Network Working Group F. Templin, Ed. Internet-Draft Boeing Phantom Works Intended status: Informational January 19, 2009 Expires: July 23, 2009 Routing and Addressing in Next-Generation EnteRprises (RANGER) draft-templin-ranger-06.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 23, 2009. 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. Abstract Next-generation enterprise networks including the global Internet itself will require an architected solution for the coordination of Templin Expires July 23, 2009 [Page 1] Internet-Draft RANGER January 2009 internal routing and addressing plans and with accommodations for Provider-Independent (PI) addressing, scalability, mobility, multi- homing and security. Additionally, next-generation networks will require support for both Internet protocol versions (IPv4 and IPv6) for an indeterminant period; perhaps even indefinitely. These considerations are particularly true for existing deployments, but the same principles apply even for clean-slate approaches. The RANGER architecture addresses these requirements, and provides a comprehensive framework for IPv6/IPv4 coexistence. Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 4 3. The RANGER Architecture . . . . . . . . . . . . . . . . . . . 6 3.1. The Enterprise-within-Enterprise Framework . . . . . . . . 6 3.2. Virtual Enterprise Traversal (VET) . . . . . . . . . . . . 8 3.2.1. Organizational Principles . . . . . . . . . . . . . . 9 3.2.2. Dynamic Routing and On-Demand Mapping . . . . . . . . 11 3.2.3. Support for Legacy Services . . . . . . . . . . . . . 13 3.3. Subnetwork Encapsulation and Adaptation Layer (SEAL) . . . 14 3.4. Mobility Management . . . . . . . . . . . . . . . . . . . 15 3.5. Multihoming . . . . . . . . . . . . . . . . . . . . . . . 16 4. Related Initiatives . . . . . . . . . . . . . . . . . . . . . 17 4.1. 6over4 and ISATAP . . . . . . . . . . . . . . . . . . . . 17 4.2. The Locator Identifier Split Protocol (LISP) . . . . . . . 17 5. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 18 6. Security Considerations . . . . . . . . . . . . . . . . . . . 18 7. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 19 8. References . . . . . . . . . . . . . . . . . . . . . . . . . . 19 8.1. Normative References . . . . . . . . . . . . . . . . . . . 19 8.2. Informative References . . . . . . . . . . . . . . . . . . 19 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . . 21 Templin Expires July 23, 2009 [Page 2] Internet-Draft RANGER January 2009 1. Introduction Next-generation enterprise networks including the global Internet itself will require an architected solution for the coordination of internal routing and addressing plans with accommodations for Provider-Independent (PI) addressing, routing scalability, mobility, multi-homing and security. Additionally, next-generation networks will require support for both Internet protocol versions (IPv4 and IPv6) for an indeterminant period; perhaps even indefinitely. These considerations are particularly true for existing deployments, but the same principles apply even for clean-slate approaches. The RANGER architecture addresses these requirements, and provides a comprehensive framework for IPv6/IPv4 coexistence [I-D.arkko-townsley-coexistence]. RANGER is a scalable architecture for routing and addressing in next- generation enterprises that contain one or more distinct interior addressing domains. Each of these domains may coordinate their own internal addressing plans independently of one another such that limited-scope addresses (e.g., [RFC1918] private-use IPv4 addresses) may be reused with impunity to provide unlimited scaling through spatial reuse. Each addressing domain therefore appears as an enterprise unto itself, such that a model of recursively nested "enterprises-within-enterprises" is enabled. Logical or physical partitioning of an enterprise into multiple sites is also possible and beneficial in many scenarios. Without an architected approach, routing and addressing within such a framework would be fragmented due to limited-scope address/prefix reuse between disjoint addressing domains. With RANGER, however, the enterprise can be unified via a virtual overlay architecture mainfested by automatic tunneling over disjoint domains interconnected via border routers. The RANGER architecture provides for operation of virtual overlay networks within a diverse range of enterprise network scenarios. Moreover, RANGER gracefully supports modes of operation that have heretofore challenged the classic IP networking model. While this document discusses the specific example of IPv6 as a virtual overlay over or IPv4 networks, it is important to note that the same architectural principles apply to any combination of IP* virtual overlays over IP* networks. The RANGER architecture builds on mechanisms documented in the IRTF and IETF communities, with composite technologies including Virtual Enterprise Traversal (VET) [I-D.templin-autoconf-dhcp], the Subnetwork Encapsulation and Adaptation Layer (SEAL) [I-D.templin-seal] and the Intra-Site Automatic Tunnel Addressing Templin Expires July 23, 2009 [Page 3] Internet-Draft RANGER January 2009 Protocol (ISATAP) [RFC5214][I-D.templin-isatapv4]. Other mechanisms such as IPsec [RFC4301] are also in scope for use within certain deployments. The RANGER architectural principles can be traced to the deliberations of the ROAD group in January 1992 [RFC1380], and also to still earlier works including NIMROD [RFC1753], the Catenet model for internetworking [CATENET][IEN48] [RFC2775], and many others. [RFC1955] captures the high-level architectural aspects of the ROAD group deliberations in a "New Scheme for Internet Routing and Addressing [ENCAPS] for IPNG". 