Network Working Group S. Jiang Internet Draft B. Liu Intended status: Best Current Practice Huawei Technologies Co., Ltd Expires: January 03, 2012 B. Carpenter University of Auckland July 01, 2011 IPv6 Enterprise Network Renumbering Scenarios and Guidelines draft-jiang-6renum-enterprise-00.txt Status of this Memo This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at http://datatracker.ietf.org/drafts/current/. 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." This Internet-Draft will expire on January 03, 2012. Copyright Notice Copyright (c) 2011 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. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License. Abstract This document analyzes the enterprise renumbering events and gives the recommendations among the existing renumbering mechanisms. According to the different stages of renumbering events, Jiang & Liu Expires January 03, 2012 [Page 1] Internet-Draft draft-jiang-6renum-enterprise-00.txt July 2011 considerations and best current recommendations are described in three categories: during network design, for preparation of renumbering, and during renumbering operation. A gap inventory is listed at the end of this document. Table of Contents 1. Introduction ................................................. 3 2. Enterprise Network Illustration for Renumbering .............. 3 3. Enterprise Network Renumbering Scenario Categories ........... 4 3.1. Renumbering caused by External Network Factors........... 4 3.2. Renumbering caused by Internal Network Factors........... 5 4. Network Renumbering Considerations and Best Current Recommendations ................................................. 5 4.1. Considerations and Recommendations during Network Design. 6 4.2. Considerations and Recommendations for the Preparation of Renumbering .................................................. 8 4.3. Considerations and Recommendations during Renumbering Operation .................................................... 9 5. Gap Inventory ............................................... 11 6. Security Considerations ..................................... 12 7. IANA Considerations ......................................... 12 8. Acknowledgements ............................................ 12 9. Change Log [RFC Editor please remove] ....................... 12 10. References ................................................. 13 10.1. Normative References .................................. 13 10.2. Informative References ................................ 14 Author's Addresses ............................................. 15 Jiang, et al. Expires January 03, 2012 [Page 2] Internet-Draft draft-jiang-6renum-enterprise-00.txt July 2011 1. Introduction IPv6 site renumbering is considered difficult. Network managers would turn to Provider Independent (PI) addressing for IPv6 to attempt to minimize the need for future renumbering. However, widespread use of PI may create very serious BGP4 scaling problems. It is thus desirable to develop tools and practices that may make renumbering a simpler process to reduce demand for IPv6 PI space. This document undertakes scenario descriptions, including documentation of current capability inventories and existing BCPs, for enterprise networks. It takes the analysis conclusions from [RFC5887] and other relevant documents as the primary input. This document focuses on IPv6 only, by leaving IPv4 out of scope. Dual-stack network or IPv4/IPv6 transition scenarios are out of scope, too. According to the different stages of renumbering events, considerations and best current recommendations are described in three categories: during network design, for preparation of renumbering, and during renumbering operation. A gap inventory is listed at the end of this document. 2. Enterprise Network Illustration for Renumbering The enterprise network architecture is illustrated as the figure below. From the renumbering perspective of view, these entities relevant to renumbering are highlighted. Address reconfiguration is fulfilled either by DHCPv6 or ND protocols. Static address assignment is not considered in this version. During the renumbering event, the DNS records need to be synchronized while routing tables, ACLs and IP filtering tables in various gateways also need to be updated, too. Jiang, et al. Expires January 03, 2012 [Page 3] Internet-Draft draft-jiang-6renum-enterprise-00.txt July 2011 Uplink 1 Uplink 2 | | +---+---+ +---+---+ +---- |Gateway| --------- |Gateway| -----+ | +-------+ +-------+ | | Enterprise Network | | +------+ +------+ +------+ | | | APP | | DHCP | | DNS | | | |Server| |Server| +Server+ | | +---+--+ +---+--+ +--+---+ | | | | | | | ---+--+---------+------+---+- | | | | | | +--+---+ +---+--+ | | |Router| |Router| | | +--+---+ +---+--+ | | | | | | -+---+----+-------+---+--+- | | | | | | | | +-+--+ +--+-+ +--+-+ +-+--+ | | |Host| |Host| |Host| |Host| | | +----+ +----+ +----+ +----+ | +----------------------------------------+ Figure 1 Enterprise network illustration It is assumed that IPv6 enterprise networks are IPv6-only, or dual- stack in which a logical IPv6 plane is independent from IPv4. The complicated IPv4/IPv6 co-existing scenarios are out of scope. This document focuses on the unicast addresses; site-local, link- local, multicast and anycast addresses are out of scope. 