Tina Strauf, Bernard Tuy, Gunter Van de Velde, and Stig Venaas.


12. References


12.1 Normative References

[1] Baker, F., Lear, E. and R. Droms, "Procedures for Renumbering an IPv6 Network without a Flag Day", draft-ietf-v6ops-renumbering-procedure-01 (work in progress), July 2004.
[2] Berkowitz, H., Ferguson, P., Leland, W. and P. Nesser, "Enterprise Renumbering: Experience and Information Solicitation", RFC 1916, February 1996.
[3] Ferguson, P. and H. Berkowitz, "Network Renumbering Overview: Why would I want it and what is it anyway?", RFC 2071, January 1997.
[4] Berkowitz, H., "Router Renumbering Guide", RFC 2072, January 1997.
[5] Draves, R., "Default Address Selection for Internet Protocol version 6 (IPv6)", RFC 3484, February 2003.
[6] Thomson, S. and T. Narten, "IPv6 Stateless Address Autoconfiguration", RFC 2462, December 1998 (TXT, HTML, XML).
[7] Crawford, M., "Router Renumbering for IPv6", RFC 2894, August 2000.
[8] Narten, T., Nordmark, E. and W. Simpson, "Neighbor Discovery for IP Version 6 (IPv6)", RFC 2461, December 1998 (TXT, HTML, XML).


12.2 Informative References

[9] Carpenter, B. and K. Moore, "Connection of IPv6 Domains via IPv4 Clouds", RFC 3056, February 2001.
[10] IAB and IESG, "IAB/IESG Recommendations on IPv6 Address Allocations to Sites", RFC 3177, September 2001.
[11] Fink, R. and R. Hinden, "6bone (IPv6 Testing Address Allocation) Phaseout", RFC 3701, March 2004.
[12] Templin, F., Gleeson, T., Talwar, M. and D. Thaler, "Intra-Site Automatic Tunnel Addressing Protocol (ISATAP)", draft-ietf-ngtrans-isatap-22 (work in progress), May 2004.
[13] Ernst, T. and H. Lach, "Network Mobility Support Terminology", draft-ietf-nemo-terminology-01 (work in progress), February 2004.
[14] Stewart, R., Xie, Q., Morneault, K., Sharp, C., Schwarzbauer, H., Taylor, T., Rytina, I., Kalla, M., Zhang, L. and V. Paxson, "Stream Control Transmission Protocol", RFC 2960, October 2000.
[15] Stewart, R., "Stream Control Transmission Protocol (SCTP) Dynamic Address Reconfiguration", draft-ietf-tsvwg-addip-sctp-09 (work in progress), June 2004.
[16] Hinden, R. and S. Deering, "IP Version 6 Addressing Architecture", RFC 2373, July 1998 (TXT, HTML, XML).
[17] Johnson, D., Perkins, C. and J. Arkko, "Mobility Support in IPv6", RFC 3775, June 2004.
[18] Droms, R., Bound, J., Volz, B., Lemon, T., Perkins, C. and M. Carney, "Dynamic Host Configuration Protocol for IPv6 (DHCPv6)", RFC 3315, July 2003.
[19] Troan, O. and R. Droms, "IPv6 Prefix Options for Dynamic Host Configuration Protocol (DHCP) version 6", RFC 3633, December 2003.
[20] Johnson, D. and S. Deering, "Reserved IPv6 Subnet Anycast Addresses", RFC 2526, March 1999.
[21] Jeong, J., "IPv6 Host Configuration of DNS Server Information Approaches", draft-ietf-dnsop-ipv6-dns-configuration-04 (work in progress), September 2004.
[22] Hinden, R. and S. Deering, "Internet Protocol Version 6 (IPv6) Addressing Architecture", RFC 3513, April 2003.
[23] Huston, G., "Architectural Approaches to Multi-Homing for IPv6", draft-ietf-multi6-architecture-01 (work in progress), October 2004.
[24] Hinden, R. and B. Haberman, "Unique Local IPv6 Unicast Addresses", draft-ietf-ipv6-unique-local-addr-06 (work in progress), September 2004.
[25] Hinden, R. and B. Haberman, "Centrally Assigned Unique Local IPv6 Unicast Addresses", draft-ietf-ipv6-ula-central-00 (work in progress), June 2004.
[26] Haberman, B. and D. Thaler, "Unicast-Prefix-based IPv6 Multicast Addresses", RFC 3306, August 2002.
[27] Venaas, S. and T. Chown, "Lifetime Option for DHCPv6", draft-ietf-dhc-lifetime-02 (work in progress), September 2004.
[28] Velde, G., "IPv6 Network Architecture Protection", draft-vandevelde-v6ops-nap-00 (work in progress), October 2004.
[29] Vixie, P., Thomson, S., Rekhter, Y. and J. Bound, "Dynamic Updates in the Domain Name System (DNS UPDATE)", RFC 2136, April 1997 (TXT, HTML, XML).
[30] Moskowitz, R., "Host Identity Protocol Architecture", draft-moskowitz-hip-arch-06 (work in progress), June 2004.
[31] Chown, T., "IPv6 Campus Transition Scenario Description and Analysis", draft-chown-v6ops-campus-transition-00 (work in progress), July 2004.


Authors' Addresses

  Tim J. Chown
  University of Southampton, UK
  Electronics and Computer Science
  University of Southampton
  Southampton SO17 1BJ
Phone:  +44 23 8059 5415
Fax:  +44 23 8059 2865
  Mark K. Thompson
  University of Southampton, UK
  Alan J. Ford
  University of Southampton, UK


Intellectual Property Statement

Disclaimer of Validity

Copyright Statement