Re: [mpls] RMR updates

["Dongjie (Jimmy)" <[e-mail]>]

Hi Kireeti,

Some comments from my side:


1.       If the “half-ring” is the physical network topology, then I’d agree with Huub that it is similar to a linear topology. Then RMR may not need to consider this scenario.


2.       As for hub nodes protection, IMO it is definitely needed. One possible solution maybe is to treat the two hub nodes as a virtual node, and allocate the same ring label. OAM can be used to detect the failure and trigger ring LSP switch over. This is described in the shared ring protection document.

Best regards,
Jie

From: mpls [mailto:[e-mail]] On Behalf Of Kireeti Kompella
Sent: Thursday, November 09, 2017 5:14 AM
To: [e-mail]
Subject: [mpls] RMR updates

Hi All,

We'd like to move to last call on the RMR document, but first there are a couple of loose ends we'd like to tie up, specifically, the "Advanced topics" in the -05 version of the draft.

a) how should the RMR document deal with "half-rings"?


Consider the following half-ring:

    A    B
    |    |
    C----D


Here are some options:
i) the RMR document shouldn't concern itself with this case
ii) create a PW from A to B to complete the ring
ii) instead of creating N LSPs (in this case 4), only create LSPs from A to B and from B to A.

b) how should the RMR document deal with protecting hub nodes?
i) not deal with it.
ii) use technology akin to PIC edge

We feel that it is important to address both these issues. We are soliciting your response, either among the options suggested, or potentially other ideas.

Thanks,
--
Kireeti