Network Working Group X. Xu Internet Draft Huawei Intended status: Informational Expires: July 2010 January 25, 2010 Routing Architecture for the Next Generation Internet (RANGI) draft-xu-rangi-02.txt Status of this Memo This Internet-Draft is submitted to IETF in full conformance with the provisions of BCP 78 and BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet-Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt. The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html. This Internet-Draft will expire on July 25, 2010. Copyright Notice Copyright (c) 2009 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. Xu Expires July 25, 2010 [Page 1] Internet-Draft Routing Architecture July, 2009 for the Next Generation Internet (RANGI) Abstract IRTF Routing Research Group (RRG) is exploring a new routing and addressing architecture to address the issues with the current Internet, e.g., mobility, multi-homing, traffic engineering, and especially the routing scalability. This document describes a new identifier (ID)/locator split based routing and addressing architecture, called Routing Architecture for the Next Generation Internet (RANGI), in an attempt to deal with the above problems. Table of Contents 1. Introduction.................................................3 2. Architecture Description.....................................3 2.1. Host Identifiers........................................3 2.2. Host Locators...........................................5 2.3. Packet Formats..........................................6 2.4. ID/Locator Mapping Resolution...........................6 2.5. Routing and Forwarding System...........................8 2.5.1. Host Behavior......................................8 2.5.2. LDBR Behavior......................................8 2.5.3. Non-LDBR Behavior..................................8 2.5.4. Forwarding Procedures..............................8 2.6. Multi-homing and Traffic-Engineering...................10 2.7. Mobility...............................................12 3. Summary.....................................................12 4. Security Considerations.....................................13 5. IANA Considerations.........................................13 6. Acknowledgments.............................................13 7. References..................................................14 Xu Expires July 25, 2010 [Page 2] Internet-Draft Routing Architecture July, 2009 for the Next Generation Internet (RANGI) 1. Introduction The Default Free Zone (DFZ) routing table size has been growing at an increasing and potentially alarming rate for several years, which has detrimental impact on the routing system scalability and the routing convergence performance. This so-called routing scalability issue has drawn significant attention from both industry and academia. After much discussion following the IAB Routing and Addressing workshop [RAWS] in Amsterdam, a common conclusion was reached that the explosive growth in the DFZ routing table is mainly caused by the wide adoption of multi-homing, traffic engineering and provider- independent address. However, the underlying reason for this issue is the overloading of IP address semantics of both identifiers and locators. This overloading makes it impossible to renumber IP addresses in a topologically aggregatable way. At present, the IRTF Routing Research Group (RRG) is chartered to explore a new routing and addressing architecture which is expected to support the multi-homing, traffic-engineering, mobility and simplified renumbering features in a more scalable way. This document describes an ID/locator split approach, called Routing Architecture for the Next Generation Internet (RANGI), which aims to deal with the above issues. Similar with Host Identity Protocol (HIP) [RFC4423], RANGI also introduces a host identifier (ID) layer between the IPv6 network layer and the transport layer. As a result, the transport-layer associations (i.e., TCP connections and UDP associations) are no longer bound to IP addresses, but to the host IDs. The major difference from the HIP is that the host IDs in RANGI are hierarchical and cryptographic host IDs which have organizational structure. As a result, the ID/locator mapping system for such identifiers has reasonable business model and clear trust boundaries. In addition, RANGI uses special IPv4-embeded IPv6 addresses as locators. With such locators, site-controlled traffic-engineering can be easily achieved and the renumbering burden during ISP change is also eliminated greatly. Besides, the