2. Terminology commons a routing region within an enterprise that provides a subnetwork for cooperative peering between the border routers of diverse organizations that may have competing interests. A prime example of a commons is the Default Free Zone (DFZ) of the global Internet. enterprise the same as defined in [RFC4852], where the enterprise deploys a unified routing and addressing plan within the commons, but may contain many disjoint interior addressing domains and/or organizational groupings that can be considered as enterprises unto themselves. An enterprise therefore need not be "one big happy family", but instead provides a commons for the cooperative interconnection of diverse organizations that may have competing interests (e.g., such as the case within the global Internet default free zone). Enterprise networks are typically associated with large corporations or academic campuses, however the RANGER architectural principles apply to any network that has some degree of cooperative active management. This definition therefore extends to home networks, small office networks, a wide variety of mobile ad-hoc networks (MANETs), and even to the global Internet itself. site a logical and/or physical grouping of interfaces within a unified addressing region of an enterprise, where the topology of the site is a proper subset of the topology of the enterprise. A site may contain many interior sites, which may themselves contain many interior sites in a recursive fashion. Templin Expires July 23, 2009 [Page 4] Internet-Draft RANGER January 2009 Throughout the remainder of this document, the term "enterprise" refers to either enterprise or site, i.e., the RANGER principles apply equally to enterprises and sites of any size or shape. At the lowest level of recursive decomposition, a singleton Border Router can be considered as an enterprise unto itself. Border Router (BR) a node at the edge of an enterprise that is also configured as a router in an overlay network. BRs connect their directly-attached networks to the overlay network, and connect to other networks via IP-in-IP tunneling across the commons to other BRs. Border Gateway (BG) a BR that also connects the enterprise to provider networks and/or to the global Internet. BGs are typically configured as default routers in the overlay, and provide forwarding services for accessing IP networks not reachable via a BR within the commons. overlay network a virtual network manifested by routing and addressing over virtual links formed through automatic tunneling. An overlay network may span many underlying enterprises. 6over4 Transmission of IPv6 over IPv4 Domains without Explicit Tunnels [RFC2529]; functional specifications and operational practices for automatic tunneling of unicast/multicast IPv6 packets over multicast-capable IPv4 enterprises. ISATAP Intra-Site Automatic Tunnel Addressing Protocol (ISATAP) [RFC5214][I-D.templin-isatapv4]; functional specifications and operational practices for automatic tunneling over unicast-only enterprises. VET Virtual Enterprise Traversal (VET) [I-D.templin-autoconf-dhcp]; functional specifications and operational practices that provide a functional superset of 6over4 and ISATAP. In addition to both unicast and multicast tunneling, VET also supports address/prefix autoconfiguration as well as additional encapsulations such as IPSec, SEAL, LISP, etc. SEAL Subnetwork Encapsulation and Adaptation Layer (SEAL) [I-D.templin-seal]; a functional specification for robust packet identification and link adaptation over tunnels. SEAL supports effective ingress filtering and adapts to subnetworks configured Templin Expires July 23, 2009 [Page 5] Internet-Draft RANGER January 2009 over links with diverse characteristics. Provider-Independent (PI) prefix an IPv6 or IPv4 prefix (e.g., 2001:DB8::/48, 192.0.2/24, etc.) that is routable within a limited scope and may also appear in enterprise mapping tables. PI prefixes that can appear in mapping tables are typically delegated to a BR by a registry, but are not aggregated by a provider network. Local-use IPv6 and IPv4 prefixes (e.g., FD00::/8, 192.168/16, etc.) are another example of a PI prefix, but these typically do not appear in mapping tables. Provider-Aggregated (PA) prefix an IPv6 or IPv4 prefix that is either derived from a PI prefix or delegated directly to a provider network by a registry. Although not widely discussed, it bears specific mention that an IPv6 prefix taken from delegating router's PI space becomes a PA prefix from the perspective of the requesting router. 