Minutes for TRILL at interim-2014-trill-1

Meeting Minutes Transparent Interconnection of Lots of Links (trill) WG
Title Minutes for TRILL at interim-2014-trill-1
State Active
Other versions plain text
Last updated 2014-12-18

Meeting Minutes

TRILL WG Interim Meeting
Chairs: Sue Hares (Hickory Hill Consulting), Jon Hudson (Brocade)
Secretary: Donald Eastlake (Huawei)

Web-Ex, 12/17/2014 from 9-10pm (21:00-22:00 ET)
Other time-zones: 
  6:00pm to  7:00pm PT on 12/17/2014
 10:00am to 11:00pm Beijing time on 12/18/2014

[Notes by the Secretary, edited by the Chairs.  Times shown are as in
original agenda, not time actually taken. 
It was raining very heavily in the silicon valley area which might
have affected some potential participants from that area.]

Susan Hares
Jon Hudson
Donald Eastlake
Pat Thaler
Weiguo Hao
Yizou Li 
Mingui Zhang 

Called to order by Sue Hares at 21:06.

Sue: I'm Sue Hares. Our co-chair Jon Hudson is also on the call as is
Donald Eastlake, our Secretary. [Sue went over the Note Well.  Went
over agenda. Announced that the new Milestones presented in Honolulu
were approved.] As noted in the slides, I'm thinking of a 2nd Interim
meeting January 22nd 2015.  Is that convenient for people?  I'll
inquire on the list.  Anything I missed in the agenda?

Donald: The Directory Assistance stuff should include

Sue: OK. We'll add that.

1) Discussion of Milestones based on schedule of WG LCs and WG
adoptions (3 minutes).

WG LC completed with WG Consensus on all these drafts: 
draft-ietf-trill-aa-multi-attach-02 [12/9/14 to 12/15/14 WG LC]
draft-ietf-trill-cmt-03             [11/16/14 to 12/12/14 WG LC]
draft-ietf-trill-pseudonode-nickname-02 [11/18/14 to 12/9/14 WG LC]
Draft-ietf-trill-oam-mib-01         [11/18 to 12/2/14 WG LC]  

Newly Adoped WG drafts 
draft-ietf-trill-centralized-replication-00    [11/26 to 12/12]
draft-ietf-trill-directory-assistance-encap-00 [11/26 to 12/10]
draft-ietf-trill-smart-endodes-00              [11/25 to 12/09]

Adoption calls i process: 
draft-hao-trill-yang-01               [12/15 to 12/29]
draft-deepak-trill-yang-pm-01         [12/15 to 12/29] 

[See Slides for milestones.] All slides are at: 

Sue: Any questions or comments on the milestones? [No response.]

2) Discussion of TRILL Yang drafts by authors [25 minutes]
draft-hao-trill-yang-01         [Weiguo Hao]    [7 minutes]

Weiguo Hao: [see Slides] Title slide should just say "TRILL Yang", not
"TRILL Yang OAM". I will talk a little about YANG, then TRILL
YANG. Draft only covers TRILL base protocol.  This protocol covers
an overview of the document and a review of the Yang data structure. 
[Due to audio noise problem, some of structure was presented by Sue Hares.]

draft-deepak-trill-yang-pm-00   [Deepak Kumar]  [7 minutes]

Sue Hares: [see Depak Slides] Deepak is probably delayed by the
rain. I can give his presentation on TRILL OAM PM YANG. 
This presentation began by describing the general OAM yang,
Trill OAM yang, NV03 yang, and TRILL OAM Yang. 

Discussion of Yang needs [10 minutes]

Sue: What's missing?

Donald Eastlake: These drafts cover the base TRILL and TRILL OAM PM,
Performance Management, with YANG. But what about covering additions 
like Fine Grained Labeling (FGL) and what about Fault Management OAM?

Sue: Maybe we also need an I2RS interface.

Weiguo: We should cover YANG for AA (Active Active).

Jon Hudson: Brocade is implementing YANG. Has someone inplemented
YANG for standards conformant TRILL and shipped it?

Weiguo: Huawei plans to implement this YANG module. To produce the
draft, we have removed some company specific features.