deployment cost of such a new architecture is reduced greatly and the new architecture could be deployed in a more incremental way. 2. Architecture Description 2.1. Host Identifiers Similar to HIP, RANGI is a host-based ID/locator split architecture. The host IDs in RANGI are hierarchical and 128-bit long. As depicted in Figure 1, a host ID consists of two parts: the leftmost n-bit Xu Expires July 25, 2010 [Page 3] Internet-Draft Routing Architecture July, 2009 for the Next Generation Internet (RANGI) (Note that the suitable value of ''n'' has not been determined yet) part is the Administrative Domain (AD) ID which has embedded organizational affiliation and global uniqueness, and the remaining part is the Local Host ID which is generated by computing a cryptographic one-way hash function from a public key of the ID owner and auxiliary parameters (e.g., the AD ID). The binding between the public key and the host ID can be verified by re-computing the hash value and by comparing the hash with the host ID. As these identifiers are expected to be used along with IPv6 addresses at both applications and APIs, especially in the transition mechanisms defined in [RANGI-PROXY], it is desired to explicitly distinguish host IDs from IPv6 addresses (i.e., locators) and vice versa. Hence, a separate prefix for identifiers should be allocated by the IANA. As a result several leftmost bits in the AD ID field should be reserved for this purpose. |<------- n bits --------->|<-- 128-n bits-->| +--------------------------+-----------------+ | Administrative Domain ID | Local Host ID | +--------------------------+-----------------+ | \ | \ | \ | \ | \ +-----------+-------------+-------------+ | Country ID| Authority ID| Region ID | <------Example +-----------+-------------+-------------+ Figure 1. Host Identifier Structure The approach of generating hierarchical host IDs in RANGI is similar to the Cryptographically Generated Addresses (CGA) [RFC3972]. The major difference is that the prefix of the RANGI host ID is AD ID, rather than ordinary IPv6 address prefix. In CGA, the process of generating a new address takes three input values: a 64-bit subnet prefix, the public key of the address owner as a DER-encoded ASN.1 structure of the type SubjectPublicKeyInfo and the security parameter Sec, which is an unsigned three-bit integer. In contrast, the process of generating a new host ID in RANGI also takes three input values: the n-bit AD ID, the public key of the host ID owner and the security parameter Sec. Therefore, if we set the value of n to 64, host IDs can be compatible with CGAs. The benefits of the hierarchical host ID in RANGI include but not limited to: 1) manage the global identifier namespace in a scalable Xu Expires July 25, 2010 [Page 4] Internet-Draft Routing Architecture July, 2009 for the Next Generation Internet (RANGI) way; 2) hold a reasonable economic model and clear trust boundaries in the corresponding ID/locator mapping system; 3) ease the transition from the current Internet to the RANGI. In RANGI, the global uniqueness of host IDs is guaranteed through some registration mechanism. Since the AD IDs are globally unique and owned by the corresponding host ID registration and administrative authorities from different countries respectively, the Local Host IDs are only required to be unique within the corresponding AD scope. The resolution infrastructure for flat labels has no ''pay-for-your- own'' model, as names are stored at essentially random nodes (See Layered Naming Architecture (LNA) [LNA]). In contrast, the resolution infrastructure for hierarchical host IDs in RANGI has reasonable business model and clear trust boundaries since host IDs can be stored in the corresponding nodes according to their organizational structures. To some extent, the business model of the ID/locator mapping system is similar with that for the Domain Name Service (DNS) In the transition mechanisms for RANGI described in [RANGI-PROXY], the identifiers of RANGI-aware hosts are treated as ordinary IPv6 addresses by legacy hosts. When a router receives a packet using a host ID as the destination address, it needs to forward the packet according to the destination IPv6 address as normal. In the end, the packet will be forwarded to a dedicated proxy that is responsible for translating the packets between RANGI and IPv6. Since the identifiers are hierarchical and delegation-oriented aggregatable, such identifier-based routing will not cause any routing scalability issue. For more details, please refer to [RANGI-PROXY]. 