3. Enterprise Network Renumbering Scenario Categories In this section, we category enterprise network renumbering scenarios mainly according to different reasons. Some of renumbering reasons described in [RFC2071] has out of date, or not suitable in IPv6, or not suitable for enterprise networks. 3.1. Renumbering caused by External Network Factors The most influential external network factor is the uplink ISP. Jiang, et al. Expires January 03, 2012 [Page 4] Internet-Draft draft-jiang-6renum-enterprise-00.txt July 2011 o The enterprise network switches to a new ISP. Of course, the prefixes received from different ISPs are different. This is the most common scenario. Whether there is an overlap time between the old and new ISPs would also influence the possibility whether the enterprise can fulfill renumbering without a flag day [RFC4192]. o The renumbering event may be initiated by receiving new prefixes from the same uplink. The typical scenario is that the DHCP server in ISP delegates a new prefix to the enterprise network. Or the enterprise network may be switched to a different location within the network topology of the same ISP due to various considerations, such as commercial, performance or services reasons, etc. The ISP itself may also be renumbered due to topology change or migration to a different or additional prefix. These ISP renumbering events would initiate enterprise network renumbering events, of course. o The enterprise network adds new uplink(s) for multihoming purpose. This may not a typical renumbering because the original addresses will not be changed. However, initial numbering may be considered as a special renumbering event. If the administrators only want part of the network to have multiple prefixes, the renumbering process should be carefully managed. 3.2. Renumbering caused by Internal Network Factors o As companies split, merge, grow, or reorganize, the enterprise network architectures may need to be re-built. This will trigger the internal renumbering. 4. Network Renumbering Considerations and Best Current Recommendations In order to carry out renumbering in an enterprise network, systematic planning and administrative preparation are needed. Carefully planning and preparation could make the renumbering process smoother. This section tries to give the recommended solutions or strategies for the enterprise renumbering among the existing mechanisms. There are a few gaps analyzed by [I-D.liu-6renum-gap-analysis]. If they are filled in the future, the enterprise renumbering may be processed more automatically, with fewer issues. Jiang, et al. Expires January 03, 2012 [Page 5] Internet-Draft draft-jiang-6renum-enterprise-00.txt July 2011 4.1. Considerations and Recommendations during Network Design This section describes the renumbering relevant considerations or issues that a network architect should carefully plan when he builds or designs a new network. - Prefix Delegation In a large or a multi-site enterprise network, the prefix should be carefully managed, particularly during renumbering events. Prefix information needs to be delegated from router to router. The DHCPv6 Prefix Delegation options [RFC3633] provide a mechanism for automated delegation of IPv6 prefixes. DHCPv6 PD options may also be used between the enterprise routers and their upstream ISPs. - Usage of FQDN It is recommended that Fully-Qualified Domain Names (FQDNs) should be used to configure network connectivity, such as tunnels. The capability to use FQDNs as endpoint names has been standardized in several RFCs, such as [RFC5996], although many system/network administrators do not realize that it is there and works well as a way to avoid manual modification during renumbering. Service Location Protocol [RFC2608] and multicast DNS with SRV records for service discovery can reduce the number of places that IP addresses need to be configured. - Address Types This document focuses on the dynamic-configured global unicast addresses in enterprise networks. They are the targets of renumbering events. Manual-configured addresses are not scalable in medium to large sites, hence be out of scope. However, some hosts such as servers may need static addresses. Manual-configured addresses/hosts should be avoided as much as possible. [Open Question to WG] What we can do regarding to manual configured hosts and static addresses, which do need to be changed? Unique Local Address (ULA, [RFC4193]) may be used on local routers or servers, which only intends for local communications, usually Jiang, et al. Expires January 03, 2012 [Page 6] Internet-Draft draft-jiang-6renum-enterprise-00.txt July 2011 inside of enterprise networks. Normally, they do NOT need to be changed during the renumbering event. [Open