IPv6 Source/Destination Routing using IS-IS

This specification builds on IS-IS for IPv6 and the critical extension TLV in . This note defines the sub-TLV for an IPv6 Source Prefix, to define routes from a source prefix to a destination prefix. This implements the Destination/Source Routing mechanism described in . This implies not simply routing "to a destination", but routing "to that destination AND from a specified source". It may be combined with other qualifying attributes, such as "traffic going to that destination AND using a specified flow label AND from a specified source prefix". The obvious application is egress routing, as required for a multihomed entity with a provider-allocated prefix from each of several upstream networks. Traffic within the network could be source/destination routed as well, or could be implicitly or explicitly routed from "any prefix", ::/0. Other use cases are described in . If a FIB contains a route to a given destination from one or more prefixes not including ::/0, and a given packet destined there that has a source address that is in none of them, the packet in effect has no route, just as if the destination itself were not in the route table.

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in .

Both IS-IS and OSPF perform their calculations by building a lattice of routers and links from the router performing the calculation to each router, and then use routes (sequences in the lattice) to get to destinations that those routes advertise connectivity to. Following the SPF algorithm, calculation starts by selecting a starting point (typically the router doing the calculation), and successively adding {link, router} pairs until one has calculated a route to every router in the network. As each router is added, including the original router, destinations that it is directly connected to are turned into routes in the route table: "to get to 2001:db8::/32, route traffic to {interface, list of next hop routers}". For immediate neighbors to the originating router, of course, there is no next hop router; traffic is handled locally. In this context, the route is qualified by a source prefix; It is installed into the FIB with the destination prefix, and the FIB applies the route if and only if the IPv6 source address also matches the advertised prefix. Of course, there may be multiple LSPs in the RIB with the same destination and differing source prefixes; these may also have the same or differing next hop lists. The intended forwarding action is to forward matching traffic to one of the next hop routers associated with this destination and source prefix, or to discard non-matching traffic as "destination unreachable". TLVs that lack a source prefix sub-TLV match any source address (i.e., the source prefix TLV defaults to ::/0), by definition. When resolving Destination/Source Reachabilities, the SPF calculation results used MUST reflect a calculation performed including only routers that advertise support for the critical Source Prefix TLV defined in section 3. The mechanism for signaling this is described in . Routers that support this extension MUST advertise support as described there.

For the purposes of this document, a route from the prefix A to the prefix B (in other words, whose source prefix is A and whose destination prefix is B) is expressed as A->B. A packet with the source address A and the destination address B is similarly described as A->B.

In any routing protocol, there is the possibility of ambiguity. For example, one router might advertise a fairly general prefix - a default route, a discard prefix (which consumes all traffic that is not directed to an instantiated subnet), or simply an aggregated prefix while another router advertises a more specific one. In source/destination routing, potentially ambiguous cases include cases in which the link state database contains two routes A->B' and A'->B, in which A' is a more specific prefix within the prefix A and B' is a more specific prefix within the prefix B. Traditionally, we have dealt with ambiguous destination routes using a "longest match first" rule. If the same datagram matches more than one destination prefix advertised within an area, we follow the route with the longest matching prefix. With source/destination routes, as noted in , we follow a similar but slightly different rule; the FIB lookup MUST yield the route with the longest matching destination prefix that also matches the source prefix constraint. In the event of a tie on the destination prefix, it MUST also match the longest matching source prefix among those options. An example of the issue is this. Suppose we have two routes: 2001:db8:1::/48 -> 2001:db8:3:3::/642001:db8:2::/48 -> 2001:db8:3::/48 and a packet 2001:db8:2::1 -> 2001:db8:3:3::1 If we require the algorithm to follow the longest destination match without regard to the source, the destination address matches 2001:db8:3:3::/64 (the first route), and the source address doesn't match the constraint of the first route; we therefore have no route. The FIB algorithm, in this example, must therefore match the second route, even though it is not the longest destination match, because it also matches the source address.

In the event that there are other constraints on routing, such as proposed in , the effect is a logical AND. The FIB lookup must yield the route with the longest matching destination prefix that also matches each of the constraints. The general mechanics for this are described in .

While not mandatory, IS-IS is often implemented as Multi Topology Routing with IPv4 or other protocols in the same or different topologies. The TLV structure in is topology-agnostic in that it always includes the topology ID, which may be zero to indicate the default topology. The mechanism in this document and its Sub-TLV are applicable to any topology that carries routing information used for IPv6 Unicast routing. Destination/Source reachability information SHOULD NOT be placed differently from "plain" destination reachabilities. A system MUST NOT originate Destination/Source Reachabilities in a topology that is exclusively configured for multicast RPF operation. If a topology is shared between unicast lookups and multicast reverse path lookups, reachabilities with a source prefix other than ::/0 MUST be ignored for multicast reverse path lookups. The statements in the previous two paragraphs currently result in applicability of Destination/Source routes as:

MT-ID designated usage applicability 0 default topology yes 1 IPv4 management no 2 IPv6 default yes 3 IPv4 multicast no 4 IPv6 multicast no 5 IPv6 management yes