Clarification of RRO Node-Id Sub-Object

describes the use of the RRO node-id sub-object to find the Merge Point (MP) address for MPLS Fast Reroute in multi-domain networks, where a domain is defined as an Interior Gateway Protocol (IGP) area or an Autonomous System (AS). Finding the MP address using the node-id sub-object for facility backup tunnels is useful to protect inter-domain TE LSPs from a failure of an Area Border Router (ABR) or Autonomous System Border Router (ASBR). However, does not define an RRO stacking order when including the node-id sub-object. Therefore different interpretations of the RRO seem acceptable, some of which if allowed do not permit an accurate parsing of the RRO to find the MP address in a multi-domain environment. provides clear guidelines regarding the order in which sub-objects are pushed on the RRO stack. Such stacking guidelines when pushing a node-id sub-object are required. The lack of a such a guideline can also present interoperability problems between different implementations. This draft specifies a RRO stacking order when including the node-id sub-object in addition to explaining some of the issues in multi-domain environments.

The Terminology used in this document is as defined in . Additionally the following notation is used describe the contents of the Record Route Object: N = Node-id (node-id sub-object) I = Interface-address (IPv4 sub-object) L = Label | = Marker distinguishing information from neighboring nodes < > = Order of RRO information from a single node

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 .

This section discusses the requirements at the PLR for finding the MP address in an inter-domain network in order to create a facility backup tunnel to provide link or node protection. As described in , the bypass tunnel needs to intersect with the primary tunnel at the MP. The MP address can be found by examining the RRO of the primary tunnel to check if any of the addresses is the MP address. The node-id sub-object was introduced to allow the PLR to pick the MP address in inter-domain environments, where the interface addresses in the RRO of the facility backup and primary tunnel cannot be correlated by the PLR as belonging to the same node (MP). This is specifically true when the PLR and MP are not in the same domain. As also describes, the problem of finding the MP address within a single IGP area be easily solved since the entire topology information is available. For example, to provide node protection using facility backup tunnels in the inter-area or inter-AS case to protect from the failure of an ABR or ASBR, the PLR, which is upstream of the failure node must be able to accurately decipher the MP address from the RRO. Hence, the PLR MUST be able to find the MP address by just parsing the received RRO and not requiring the existence of a Traffic Engineering database in order to provide link or node protection. was also written with this goal as is implied in its Introduction section though this requirement is not explicitly stated.

When adding the node-id sub-object in the RRO, section 3 of seems to allow different RRO formats when interpreted along with the RRO definition and usage from . From section 3 of : Add the node-id sub-object in the RRO carried in an RSVP Resv message and, when required, also add another IPv4/IPv6 sub-object to record interface address. It seems like the following RRO formats are acceptable: <N, L>, <N, I, L>, <I, N, L>, <N, L, I>, <I, L, N> Add an IPv4/IPv6 sub-object recording the interface address and, when required, add a node-id sub-object in the RRO. It seems like the following RRO formats are acceptable: <I, L, N> though the order of the sub-objects is not specified. The RRO does not have a defined delimiter using which a node can parse the RRO in the Resv message to find the set of information that has been added by a specific downstream node. As a result, the PLR might not be able to deterministically find the MP address from the received RRO in the Resv message. The following examples illustrate how the PLR might find it difficult to differentiate between two RRO formats just by parsing the received RRO: <N, L> | <I, L> | <I, L> and <N, L, I, L> | <I, L> <I, L, N> | <I, L> and <I, L> | <N, I, L> <N, L, I> | <N, L> and <N, L> | <I, N, L> <N, I, L> | <N, L, I, L> and <N, I, L> | <N, L> | <I, L> Using the traffic engineering database to correlate whether an interface and node address received in the RRO belong to the same node is not possible in inter-area or inter-AS environments.

The following rules need to be followed by an implementation of : If the node-id sub-object is included in the RRO, then the IPv4 or IPv6 sub-object MUST also be included. The IPv4 or IPv6 sub-object MUST be pushed onto the RRO prior to pushing on the node-id sub-object. As specified in , if Label_Recording is requested, then the Label Record sub-object is pushed onto the RRO prior to pushing the IP address. All implementations MUST generate either <N, I, L> or <N, L, I, L> and MUST be able to receive and parse both formats in order to find the MP address. No other formats can be generated by a node when the node-id sub-object needs to be added to the RRO. In the <N, L, I, L> format, both values of L MUST be identical. The above rules are based on already deployed implementations of .

No new security issues are introduced beyond those that are described in .

This draft does not have any request for IANA.

The authors would like to thank Miya Kohno, Yakov Rekhter, Avneesh Sachdev and Yasuyuki Matsuoka for their valuable and detailed discussions, reviews and feedback.