3. The RANGER Architecture The RANGER architecture enables scalable routing and addressing in next-generation enterprise networks with sustaining support for legacy networks and services. Key to this approach is a framework that accommodates interconnection of diverse organizations within the enterprise with a mutual spirit of cooperation, but with the potential for competing interests. The following sections outline the RANGER architecture within the context of anticipated use cases: 3.1. The Enterprise-within-Enterprise Framework Enterprise networks traditionally distribute routing information via Interior Gateway Protocols (IGPs) such as Open Shortest Path First (OSPF), while large enterprises may even use an Exterior Gateway Protocol (EGP) internally in place of an IGP. In particular, it is becoming increasingly commonplace for large enterprises to use the Border Gateway Protocol (BGP) internally and independently from the BGP instance that maintains the routing information base within the global Internet Default Free Zone (DFZ). As such, large enterprises may run an internal instance of BGP across many internal Autonomous Systems (ASs). Such a large enterprise can therefore appear as an Internet unto itself, albeit with default routes leading to the true global Internet. Additionally, each internal AS within such an enterprise may itself run BGP internally in place of an IGP, and can therefore also appear as an independent lower-tier enterprise unto itself. This enterprise-within-enterprise framework can be extended in a hierarchical fashion as broadly and as Templin Expires July 23, 2009 [Page 6] Internet-Draft RANGER January 2009 deeply as desired to acheive scaling factors as well as organizational and/or functional compartmentalization, e.g., as shown in Figure 1. ,---------------. ,-' Global `-. <--------+ ( IPv6/IPv4 ) ,----|-----. `-. Internet ,-' ( Enterprises) `+--+..+--+ ...+--+ ( E2 thru EN ) _.-|R1|--|R2+----|Rn|-._ `.---------/ _.---'' +--+ +--+ ...+--+ -. ,--'' ,---. `---. ,-' X5 X6 .---.. `-. ,' ,.X1-.. / \ ,' `. `. ,' ,' `. .' E1.2 '. X8 E1.m \ `. / / \ | ,--. | / _,.._ \ \ / / E1.1 \ | Y3 `. | | / Y7 | \ ; | ___ | | ` W Y4 |... | `Y6 ,' | : | | ,-' `. X2 | `--' | | `'' | | : | | V Y2 | \ _ / | | ; \ | `-Y1,,' | \ .' Y5 / \ ,-Y8'`- / / \ \ / \ \_' / X9 `. ,'/ / `. \ X3 `.__,,' `._ Y9'',' ,' ` `._ _,' ___.......X7_ `---' ,' ` `---' ,-' `-. -' `---. `. E1.3 Z _' _.--' `-----. \---.......---' _.---'' `----------------'' <---------------- Enterprise E1 ----------------> Figure 1: Enterprise-within-Enterprise Framework Figure 1 depicts an enterprise 'E1' connected to the global IPv6/IPv4 Internet via routers 'R1' through 'Rn' and additional enterprises 'E2' through 'EN' that also connect to the global Internet. Within the 'E1' commons, there may be arbitrarily-many hosts, routers and networks (not shown in the diagram) over which both unencapsulated IP packets and IP-in-IP encapsulated packets can be forwarded. There may also be many lower-tier enterprises 'E1.1' through 'E1.m' (shown in the diagram) that interconnect within the 'E1' commons via Border Routers (BRs) depicted as 'X1' through 'X9' (where 'X1' through 'X9' see 'R1' through 'Rn' as Border Gateways (BGs)). Within each 'E1.*' enterprise, there may also be arbitrarily-many lower-tier enterprises that interconnect within the 'E1.*' commons via BRs depicted as 'Y1' through 'Y9' in the diagram (where 'Y1' through 'Y9' see 'X1' through 'X9' as BGs). This hierarchical decomposition can be recursively nested as deeply as desired, and ultimately terminates at singleton nodes such as those depicted as 'V', 'W' and 'Z' in the diagram. Templin Expires July 23, 2009 [Page 7] Internet-Draft RANGER January 2009 It is important to note that nodes such as 'V', 'W' and 'Z' may be simple hosts, or they may be BRs that attach arbitrarily-complex edge networks. Such edge networks could be as simple as a home network behind a residential gateway or as complex as a major corporate/ academic campus, a large service provider network, etc. Again, this enterprise-within-enterprise framework can be recursively nested as broadly and deeply as desired. From the ultimate level of the hierarchy, consider now that the global Internet itself can be viewed as an "enterprise" that interconnects lower-tier enterprises E1 through EN such that all RANGER architectural principles apply equally within that context. As a specific case in point, the future global Aeronautical Telecommuncations Network (ATN) under consideration within the civil aviation industry [I-D.bauer-mext-aero-topology] will take the form of a large enterprise network that appears as an Internet unto itself, i.e., exactly as depicted for 'E1' in Figure 1. Within the ATN, there will be many Air Communications Service Provider (ACSP) and Air Navigation Service Provider (ANSP) networks organized as autonomous systems internal to the ATN, i.e., exactly as depicted for 'E1.*' in the diagram. The ACSP/ANSP network BGs will participate in a BGP instance internal to the ATN, and may themselves run independent BGP instances internally that are further sub-divided into lower-tier enterprises organized as regional, organizational, functional, etc. compartments. It is important to note that, while ACSPs/ANSPs within the ATN will share a common objective of safety- of-flight for civil aviation services, there may be competing business/social/political interests between them such that the ATN is not necessarily "one big happy family". Therefore, the model parallels that of the global Internet itself. Such an operational framework may indeed be the case for many next- generation enterprises. In particular, although the routing and addressing arrangements of all enterprises will require a mutual level of cooperative active management at a certain level, free market forces, business objectives, political alliances, etc. may drive internal competition. 