2.2. Host Locators The host locators in RANGI are ordinary IPv6 addresses. Since the IPv4/IPv6 coexistence and transition will last for a long period, in order to reduce the deployment cost of this new routing and addressing architecture, RANGI uses specific IPv4-embeded IPv6 addresses as locators. As shown in Figure 2, the leftmost 96-bit part of the locator is Locator Domain Identifier (LD ID) while the rightmost 32-bit part is filled with an IPv4 address which is required to be locally unique within the scope of corresponding LD. LD IDs are used to globally identify each autonomous site network which could adopt independent IPv4 address space (either public or private IPv4 addresses). Actually, LD IDs are Provider-Assigned (PA) /96 IPv6 prefixes which are topologically aggregatable in providers' networks. Xu Expires July 25, 2010 [Page 5] Internet-Draft Routing Architecture July, 2009 for the Next Generation Internet (RANGI) |<------- 96 bits -------->|<---- 32 bits--->| +--------------------------+-----------------+ | LD ID | IPv4 | +--------------------------+-----------------+ Figure 2. Host Locator Structure Similar with the Intra-Site Automatic Tunnel Addressing Protocol (ISATAP) [RFC5214], this specific locator can be used for automatically tunneling IPv6 packets over the destination IPv4 site networks since the embedded IPv4 address of the destination locator is used as the tunnel destination address directly. 2.3. Packet Formats RANGI reuse IPv6 protocol stack and packet format to maximum extent. The host ID simply appears as an option in the Destination Option Header, whereas the locator is filled as the IPv6 address in the IPv6 header. Packets sent from a RANGI host can be protected by attaching the public key and auxiliary parameters and by signing the packets with the corresponding private key. The protection works without a certification authority or any security infrastructure. 2.4. ID/Locator Mapping Resolution ID/locator split implies a need for storing and distributing the mappings from identifiers to locators. In RANGI, the mappings from Fully Qualified Domain Names (FQDNs) to identifiers are stored in the DNS system, while the mappings from host IDs to locators are stored in a distributed id/locator mapping system (e.g., a reverse DNS system). In a reverse DNS based mapping infrastructure, if there are too many entries to be maintained by the authoritative server of a given AD, Distribute Hash Table (DHT) technology can be used to scale that authoritative server because the Local Host ID part is a flat label. As a result, the mappings belonging to a given AD will be maintained by a group of DHT peers in a distributed way. As a result, the robustness of DHT is inherited naturally into the ID/Locator mapping system. Meanwhile, there is no trust issue since each of these ADs runs their own DHT ring which only maintains their presidial mappings. A detailed lookup example is given as follows: Xu Expires July 25, 2010 [Page 6] Internet-Draft Routing Architecture July, 2009 for the Next Generation Internet (RANGI) 1. A host ID will be transformed to a FQDN format string. Firstly, a host ID is expressed as "country-code.authority-code.region- code.local-host-ID" by inserting dots between adjacent fields, then by reversing the fields and attaching with the suffix "RANGI.EXAMPLE." it is transformed into a FQDN-format string "local- host-ID.region-code.authority-code.country-code.RANGI.EXAMPLE." 2. The FQDN-format string is used as a key to locate the DNS server or the DHT ring maintaining the desired resource records in the reverse DNS infrastructure. If the DHT ring is located, the Local Host ID part SHOULD be used as a key to locate the associated peer in the DHT ring. 3. After an authoritative DNS server or a DHT peer has been located, the Local Host ID is used to find out the related entries for that identifier. In order to facilitate such a lookup process, a new sub-domain called ''RANGI.EXAMPLE.'' needs to be inserted into the current domain name space tree structure. This domain can delegate its sub-domains according to the hierarchy of the FQDN-format string of the host ID. A new Resource Record (RR) named RANGI is also defined for the ID-to- Loc mappings, in which the NAME field is filled with the FQDN-format string of a host ID, while the RDATA field is filled with the corresponding locator information, including but not limited to an IPv6 address and its preference, and so on. The resolution infrastructure for flat names has no ''pay-for-your- own''model, as the flat names