The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2001-Sep> msg00316



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

draft-gan-fast-reroute-0

  • From: francis.arts@alcatel.be
  • Date: Sat, 22 Sep 2001 16:07:05 -0400




Hello Ping et al,

Please find included some comments  / questions to your draft <draft-gan-fast-reroute-00>.

     Francis.

----------------------------

1) Section 2 shows an example of how the detours are established. Assume that node 3 cannot be avoided when calculating the **detour** [node 2 , node 4]. However, assume that there are several parallel links between nodes 2 and 3 as well as between nodes 3
and 4. Should the detour LSP then still be established but traversing on different links between nodes 2 and 3 / between nodes 3 and 4 than the main LSP? [This would protect against link failures between nodes 2 and 3, as well as link failures between
nodes 3 and 4; however no protection against node failures of node 3.]


2) Section 3 states that the presence of the DETOUR object indicates that this is a detour LSP. What are the downstream nodes expected to do with the contents ("source id" and "downstream node id") of the DETOUR object?


3) Section 3 indicates that the FAST_REROUTE object contains the bandwidth. Is this bandwidth the commited information rate?


4) Section 4.1 states that "the detour path computation results in enough information to establish an LSP; this information includes a DETOUR object that specifies the detour's destination" However, from section 3 I understand that the DETOUR object does
not specify a destination; rather it specifies the downstream node id that the detour tries to avoid. Please clarify.


5) Section 4.3 describes the expected behavior of the merging node in the control plane. Is it expected that in the data plane the merging node "throughconnects" the detour LSP immediately onto the main LSP? Or should this only be done when the merging
node detects a failure of the main LSP? [If only done after the main LSP has failed, this results in some latency for the merging node to detect the failure of the main LSP.]


6) Section 4.4 states that "it is possible to have detours traversing through the same nodes as the main LSP". However, when looking into section 2, I understand that it is exactly the idea for the detour LSP not to traverse the same links and the same
nodes as the main LSP (except for the ingress and egress nodes of the detour LSP)? Please clarify.