[spb-isis] SPB-802.1aq Interoperability discussions

[Peter Ashwood-Smith <[e-mail]>]

Folks, there has been some off list interest in doing some interop testing
of the new TLV formats etc. as given in
http://tools.ietf.org/html/draft-ietf-isis-ieee-aq-00 so I'd like to bring
the discussion to the SPB list (which is copied).

Anyway what I'd propose is having a brief discussion on the TLV's we want to
test. I don't see much point in testing the data paths or OA&M as that's all
vanilla IEEE stuff so visual inspections of FIBs and the LSDBs is probably
adequate as a first pass, also means a simple LINUX control plane can run
some tests too easily enough.

Anyway what I'd suggest is SPB NLPID in the hello, standard IS-IS
addressing, Level 1, don't worry about the MCID and TOP digest yet.

Then SPB-B-VID TLV for 4 B-VID/ECT-ALGORITHM tuples, suggest B-VIDs 101-104
for ECT-ALGORITHM 1-4.

Then the SPB-Instance sub-TLV, CIST information ignored, Bridge Priority as
configured. M mode, Sp SourceID same as low 20 bits of box mac. Number of
Trees = 4 and then the VLAN-IDs 101..104 for ECT-ALGORITHMs 1..4 with of
course SPVID = 0.

SPB-LINK metric SUB TLV with METRIC as configured.

SPBM-SERVICE IDENTIFIER TLV with BOX-MAC, base VID chosen from 101-104 and
ISIDs as configured.

Suggest for simplicity we pick 4 ISIDs one per B-VID/ECT for simplicity.
ISIDs 1-4 would seem to make sense and all can be tx & rx.

Anyway the above would make a useful first simple test.

Some options include running an adjacency over a static IP port between
different vendors to avoid shipping equipement around, or we can host
something in Ottawa as many of the interested parties all have sites here
within a few kilometers of each other and any one of our labs could be used
easily I suppose.

Anyway let me know your thoughts on the subject. From my perspective I don't
want to make this a big complicated affair, (at least for a first round),
the changes to IS-IS are pretty simple, the datapaths are out of scope
really as they are ordinary IEEE, mostly its just a check that you got the
computations right for tie breaking/B-VID assignment etc.

Peter