are stored at essentially random nodes. In contrast, the resolution infrastructure for the hierarchical host IDs, as used in RANGI, has reasonable business and trust models, because the hierarchical host IDs have clear organization affiliation and the identifier resources can be allocated and managed with clear administrative boundaries. To prevent the Man-in-the-Middle attack during mapping lookups, the DNS Security Extensions (DNSSEC) [RFC2065] is strongly recommended for the origin authentication and integrity assurance of the DNS data. The mechanism defined in DNS UPDATE [RFC2136] is directly used for dynamically updating the RRs in the corresponding zone. To support mobility, the TTL of a RANGI RR should be set to 0 or a very small value to prevent a DNS resolver caching antique mapping information. However, if a host (i.e., Correspondence Node) wants to cache the RR of the communicating host (i.e., Mobile Node), it can reset the TTL to a reasonable value internally. Xu Expires July 25, 2010 [Page 7] Internet-Draft Routing Architecture July, 2009 for the Next Generation Internet (RANGI) To secure the dynamic update of ID-to-Locator mappings, the mechanism defined in the Cryptographically Generated Addresses (CGA) [RFC3972] is used for the purposes of data integrity protection and origin authentication, e.g., the update message is attached with the public key of the update sender and auxiliary parameters, and is signed with the corresponding private key. 2.5. Routing and Forwarding System Within RANGI, LDs are connected via Locator Domain Border Routers (LDBRs). A LDBR has at least one locally unique IPv4 address in each LD to which it is connected. The adjacent LDBRs exchange LD ID or LD prefix (Specific LD IDs can be aggregated into one LD prefix) reachability information with an inter-LD routing protocol. In fact, the Border Gateway Protocol (BGP) for IPv6 address family can be used directly as the inter-LD routing protocol. 2.5.1. Host Behavior Generally, a RANGI host needs to obtain the locator of the destination host from the ID/locator mapping system before initializing a communication. If the communication parties share a same LD ID, they can exchange packets directly over an IPv4 tunnel. Otherwise, the traffic will be relayed by LDBRs through IPv4 tunnels. Hosts can get the IPv4 addresses of their local LDBRs in several ways, e.g., a new Dynamic Host Configuration Protocol (DHCP) option, or a well-known LD-scope anycast address specific for LDBRs. 2.5.2. LDBR Behavior LDBRs establish BGP sessions among them so as to exchange LD reachability information (i.e., IPv6 routing information with the mask length less than /96). LDBRs forward RANGI packets according to the destination IPv6 addresses (i.e., locators) as normal. 2.5.3. Non-LDBR Behavior The non-LDBRs just need to support IPv4 forwarding capability. So there is no need to upgrade them. 2.5.4. Forwarding Procedures RANGI introduces a two-level routing mechanism which is composed of LD ID based routing and IPv4 address based routing. The former is Xu Expires July 25, 2010 [Page 8] Internet-Draft Routing Architecture July, 2009 for the Next Generation Internet (RANGI) used for inter-LD routing while the latter is used for intra-LD routing. A simple RANGI routing procedure is illustrated in Figure 3. Host A (as source host) looks up the current locator of host B (as destination host) through the ID/Locator mapping system. After obtaining that information, host A will tunnel the packets with destination address being host B to one of its local LDBRs. The LDBR shall find the next-hop LDBR based on the IPv6 globally routable locator, and forward the packets to it. For the intermediate transit networks, if the Non-LDBR routers which the packets have to traverse are legacy IPv4 routers, the ingress LDBR (for that locator domain) forwards the packet to the egress LDBR of the same locator domain over IPv4 tunnels. +-------------+ +-------------+ +-------------+ +-------------+ | Payload | | Payload | | Payload | | Payload | +-------------+ +-------------+ +-------------+ +-------------+ |HI(A)->HI(B) | |HI(A)->HI(B) | |HI(A)->HI(B) | |HI(A)->HI(B) | +-------------+ +-------------+ +-------------+ +-------------+ |IPv6->IPv6 | |IPv6->IPv6 | |IPv6->IPv6 | |IPv6->IPv6 | | (A) (B) | | (A) (B) | | (A) (B) | | (A) (B) | +-------------+ +-------------+ +-------------+ +-------------+ |IPv4->IPv4 | | IPv4->IPv4 | |IPv4->IPv4 | | (A) (BR1) | |(BR2) (BR3) | | (BR4) (B) | +-------------+ +-------------+ +-------------+ |<- A to BR1 ->|<-BR1 to BR2 ->|<-BR2 to BR3 ->| |<-BR4 to B ->| +--------- ------ ---------| +---+ \ / \ / +---+ | A | \ / \ / /| B | +---+\\ \ / \ / // +---+ | \\ | | | | / | | \\ +---+ +---+ +---+ +---+// | | \|BR1+----+BR2+------+BR3+---+BR4+/ | | +---+ +---+ +---+ +---+ | | | | | | | \ / \ / \ / \ LD #1 / \ LD #2 / \ LD #3 / \ / \ / \ / ------ ------ ------ Figure 3. Routing Procedure Xu Expires July 25, 2010 [Page 9] Internet-Draft Routing Architecture July, 2009 for the Next Generation Internet (RANGI) In RANGI, IPv6-over-IPv4 tunnels are employed for intra-LD routing between LDBRs (or between LDBRs and hosts). Hence, RANGI can achieve a smooth IPv4/IPv6 transition. Once the internal routers within LDs are upgraded to IPv6, the requirement of IPv6 over IPv4 tunnel between the LDBRs or between LDBRs and hosts will be eliminated and the packets will be delivered by the LDBRs and the internal IPv6 routers hop by hop without tunneling as shown in Figure 4. +-------------------+ +-------------------+ +-------------------+ | Payload | | Payload | | Payload | +-------------------| +-------------------| +-------------------| | HI(A)->HI(B) | | HI(A)->HI(B) | | HI(A)->HI(B) | +-------------------| +-------------------| +-------------------| | IPv6(A)->IPv6(B) | | IPv6(A)->IPv6(B) | | IPv6(A)->IPv6(B) | +-------------------| ++------------------| +-------------------| |IPv4(A)->IPv4(BR1) | +-------------------+ |<--- A to BR1 --->|<----- BR1 to BR4 --->|<--- BR4 to B --->| +--------- ------ ---------| +---+ \ / \ / +---+ | A | \ / \ / /| B | +---+\\ \ / \ / // +---+ | \\ | | | | / | | \\ +---+ +---+ +---+ +---+// | | \|BR1+----+BR2+------+BR3+---+BR4+/ | | +---+ +---+ +---+ +---+ | | | | | | | \ LD #1 / \ LD #2 / \ LD #3 / \ (IPv4) / \ (IPv6) / \ (IPv6) / \ / \ / \ / ------ ------ ------ Figure 4. IPv4/IPv6 Transition 2.6. Multi-homing and Traffic-Engineering In RANGI, Each multi-homed stub LD shall be assigned a LD ID by each upstream ISP. In fact, these LD IDs are /96 IPv6 prefixes which are topologically aggregatable in provider networks. Each Host within the multi-homed site, in turn, has multiple locators by concatenating the provider-assigned LD IDs with its locally unique IPv4 address. These hosts register the mappings from their identifiers to locators with the ID/Locator mapping system. As shown in Figure 5, host A which is located in a multi-homed site, has two LD IDs, LD ID_1 and LD ID_2, assigned separately from ISP1 and ISP2. Host A chooses either one as Xu Expires July 25, 2010 [Page 10] Internet-Draft Routing Architecture July, 2009 for the Next Generation Internet (RANGI) the source LD ID of the outgoing packets. Upon receiving the packets, the site exit LDBR, BR1, implements source-based policy routing. For example, if the source LD is LD ID_1, the packets will be forwarded to the ISP1's network, otherwise, they will be forwarded to ISP2's network. ------ / \ / \ / \ | | +---+ | +BR2| | /+---+ | / | | / \ / /------ / \ ISP#1 / +---* \ / \ / | A | \ / ------ +---+\\ \ / | \\ | / | \\ +---+ / | \|BR1+/ ------ | +---+-- / \ | | -- / \ \ / -- / \ \ Site A / -- | | \ / -- +---+ | ------ --+BR3| | +---+ | | | \ / \ ISP#2 / \ / ------ Figure 5. Site Multi-homing and Traffic-engineering The site-controlled traffic-engineering works as follows: 1) The source host sends out packets with the source LD ID being one of its LD IDs (assuming LD ID 1 being used). 2) The packets are intercepted by the LDBR BR1, and according to the traffic-engineering policy, the source LD IDs of the packets may Xu Expires July 25, 2010 [Page 11] Internet-Draft Routing Architecture July, 2009 for the Next Generation Internet (RANGI) be re-written from LD ID_1 to LD ID_2. Then BR1 forwards the packets into ISP2's network according to source-based routing policies. 3) Once the packets arrive at the destination host, that host will use the source LD IDs in the received packets as the destination LD IDs in the response packets. So the response packets will also enter site A through ISP2's network. 4) The source host could accept this change and use LD ID 2 as source LD ID in the subsequent packets. Similar to the GSE [GSE], the site-controlled traffic-engineering by rewriting the source LD ID will have effect on the path (upstream ISP) selection for both the outgoing packets and the incoming packets. In addition, the multi-homing and traffic-engineering usages in RANGI will not cause any routing scalability issue. 