3.2. Virtual Enterprise Traversal (VET) Within the enterprise-within-enterprise framework outlined in Section 3.1, the RANGER architecture is based on an overlay network approach manifested through Virtual Enterprise Traversal (VET) [I-D.templin-autoconf-dhcp] [RFC5214][I-D.templin-isatapv4]. The approach uses automatic IP-in-IP tunneling within a hierarchy of lower-tier enterprises that are either configured within the same addressing region of a parent enterprise or use their own enterprise- Templin Expires July 23, 2009 [Page 8] Internet-Draft RANGER January 2009 local routing and addressing internally. VET and its related works specify the necessary mechanisms and operational practices to manifest the RANGER architecture. The use of VET in conjunction with SEAL Section 3.3 is essential in certain deployments to avoid issues related to source address spoofing and black holing due to path Maximum Transmission Unit (MTU) limitations. (The use of VET in conjunction with IPsec [RFC4301] can also be beneficial in some enterprise network scenarios.) The following sections discuss operational considerations and use cases within the VET approach: 3.2.1. Organizational Principles In the VET approach, logically and/or physically disjoint lower-tier enterprises are connected by BGs that appear as BRs within a parent enterprise. In turn, BRs within the parent enterprise coordinate their IP prefixes with BGs that connect to upper-tier enterprises. The prefixes could be either Provider-Independent (PI) prefixes owned by the BR or Provider-Aggregated (PA) prefixes delegated by the BG, but the prefixes are linked with mapping and routing over the virtual topology in both cases. Additionally, fault tolerance and multihoming is naturally afforded through configuration of multiple BGs per enterprise. Figure 2 below depits a vertical slice (albeit represented horizontally) from the enterprise-within-enterprise framework shown in Figure 1. In this example, an IPv6 host 'H' that is deeply nested within Enterprise 'E1' connects to IPv6 server 'S1' located somewhere on the IPv6 Internet. Again, while this and other examples speak of the specific instance of IPv6-in-IPv4 encapsulation, the same principles apply to any IP-in-IP encapsulation: Templin Expires July 23, 2009 [Page 9] Internet-Draft RANGER January 2009 +------+ | IPv6 | |Server| " " " " " " " "" " " " " " " " " " " " " " " " | S1 | " <----------------- 2001:DB8::/40 (PA) " +--+---+ " 2001:DB8:10::/56 (PI) ----------------> " | " . . . . . . . . . . . . . . . " | " . . . . . . " | " . +----+ v +--- + v +----+ v +----+ +-----+-------+ " . | V += e =+ Y1 += e =+ X2 += e =+ R2 +==+ Internet | " . +-+--+ t +----+ t +----+ t +----+ +-----+-------+ " . | 1 . . 2 . . 3 . " | " . H . . . . . " | " . . . . . . . . . . . . . . " +--+---+ " " | IPv4 | " 10/8 10/8 10/8 " |Server| " " " " " " " " " " " " " " "" " " " " " " " | S2 | <-- Enterprise E1 --> +------+ Figure 2: Virutal Enterprise Traversal Within this vertical slice, Figure 2 depicts each enterprise within the 'E1' hierarchy as spanned by automatic IPv6-in-IPv4 tunnels 'vet1' through 'vet3'. Each 'vet*' interface within this framework is a Non-Broadcast, Multiple Access (NBMA) interface that connects all BRs within the same enterprise. Each enterprise within the 'E1' hierarchy may comprise a smaller topological region within a larger IPv4 routing region, or they may configure an independent routing and addressing plan from a common (but spatially reused) limited-scope IPv4 prefix, e.g., depicted as '10/8' in the diagram. When PA addressing is used, the BR for each 'E1*' enterprise receives an IPv6 prefix delegation from a delegating BG in a parent enterprise. In this example, when 'R2' is a delegating router for the prefix '2001:DB8::/40, it may delegate '2001:DB8::/48' to X2, whichin turn delegates '2001:DB8::/52' to Y1, which in turn delegates 2001:DB8::/56' to V. When PI addressing is used, individual BRs (e.g., V) register their PI prefixes (e.g., 2001:DB1:10::/56) with their BGs which in turn register them with their parent BGs, etc. This registration results in updates to the mapping system, which can later be used to populate a BR's Forwarding Information Base (FIB). In the example in Figure 2, a simple IPv6 host 'H' is attached to a shared link with IPv6/IPv4 dual stack router 'V', which advertises the IPv6 prefixes '2001:DB8:0:0::/64' and '2001:DB8:10:0::/64. IPv6 host 'H' uses standard IPv6 neighbor discovery mechanisms to discover 'V' as a default IPv6 router that can forward its packets off the local link, and configures addresses from both of the advertised Templin Expires July 23, 2009 [Page 10] Internet-Draft RANGER January 2009 prefixes. 'V' in turn sees node 'Y1' as a BG that can be reached via VET interface ''vet1' and that can forward packets toward IPv6 server 'S1'. Similarly, node 'Y1' is a BR on the enterprise spanned by 'vet2' that sees 'X2' as a BG, and node 'X2' is a BR on 'vet3' that sees 'R2' as a BG. Ultimately, 'R2' is a BR that connects 'E1' to the global Internet. In this and other examples, it is specifically worth noting that a BR may have potentially many VET interfaces over which it can connect to the BRs/BGs of neighboring enterprises. 