Question to WG] Is anyone actually using ULAs? - Address configuration models In IPv6 networks, there are two auto-configuration models for address assignment: the Stateless Address Auto-Configuration (SLAAC) by Neighbor Discovery (ND, [RFC4861, RFC4862]) and the stateful address configuration by Dynamic Host Configuration Protocol for IPv6 (DHCPv6, [RFC3315]). In the latest work, DHCPv6 can also support host-generate address model by assigning prefix through DHCPv6 messages [I-D.ietf-dhc-host-gen-id]. ND is considered easier to renumber by broadcasting a Router Advertisement message with a new prefix. DHCPv6 can also trigger the renumbering process by sending unicast RECONFIGURE messages though it may cause a large number of interactions between hosts and DHCPv6 server. In principle, a network should choose only one address configuration model and employs either ND or DHCPv6. This document has no preference between ND and DHCPv6 address configuration models. However, since DHCPv6 is also used to configure many other network parameters, there are ND and DHCPv6 co-existing scenarios. The current protocols do not effectively prevent that both SLAAC and DHCPv6 address assignment are used in the same network (see M bit analysis in section 5.1.1 [RFC5887]). It is network architects' job to make sure only one configuration model is employed. Even in a large network that contains several subnet works, it is recommended not to mix the two address configuration models though isolately using them in different subnet works may reduce the risk partly. - DNS It is recommended that the site have an automatic and systematic procedure for updating/synchronising its DNS records, including both forward and reverse mapping [RFC2874]. Manually on-demand updating model is considered as a harmful problem creator in renumbering event. A6 DNS record model is recommended over AAAA record model for renumbering purpose [RFC2874, RFC3364]. Jiang, et al. Expires January 03, 2012 [Page 7] Internet-Draft draft-jiang-6renum-enterprise-00.txt July 2011 In order to simplify the operation procedure, the network architect should combine the forward and reverse DNS updates in a single procedure. If a small site depends on its ISP's DNS system rather than maintains its own one. When renumbering, it requires administrative coordination between the site and its ISP. Alternatively, the DNS synchronizing may be completed through the Secure Dynamic DNS Update. - Security Any automatic renumbering scheme has a potential exposure to hijacking at the moment that a new address is announced. Proper network security mechanisms should be employed. Secure Neighbor Discovery (SEND, [RFC3971]), which does not widely deployed, is recommended to replace ND. Alternatively, certain lightweight renumbering specific security mechanism may be developed in the future. DHCPv6 build-in secure mechanisms, like Secure DHCPv6 [I-D.ietf-dhc-secure-dhcpv6] or authentication of DHCPv6 messages [RFC3315] are recommended. - Miscellaneous A site or network should also avoid to embed addresses from other sites or networks in its own configuration data. Instead, the Fully-Qualified Domain Names should be used. Thusness, these connectivities can survive after renumbering events. This also applies to host-based connectivities. 4.2. Considerations and Recommendations for the Preparation of Renumbering It is not possible to reduce a prefix's lifetime to below two hours. So, renumbering should not be an unplanned sudden event. This issue could only be avoided by early planning and preparation. This session describes several recommendations for the preparation of enterprise renumbering event. By adopting these recommendations, a site could be renumbered easier. However, these recommendations are not cost free. They might increase the daily burden of network operation. Therefore, only these networks that are expected to be renumbered soon or very frequently should adopt these recommendations with the balance consideration between daily cost and renumbering cost. - Reduce the address preferred time or valid time or both. Jiang, et al. Expires January 03, 2012 [Page 8] Internet-Draft draft-jiang-6renum-enterprise-00.txt July 2011 Long-lifetime addresses may cause issues for renumbering events. Particularly, some offline hosts may reconnect back using these addresses after renumbering events. Shorter preferred lifetime with relevant long valid lifetime may get short transition period for renumbering event and avoid address renew too frequent. - Reduce the DNS record TTL. The DNS record TTL on the local DNS server should be manipulated to ensure that stale addresses are not cached. - Reduce the DNS configuration lifetime on the hosts. Since the DNS server could be renumbered as well, the DNS configuration lifetime on the hosts should also be reduced if renumbering events are expected. The DNS configuration can be done through either ND [RFC6106] or DHCPv6 [RFC3646]. However, DHCPv6 DNS option does not include associated lifetime. It should be updated. 