2.7. Mobility In RANGI, when a host physically moves from one attachment to another, its host ID remains unchanged. The host needs to register the new locator with the ID/locator mapping system. Meanwhile, it should notify the corresponding entity of its new locator as soon as possible. 3. Summary RANGI achieves almost all of goals set by RRG, which are listed as follows: 1) Routing Scalability: Scalability is achieved by separating identifiers from locators. Global routing is done based on provider assigned locators. 2) Traffic Engineering: Site border router can overwrite the source locator of the outgoing packets before performing source-based policy routing. That is to say, hosts located in a multi-homed site can suggest the upstream ISP for outbound and inbound traffics, while the first-hop LDBR (i.e., site border router) has the final decision right on the upstream ISP selection. 3) Mobility and Multi-homing: Applications and transport layers are bound to host IDs and so the sessions will not be interrupted due to locator change in cases of mobility or multi-homing. Xu Expires July 25, 2010 [Page 12] Internet-Draft Routing Architecture July, 2009 for the Next Generation Internet (RANGI) 4) Simplified Renumbering: When changing providers, the local IPv4 addresses of the site do not need to change. Hence the internal routers within the site don't need renumbering. 5) Decoupling Location and Identifier: Obvious. 6) Routing Quality: Since LDBRs only exchange LD reachability and the topology within LD will not be disclosed outside, the routing stability is improved significantly. 7) Routing Security: RANGI reuses existing routing system and does not introduce any new security risk into the routing system. 8) Incremental Deployability: the non-LDBRs within a LD can still be IPv4-only. RANGI allows easy transition from IPv4 networks to IPv6 networks. In addition, the transition mechanisms for RANGI defined in [RANGI-PROXY] allow RANGI hosts to communicate with the legacy IPv4 or IPv6 hosts, and vice versa. 4. Security Considerations TBD. 5. IANA Considerations A specific prefix for host IDs needs to be assigned from the IPv6 address space. Two new options in the Destination Option Header need to be assigned for the host ID and its corresponding parameter date structure respectively. 6. Acknowledgments The author would like to thank Raj Jain, Xuewei Wang and Dacheng Zhang for their valuable contributions. Thanks should also be given to Paul Francis, Lixia Zhang, Brain Carpenter, Dave Oran, Joel Halpern, and Tony Li for their insightful comments. This research project is partially funded by the National''863'' Hi- Tech Program of China. Xu Expires July 25, 2010 [Page 13] Internet-Draft Routing Architecture July, 2009 for the Next Generation Internet (RANGI) 7. References [RAWS] D. Meyer, L. Zhang, and K. Fall. "Report from the IAB Workshop on Routing and Addressing". Internet draft, draft-iab-raws- report-01.txt, work in progress, February 2007. [GOALS] T. Li, "Design Goals for Scalable Internet Routing", draft- irtf-rrg-design-goals-01, July 2007. [RFC4423] R. Moskowitz and P. Nikander, "Host Identity Protocol (HIP) Architecture", RFC 4423, May 2006. [RFC3972] T. Aura, "Cryptographically Generated Addresses (CGA)", RFC3972, Mar 2005. [RFC5214] F. Templin, T. Gleeson, ''Intra-Site Automatic Tunnel Addressing Protocol (ISATAP),'' RFC 5214, March, 2008. [RFC2136] P. Vixie, S. Thomson, Y. Rekhter, J. Bound, ''Dynamic Updates in the Domain Name System (DNS UPDATE)'', RFC 2136, April 1997. [RFC2065] Eastlake, D., and C. Kaufman, "Domain Name System Protocol Security Extensions", RFC 2065, January 1997. [RFC2137] Eastlake, D., "Secure Domain Name System Dynamic Update", RFC 2137, April 1997. [H-DHT] L. Garces-Erice, E. Biersack, P. Felber, K. Ross, and G. Urvoy-Keller, "Hierarchical Peer-to-peer Systems" In Proc. Euro-Par 2003, Klagenfurt,Austria, 2003. [GSE] M. O'Dell, "GSE-An Alternative Addressing Architecture for IPv6", Internet-Draft, Feb 1997. [LNA] Hari Balakrishnan, Karthik Lakshminarayanan, Sylvia Ratnasamy,Scott Shenker, Ion Stoica and Michael Walfish, "A Layered Naming Architecture for the Internet", Proc. ACM SIGCOMM, Portland, Oregon, USA, August 30 - September 3, 2004. [RANGI-PROXY] X. Xu, ''Transition Mechanisms for Routing Architecture for the Next Generation Internet (RANGI)'', draft-xu-rangi- proxy-01.txt, July 2009. Xu Expires July 25, 2010 [Page 14] Internet-Draft Routing Architecture July, 2009 for the Next Generation Internet (RANGI) Authors' Addresses Xiaohu Xu Huawei Technologies, No.3 Xinxi Rd., Shang-Di Information Industry Base, Hai-Dian District, Beijing 100085, P.R. China Phone: +86 10 82836073 Email: xuxh@huawei.com