3.2.2. Dynamic Routing and On-Demand Mapping In the example shown in Figure 2, 'V', 'Y1', 'X2' and 'R2' configure separate 'vet*' interfaces for each enterprise they connect to and to discover BRs/BGs through a dynamic routing protocol and/or mapping database lookups. After tunnels 'vet1' through 'vet3' are established and BG's discovered, the BRs connected to a 'vet*' interface can run a dynamic routing protocol such as OSPVFv3 [RFC5340] and form adjacencies between themselves in an on-demand fashion while treating the 'vet*' interface as an ordinary link. It is important to note that adjacencies can be formed on-demand and allowed to expire after idle periods such that a full mesh of links need not be maintained. This allows an overlay network that spans 'E1' to dynamically adapt to changing conditions within the enterprise. In a second example, Figure 3 depicts an instance of on-demand discovery of more-specific routes in which an IPv6 host 'H' connects to an IPv6 server 'J' located in a different organizational entity within 'E1': Templin Expires July 23, 2009 [Page 11] Internet-Draft RANGER January 2009 +------+ | IPv6 | |Server| " " " " " " " "" " " " " " " " " " " " " " " " | S1 | " <----------------- 2001:DB8::/40 (PA) " +--+---+ " 2001:DB8:10::/56 (PI) ----------------> " | " . . . . . . . . . . . . . . . " | " . . . . . . " | " . +----+ v +----+ v +----+ +----+ +-----+-------+ " . | V += e =+ Y1 += e =+ X2 += =+ R2 +==+ Internet | " . +-+--+ t +----+ t +----+ +----+ +-----+-------+ " . | 1 . . 2 . . . " | " . H . . . . v . " | " . . . . . . . . . . . e . " +--+---+ " . t . " | IPv4 | " . . . . . . , . 3 . " |Server| " . +----+ v +----+ . " | S2 | " . | Z += e =+ X7 += . " +------+ " . +-+--+ t +----+ . " " . | 4 . . . " " . J . . . . . " " . . . . . . . " " 2001:DB8:20::/56 (PI) --------> " " " " " " " " " " " " " " " "" " " " " " " " <-- Enterprise E1 --> Figure 3: On-Demand Discovery In this example, tunnel interfaces 'vet1' through 'vet4' as well as IPv6 PI prefix registrations have been established through VET enterprise autoconfiguration procedures [I-D.templin-autoconf-dhcp]. When host 'H' with IPv6 address '2001:DB8:10::1' sends packets to server 'J' with IPv6 address '2001:DB8:20::1', unless IPv6 routing information is available BR 'X2' must determine that 'J' can be reached using a more direct route via 'X7' across the 'E1' commons. To do so, 'X2' can perform an on-demand mapping lookup by consulting the enterprise mapping service (e.g., an enterprise name service). Alternatively, 'X2' can send the packet to a default router 'R2', and 'R2' can return an ICMPv6 redirect message indicating that 'J' can be reached via a more direct route through 'X7'. When 'X2' receives a redirect from 'R2', it can send an IPv6 Router Advertisement (RA) message using SEcure Neighbor Discovery (SEND) [RFC3971] to 'X7' then forward subsequent packets directly via the route-optimized path to 'X7'. In some enterprise scenarions, dynamic routing and on-demand mapping can be combined as complementary functions. In other scenarios, it may be sufficient to use on-demand mapping alone. Templin Expires July 23, 2009 [Page 12] Internet-Draft RANGER January 2009 3.2.3. Support for Legacy Services While the overlay network that spans 'E1' provides a fully-connected routing and addressing capability for IP overlay services, access to legacy services will still be required for an extended (and possibly indefinite) period. For example, Figure 4 below depicts the applicable IPv4 service access models for the RANGER architecture when IPv6 is used as the overlay: +------+ | IPv6 | |Server| " " " " " " " "" " " " " " " " " " " " " " " " | S1 | " <----------------- 2001:DB8::/40 (PA) " +--+---+ " 2001:DB8:10::/56 (PI) -----------------> " | " . . . . . . . . . . . . . . . " | " . . . . . . " | " . +----+ v +--- + v +----+ v +----+ +-----+-------+ " . | V += e =+ Y1 += e =+ X2 += e =+ R2 +==+ Internet | " . +----+ t +----+ t +----+ t +----+ +-----+-------+ " . 1 . . 2 . . 3 . " | " . K L . . . . M . " | " . . . . . . . . . . . . . . " +--+---+ " " | IPv4 | " " |Server| " " " " " " " " " " " " " " "" " " " " " " " | S2 | <-- Enterprise E1 --> +------+ Figure 4: Legacy IPv4 Service Support In a first instance, an IPv4 client 'K' within enterprise 'E1.1.1' can access IPv4 service 'L' within the same enterprise as-normal and without the need for any encapsulation. Instead, a "mapping" is done through a simple name lookup within the enterprise-local name service deployed in 'E1.1.1', and enterprise-local native IPv4 services are used. In many instances, this may indeed be the preferred service access model even when IPv6 services are widely deployed due to factors such as inability to replace legacy IPv4 applications, IPv6 header overhead avoidance, etc. In a second instance, IPv4 client 'K' can access IPv4 server 'S2' on the global IPv4 Internet in a number of ways. First, if the recursively nested enterprises are all configured within the same IPv4 routing region within E1, 'K' can simply forward its packets toward 'R2' that then acts as an IPv4 Network Address Translator (NAT) and/or an ordinary IPv4 enterprise border router. Secondly, if the recursively nested enterprises are configured within disjoint IPv4 routing regions, all routers 'Y1', 'X2' and 'R2' can provide an Templin Expires July 23, 2009 [Page 13] Internet-Draft RANGER January 2009 IPv4 NAT capability however this approach requires multiple instances of stateful NAT devices on the path which can lead to an overall degree of brittleness and intolerance to routing changes. Instead, 'E1' could deploy a "Carrier-Grade NAT (CGN)" at 'R2' (i.e., at the enterprise border with the global Internet) and E1.1.1 could discover 'Y1' as an IPv4 default router. 