4.3. Considerations and Recommendations during Renumbering Operation Renumbering events are not instantaneous events. Normally, there is a transition period, in which both the old prefix and the new prefix are used in the site. Better network design and management, better pre-preparation and longer transition period are helpful to reduce the issues during renumbering operation. - Within/without a flag day As is described in [RFC4192], "a 'flag day' is a procedure in which the network, or a part of it, is changed during a planned outage, or suddenly, causing an outage while the network recovers." If renumbering event is processed within a flag day, the network service/connectivity will be outage for a period till the renumbering event is completed. It is efficient and provides convenient for network operation and management. But network outage is usually unacceptable for end users and the enterprises. Renumbering procedure without a flag day provides smooth addresses switching, but much more operational complexity and difficulty is introduced. - Transition period Jiang, et al. Expires January 03, 2012 [Page 9] Internet-Draft draft-jiang-6renum-enterprise-00.txt July 2011 If renumbering transition period is longer than all addresses lifetime, after which the addresses lease expire, each host will automatically pick up its new IP address. In this case, it would be the DHCP server or Router Advertisement itself that automatically accomplishes client renumbering. - Network initiative enforced renumbering If the network has to enforce renumbering before addresses lease expire, the network should initiate enforcement messages, either in Router Advertisement messages or DHCPv6 RECONFIGURE messages. - Impact to branch/main sites Renumbering in main/branch site may cause impact on branch/main site communication. The routes, ingress filtering of site's gateways, and DNS may need to be updated. This needs carefully planning and organizing. - DNS record update and DNS configuration on hosts DNS records should be updated if hosts are renumbered. If the site depends on ISP's DNS system, it should report the new host's DNS records to its ISP. During the transition period, both old and new DNS records are valid. If the TTL of DNS records is shorter than the transition period, administrative operation may not be necessary. DNS configuration on hosts should be updated if local recursive DNS servers are renumbered. During the transition period, both old and new DNS addresses may co-exist on the hosts. If the lifetime of DNS configuration is shorter than the transition period, name resolving failure may not be reduced to minimum. A notification mechanism may be needed to indicate the hosts that a renumbering event of local recursive DNS happens or is going to take place. - Router awareness In a site with multiple border routers, all border routers should be aware of partial renumbering in order to correctly handle inbound packets. Internal forwarding tables need to be updated. - Border filtering In a multihomed site, an egress router to ISP A could normally filter packets with source addresses from other ISPs. The egress router connecting to ISP A should be notified if the egress router Jiang, et al. Expires January 03, 2012 [Page 10] Internet-Draft draft-jiang-6renum-enterprise-00.txt July 2011 connecting to ISP B initiates a renumbering event in order to properly act filter function. - Tunnel concentrator renumbering Tunnel concentrator itself might be renumbered. This change should be reconfigured to relevant hosts or router, unless the configuration of tunnel concentrator was based on FQDN. 5. Gap Inventory This section lists a few issues that still remain unsolvable. Some of them may be inherently unsolvable. - Manual or script-driven procedures will break the completely automatic host renumbering. - Some environments like embedded systems might not use DHCP or SLAAC and even configuration scripts might not be an option. This creates special problems that no general-purpose solution is likely to address. - TCP and UDP flows can't survive at renumbering event at either end. - Some address configuration data might be widely dispersed and much harder to find, even will inevitably be found only after the renumbering event. - The embedding of IPv6 unicast addresses into multicast addresses and the embedded-RP (Rendezvous Point) [RFC3956] will cause issues when renumbering. - Changing the unicast source address of a multicast sender might also be an issue for receivers. - When a renumbering event takes place, entries in the state table of tunnel concentrator that happen to contain the affected addresses will become invalid and will eventually time out. However, this can be considered as harmless though it takes sources on these devices for a while. - A site that is listed in a black list can escape that list by renumbering itself. The site itself of course will not initiatively to report its renumbering and the black list may not be able to monitor or discover the renumbering event. Jiang, et al. Expires January 03, 2012 [Page 11] Internet-Draft draft-jiang-6renum-enterprise-00.txt July 2011 - Multihomed site, using SLAAC for one address prefix and DHCPv6 for another, would clearly create a risk of inconsistent host behaviour and operational confusion. - The impact of portion renumbering may need to be analyzed further. Some of these issues can be considered as harmless or have minimum impacts. 