Sue: In my discussion with Deepak last week, he indicated his specification
on yang for OAM PM functions was already implemented.  

Jon: I think it will be more common than not that people will
implement it.  I think other companies will be doing things similarly
to Huawei.

Yizhou Li: I think we should cover FGL now but hold off a bit on AA
until it is further through the process.

Weiguo: I do not think the base TRILL YANG draft should have any more
added to it. Other things, like FGL, should be augmentations in their
own separate documents.

Sue: That's how most people do it. A base YANG module and then
separate documents augmenting it. Is that OK?

Yizhou and others: Yes. That will work. 

Sue: OK, we will go that way. Separate drafts for FGL and AA.

Sue: I'd be interested if you need to pull the LSDB in a rapid state
model in which case we may want I2RS.

Jon: Yes, I'll check, but I think devices being shipped are lower
latency device so we should offer something more real time. OK if you
can optionally slow it down but not good if you need it faster and
can't get that.

Sue: OK, any further comments before Directory Assist? [None]

3) Directory Service Considerations [30 minutes]

 3.1 3 Drafts on TRILL directory assistance 
 draft-ietf-trill-ia-appsubtlv-02   [Donald Eastlake] [5 minutes]

  Donald: [see Slides] 
[Donald reviewed: draft-ietf-trill-directory-assist-mechanisms,
 draft-ietf-trill-ia-appsubtlv, draft-ietf-trill-channel0-tunnel.] 

3.2 draft-dunbar-trill-directory-assisted-encap [Linda Dunbar] 
[5 minutes]

Linda Dunbar: [see Slides] 
Linda described the directory reliant smart end
node.  This draft has a long history starting in 2011.  
Can use any port, not limited by Appointed Forwarder. 
Linda described the key benefits of the protocol.  

Weiguo: I understand this for unicast but what about multicast? How do
you pick which ingress to use? How would you handle multicast ECMP.

Linda: The end node can encapsulated to the root of the distribution

[After some back and forth between Weiguo and Linda, they decided to
discuss off line handling multi-destination traffic under the
trill-directory-assisted-encap draft.]

Yizhou: There is another draft on smart end nodes. What about that?

Linda: Draft I presented uses Directory. Other draft uses data plane

Donald: Well, I believe the trill-smart-endnodes draft, while
currently written to use data plane learning, could use a directory.
It does require changes at the edge RBridges but it also supports FGL
aware end stations and potentially multi-topology aware end
stations. The directory-assisted-encap draft assumes de-capsulation at
the egress RBridge so only a VLAN can be conveyed to the end station
on output.

Jon: Good points, I think people do want to move more work to end
stations and we probably need to think more about both of these drafts

Sue: It seems clear there is work to do related to multicast and
various forms of smart end nodes. These would be good topics for the
possible January Interim meeting along with YANG models and directory
services. I'll see who I can find to start work on a YANG module work
for AA and FGL.

draft-yizhou-trill-arp-optimization-00 [Yizhou Li] [5 minutes]

Yizhou: [see Slides] Overview. This draft was split out of the
mechanisms draft. Yizhou covered the basic Idea, and the next Step. S
uggest call for adoption after draft has been polished a bit.

Sue: What do people think about a call for adoption on this topic?

Jon: I've been frustrated that very little work has been done on this.
I think users will want this.

Mingui Zhang: This is an item we should be working on.

Sue: I think this an important piece of the Directory Service.

[draft-ietf-trill-smart-endodes-04.txt [TBD] [5 minutes] 
Mentioned above but not presented.]

Discussion on Directory Service [10 minutes]

Sue: Anything else in the Directory Service tool kit not covered by
these drafts?

Jon: I want to think about his a bit. The directory could be a very
valuable resource for the data center. Good to explore further...

Sue: Will bring this up again in January.

Yizhou: Donald presented 3 draft, I'll polish mine to be consistent
with those...

Sue: I'd encourage you to do this and try to get it done by the 15th
of January. Jon, anything else?

Jon: No but polishing up these drafts would be great.

Sue, Jon and Donald: Thanks to everyone for calling in.

Meeting adjourned at 22:16.