'Y1' could then use the "dual-stack- lite" approach in which IPv4-in-IPv6-in-IPv4 tunneling conveys the IPv4 packets from 'K' to the CGN at 'R2', which then decapsulates and translates the inner IPv4 packets before sending them on to 'S2'. Finally, 'K' could be configured as an IPv6-only node and use standard IPv6 routing to reach an IPv6/IPv4 translator located at an IPv6 BR for the enterprise in which 'S2' resides'. The translator would then use IPv6-to-IPv4 translation before sending packets onwards toward 'S2'. These and other alternatives are discussed in [I-D.wing-nat-pt-replacement-comparison]. In a final instance, IPv4 client 'K' can access an IPv4 server 'M' in a different enterprise within E1 as long as both enterprises are configured over the same underlying IPv4 routing region. If the enterprises are configured over disjoint IPv4 routing regions, however, 'K' would still be able to access 'M' using IPv6-only services, or by using IPv4 services if an IPv4 overlay were configured in parallel with the IPv6 overlay [I-D.templin-isatapv4]. 3.3. Subnetwork Encapsulation and Adaptation Layer (SEAL) Whenever the BRs of disjoint enterprises are joined across a commons, mechanisms that rely on ICMP feedback from routers within the network may become brittle or susceptible to spoofing attacks. This is due to the fact that ICMP messages can be lost due to congestion or packet filtering gateways, and that network middleboxes are essentially "anonymous" and may include insufficient information in ICMPs that can be used to authenticate their source. Of even greater concern is the fact that a rogue node from a different enterprise could send spoofed packets of any kind, e.g., for the purpose of mounting denial-of-service and/or traffic amplification attacks targeting underprivileged links. The Subnetwork Encapsulation and Encapsulation Layer (SEAL) provides effective mitigations by only accepting packets from correspondent BRs that can be validated as topologically-correct routers within the commons (i.e., the subnetwork) using the VET Potential Router List (PRL) and ingress filtering [I-D.templin-autoconf-dhcp] in conjunction with the 32-bit SEAL_ID in the pacekt. Moreover, SEAL does not require reliable delivery of all ICMPs, but rather supports continued operation even if some/many ICMPs are lost. Finally, SEAL adapts to subnetworks that contain links with diverse MTUs properties, and can use probing to identifiy links in the path that Templin Expires July 23, 2009 [Page 14] Internet-Draft RANGER January 2009 configure marginal MTUs. The advantages of using SEAL in conjunction with the RANGER enterprise-within-enterprise framework are tangible, and compare favorably with the alternative of deploying an all-IPv6 infrastructure even for clean-slate deployments. This is especially true within enterprises that provide a commons for joining organizational/political/functional entities with a spirit of mutual cooperation but with competing interests or objectives. 3.4. Mobility Management Mobility management use cases must be considered along several different vectors: o nomadic devices may be satisfied by client-only basic connectivity and can tolerate address renumbering events as they move between enterprise network attachment points. o mobile routers with PI prefixes may be satisfied by updates to the mapping system as long as they do not impart unacceptable churn. o mobile routers and end systems with PA addresses/prefixes may require additional supporting mechanisms that can accomodate address/prefix renumbering. Nomadic device mobility is already satisfied by currently deployed technologies. For example, transporting a laptop computer from a wireless access hot spot to a home network LAN would allow the nomadic device to regain client-only basic network connectivity at the expense of address renumbering resulting in terminated communication sessions. Mobile routers that use VET and SEAL can move freely between enterprises as long as they withdraw their PI prefixes from the mapping/routing systems of departed enterprises and inject them into the mapping/routing systems of new enterprises. In many cases, a more localized event mobility may result in no changes to mapping/ routing in parent enterprises. For enterprises that require in-the- network confidentiality, MobIKE [RFC4555] may also be useful within this context. Mobile routers and end systems that move quickly between disparate enterprise edge network attachment points may impart unacceptable mapping/routing churn and packet loss in service networks that cannot easily support a well-structure enterprise-within-enterprise framework. Devices that connect to such networks should use PA addresses/prefixes that can be coordinated via a rendezvous service Templin Expires July 23, 2009 [Page 15] Internet-Draft RANGER January 2009 in a home enterprise when the device moves to a visited enterprise. Mobility management mechanisms such as Mobile IPv6 [RFC3775] and HIP [RFC4423] can be used to maintain a stable identifier for fast moving devices even as they move quickly between visited enterprise edge network attachment points. As a use case in point, consider an aircraft with a mobile router moving between ground station points of attachment. If the ground stations are located within the same enterprise, or within lower-tier sites of the same parent enterprise, it should suffice for the aircraft to announce its PI prefixes at its new point of attachment and withdraw them from the old. This would avoid excessive routing/ mapping system churn, since the prefixes need not be announced/ withdrawn within the parent enterprise, i.e., the churn is isolated to lower layers of the recursive hierarchy. Note also that such movement would not entail an aircraft-wide readdressing event. As a second example, consider a wireless handset moving between service coverage areas maintained by independent providers with peering arrangements. Since the coverage range of terrestrial cellular wireless technologies is limited, mobility events may occur on a much more aggressive timescale than some other examples. In this case, the handset may expect to incur a readdressing event for its access interface at least, and may be obliged to arrange for a rendezvous linkage with its home network. It should specifically be noted that the contingency of mobility management solutions is not necessarily mutually exclusive, and must be considered in relation to specific use cases. The RANGER architecture is therefore naturally inclusive in this regard. 