6. Security Considerations A site that is listed in a black list can escape that list by renumbering itself. Any automatic renumbering scheme has a potential exposure to hijacking at the moment that a new address is announced. Proper network security mechanisms should be employed. SEND is recommended to replace ND. Alternatively, certain lightweight renumbering specific security mechanism may be developed in the future. DHCPv6 build-in secure mechanisms, like Secure DHCPv6 [I-D.ietf-dhc-secure-dhcpv6] or authentication of DHCPv6 messages [RFC3315] are recommended. The security updates will need to be made in two stages (immediately before and immediately after the event). [Editor note: this section needs further work.] 7. IANA Considerations This draft does not request any IANA action. 8. Acknowledgements This work is illumined by RFC5887, so thank for RFC 5887 authors, Randall Atkinson and Hannu Flinck. Useful ideas were also illumined by documents from Tim Chown and Fred Baker. The authors also want to thank Wesley George, Olivier Bonaventure and other 6renum members for valuable comments. 9. Change Log [RFC Editor please remove] draft-jiang-6renum-enterprise-00, original version, 2011-07-01 Jiang, et al. Expires January 03, 2012 [Page 12] Internet-Draft draft-jiang-6renum-enterprise-00.txt July 2011 10. References 10.1. Normative References [RFC2608] Guttman, E., Perkins, C., Veizades, J., and M. Day "Service Location Protocol, Version 2", RFC 2608, June 1999. [RFC3315] Droms, R., Bound, J., Volz, B., Lemon, T., Perkins, C., and M. Carney, "Dynamic Host Configuration Protocol for IPv6 (DHCPv6)", RFC 3315, July 2003. [RFC3633] Troan, O., and R. Droms, "IPv6 Prefix Options for Dynamic Host Configuration Protocol (DHCP) version 6", RFC 3633, December 2003. [RFC3646] R. Droms, "DNS Configuration options for Dynamic Host Configuration Protocol for IPv6 (DHCPv6)", RFC 3646, December 2003. [RFC3956] Savola, P., and B. Haberman, "Embedding the Rendezvous Point (RP) Address in an IPv6 Multicast Address", RFC 3956, November 2004 [RFC3971] Arkko, J., Ed., Kempf, J., Zill, B., and P. Nikander "SEcure Neighbor Discovery (SEND)", RFC 3971, March 2005 [RFC4193] Hinden, R., and B. Haberman, "Unique Local IPv6 Unicast Addresses", RFC 4193, October 2005. [RFC4861] Narten, T., Nordmark, E., Simpson, W., and H. Soliman, "Neighbor Discovery for IP version 6 (IPv6)", RFC 4861, September 2007. [RFC4862] Thomson, S., Narten, T., and T. Jinmei, "IPv6 Stateless Address Autoconfiguration", RFC 4862, September 2007. [RFC5996] Kaufman, C., Hoffman, P., Nir, Y., and P. Eronen, "Internet Key Exchange Protocol Version 2 (IKEv2)", RFC 5996, September 2010. [RFC6106] Jeong, J., Ed., Park, S., Beloeil, L., and S. Madanapalli "IPv6 Router Advertisement Option for DNS Configuration", RFC 6106, November 2011. Jiang, et al. Expires January 03, 2012 [Page 13] Internet-Draft draft-jiang-6renum-enterprise-00.txt July 2011 10.2. Informative References [RFC2071] Ferguson, P., and H. Berkowitz., "Network Renumbering Overview: Why would I want it and what is it anyway?", RFC 2071, January 1997. [RFC2874] Crawford, M., and C. Huitema, "DNS Extensions to Support IPv6 Address Aggregation and Renumbering", RFC 2874, July 2000. [RFC3364] R. Austein, "Tradeoffs in Domain Name System (DNS) Support for Internet Protocol version 6 (IPv6)", RFC 3364, August 2002. [RFC4192] Baker, F., Lear, E., and R. Droms, "Procedures for Renumbering an IPv6 Network without a Flag Day", RFC 4192, September 2005. [RFC5887] Carpenter, B., Atkinson, R., and H. Flinck, "Renumbering Still Needs Work", RFC 5887, May 2010. [I-D.ietf-dhc-secure-dhcpv6] Jiang, S., and S. Shen, "Secure DHCPv6 Using CGAs", working in progress. [I-D.ietf-dhc-host-gen-id] S. Jiang, F. Xia, and B. Sarikaya, "Prefix Assignment in DHCPv6", draft-ietf-dhc-host-gen-id (work in progress), April, 2011. [I-D.liu-6renum-gap-analysis] Liu, B., and S. Jiang, "IPv6 Site Renumbering Gap Analysis", working in progress. Jiang, et al. Expires January 03, 2012 [Page 14] Internet-Draft draft-jiang-6renum-enterprise-00.txt July 2011 Author's Addresses Sheng Jiang Huawei Technologies Co., Ltd Huawei Building, No.3 Xinxi Rd., Shang-Di Information Industry Base, Hai-Dian District, Beijing P.R. China EMail: jiangsheng@huawei.com Bing Liu Huawei Technologies Co., Ltd Huawei Building, No.3 Xinxi Rd., Shang-Di Information Industry Base, Hai-Dian District, Beijing P.R. China EMail: leo.liubing@huawei.com Brian Carpenter Department of Computer Science University of Auckland PB 92019 Auckland, 1142 New Zealand EMail: brian.e.carpenter@gmail.com Jiang, et al. Expires January 03, 2012 [Page 15]