3.5. Multihoming As with mobility management, multi-homing use cases must be considered along multiple vectors. Within an enterprise, BRs can discover multiple BGs and use them in a fault tolerant and load- balancing fashion as long as they register their PI prefixes with each such BG. At the enterprise edge, a true location/identity split approach such as HIP may be necessary for supporting true multihoming across multiple physical links with diverse properties. As a first case in point, consider the enterprise network of a major corporation that obtains services from a number of ISPs. The corporation should be able to register its PI prefixes with all of its ISPs, and use any of the ISPs for its Internet access services. As a second use case, consider an aircraft with a diverse set of wireless links (e.g., VHF, 802.16, directional, SATCOM, etc.). The aircraft should be able to select and utilize the most appropriate Templin Expires July 23, 2009 [Page 16] Internet-Draft RANGER January 2009 link(s) based on phase of flight, and change seamlessly between links as necessary. Other examples include a nomadic laptop with both 802.11 and Ethernet links, a wireless handset with both CDMA wireless and 802.11, etc. As with mobilitiy management, the contintingency of solutions is not necessarily mutually exclusive and can combine to suit use cases within the scope of the RANGER architecture. 4. Related Initiatives 4.1. 6over4 and ISATAP Long before the RANGER architecture and VET/SEAL specifications were published, 6over4 [RFC2529] specified a means for automatic tunneling of unicast/multicast IPv6 packets over multicast-capable IPv4 enterprises, however it was unable to function in enterprises that did not support a full deployment of IPv4 multicast services. To address these shortcomings, ISATAP [RFC5214][I-D.templin-isatapv4] was specified as a unicast-only variant of 6over4 and widely implemented among major software and equipment vendor products. ISATAP provides unicast-only neighbor discovery mechanisms and also adds a means for determining whether a node on an ISATAP interface is authorized to act as an IPv6 router; namely, the Potential Router List (PRL). VET provides a functional superset of the 6over4 and ISATAP mechanisms; VET further combines with SEAL, IPSec, etc. to provide the functional elements of the RANGER architecture. 4.2. The Locator Identifier Split Protocol (LISP) The Locator-Identifier Split Protocol (LISP) [I-D.farinacci-lisp] proposes a map-and-encaps architecture for scalable routing and addressing within the global Internet Default Free Zone (DFZ). Several companion documents (e.g., LISP-ALT, LISP-CONS, LISP-EMACS, LISP-NERD) propose mapping function solutions. A related mapping function solution proposal is found in [I-D.jen-apt]. LISP, and a number of related proposals being discussed in the Routing Research Group, share common properties with the solution proposed here. They may therefore be architecturally consistent with the RANGER architecture. Templin Expires July 23, 2009 [Page 17] Internet-Draft RANGER January 2009 5. IANA Considerations There are no IANA considerations for this document. 6. Security Considerations Communications between endpoints within different sites inside an enterprise are carried across a commons that joins organizational entities with a mutual spirit of cooperation, but between which there may be competing business/sociological/political interests. As a result, mechanisms that rely on feedback from routers on the path may become brittle or susceptible to spoofing attacks. This is due to the fact that IP packets can be lost due to congestion or packet filtering gateways, and that the source addresses of IP packets can be forged. Moreover, IP packets in general can be generated by anonymous attackers, e.g., from a rogue node within a third-party enterprise that has malicious intent toward a victim. Path MTU discovery is an example of a mechanism that relies on ICMP feedback from routers on the path, and as such is susceptible to these issues. For IPv4, a common workaround is to disable Path MTU Discovery and let fragmentation occur in the network if necessary. For IPv6, lack of fragmentation support in the network precludes this option such that the mitigation typically recommended is to discard ICMP messages that contain insufficient information and/or to operate with the minimum IPv6 path MTU. This example extends also to other mechanisms that either rely on or are enhanced by feedback from network devices, however attack vectors based on non-ICMP messages are also subject for concern. The RANGER architecture supports effective mitigations for attacks such as distributed denial-of-service, traffic amplification, etc. In particular, when VET and SEAL are used, enterprise BGs can use the 32-bit identification encoded in the SEAL header as well as ingress filtering to determine if a message has come from a topologically- correct enterprise located across the commons. This allows enterprises to employ effective mitigations at their borders without the requirement for mutual cooperation from other enterprises. When source address spoofing by on-path attackers located within the commons is also subject for concern, additional securing mechanisms such as tunnel-mode IPsec between enterprise BGs can also be used. BRs can obtain PI prefixes through arrangements with a prefix delegation authority. Thereafter, the BR must have a means of proving its ownership when it announces or withdraws the prefixes in enterprise routing systems. This can be accommodated through the use of SEcure Neighbor Discovery (SEND) [RFC3971] as well as a means for Templin Expires July 23, 2009 [Page 18] Internet-Draft RANGER January 2009 confirming prefix ownership, e.g., through name service lookup. The SEND mechanism is also useful for route optimization between lower- tier enterprises across a parent enterprise commons. While the RANGER architecture does not in itself address security considerations, it proposes an architectural framework for functional specifications that do. Security concerns with tunneling along with recommendations that are compatible with the RANGER architecture are found in [I-D.ietf-v6ops-tunnel-security-concerns]. 7. Acknowledgements This work was inspired through the encouragement of the Boeing DC&NT network technology team and through the communications of the IESG. Many individuals have contributed to the architectural principles that form the basis for RANGER over the course of many years. 8. References 8.1. Normative References [RFC0791] Postel, J., "Internet Protocol", STD 5, RFC 791, September 1981. [RFC2460] Deering, S. and R. Hinden, "Internet Protocol, Version 6 (IPv6) Specification", RFC 2460, December 1998. 8.2. Informative References [CATENET] Pouzin, L., "A Proposal for Interconnecting Packet Switching Networks", May 1974. [I-D.arkko-townsley-coexistence] Arkko, J. and M. Townsley, "IPv4 Run-Out and IPv4-IPv6 Co- Existence Scenarios", draft-arkko-townsley-coexistence-00 (work in progress), September 2008. [I-D.bauer-mext-aero-topology] Bauer, C. and S. Ayaz, "ATN Topology Considerations for Aeronautical NEMO RO", draft-bauer-mext-aero-topology-00 (work in progress), July 2008. [I-D.farinacci-lisp] Farinacci, D., Fuller, V., Oran, D., Meyer, D., and S. Brim, "Locator/ID Separation Protocol (LISP)", Templin Expires July 23, 2009 [Page 19] Internet-Draft RANGER January 2009 draft-farinacci-lisp-11 (work in progress), December 2008. [I-D.ietf-v6ops-tunnel-security-concerns] Hoagland, J., Krishnan, S., and D. Thaler, "Security Concerns With IP Tunneling", draft-ietf-v6ops-tunnel-security-concerns-01 (work in progress), October 2008. [I-D.jen-apt] Jen, D., Meisel, M., Massey, D., Wang, L., Zhang, B., and L. Zhang, "APT: A Practical Transit Mapping Service", draft-jen-apt-01 (work in progress), November 2007. [I-D.templin-autoconf-dhcp] Templin, F., "Virtual Enterprise Traversal (VET)", draft-templin-autoconf-dhcp-27 (work in progress), January 2009. [I-D.templin-isatapv4] Templin, F., "Transmission of IPv4 Packets over ISATAP Interfaces", draft-templin-isatapv4-00 (work in progress), December 2008. [I-D.templin-seal] Templin, F., "The Subnetwork Encapsulation and Adaptation Layer (SEAL)", draft-templin-seal-23 (work in progress), August 2008. [I-D.wing-nat-pt-replacement-comparison] Wing, D., Ward, D., and A. Durand, "A Comparison of Proposals to Replace NAT-PT", draft-wing-nat-pt-replacement-comparison-02 (work in progress), September 2008. [IEN48] Cerf, V., "The Catenet Model for Internetworking", July 1978. [RFC1380] Gross, P. and P. Almquist, "IESG Deliberations on Routing and Addressing", RFC 1380, November 1992. [RFC1753] Chiappa, J., "IPng Technical Requirements Of the Nimrod Routing and Addressing Architecture", RFC 1753, December 1994. [RFC1918] Rekhter, Y., Moskowitz, R., Karrenberg, D., Groot, G., and E. Lear, "Address Allocation for Private Internets", BCP 5, RFC 1918, February 1996. Templin Expires July 23, 2009 [Page 20] Internet-Draft RANGER January 2009 [RFC1955] Hinden, R., "New Scheme for Internet Routing and Addressing (ENCAPS) for IPNG", RFC 1955, June 1996. [RFC2529] Carpenter, B. and C. Jung, "Transmission of IPv6 over IPv4 Domains without Explicit Tunnels", RFC 2529, March 1999. [RFC2775] Carpenter, B., "Internet Transparency", RFC 2775, February 2000. [RFC3775] Johnson, D., Perkins, C., and J. Arkko, "Mobility Support in IPv6", RFC 3775, June 2004. [RFC3971] Arkko, J., Kempf, J., Zill, B., and P. Nikander, "SEcure Neighbor Discovery (SEND)", RFC 3971, March 2005. [RFC4301] Kent, S. and K. Seo, "Security Architecture for the Internet Protocol", RFC 4301, December 2005. [RFC4423] Moskowitz, R. and P. Nikander, "Host Identity Protocol (HIP) Architecture", RFC 4423, May 2006. [RFC4555] Eronen, P., "IKEv2 Mobility and Multihoming Protocol (MOBIKE)", RFC 4555, June 2006. [RFC4852] Bound, J., Pouffary, Y., Klynsma, S., Chown, T., and D. Green, "IPv6 Enterprise Network Analysis - IP Layer 3 Focus", RFC 4852, April 2007. [RFC5214] Templin, F., Gleeson, T., and D. Thaler, "Intra-Site Automatic Tunnel Addressing Protocol (ISATAP)", RFC 5214, March 2008. [RFC5340] Coltun, R., Ferguson, D., Moy, J., and A. Lindem, "OSPF for IPv6", RFC 5340, July 2008. Author's Address Fred L. Templin (editor) Boeing Phantom Works P.O. Box 3707 MC 7L-49 Seattle, WA 98124 USA Email: fltemplin@acm.org Templin Expires July 23, 2009 [Page 21]