Re: [trill] Any comments on Coordinated Multicast Trees (CMT) for TRILL - draft-ietf-trill-cmt-00

Hi Minggui and Sunny，

I also think the expensive solution for link failure is the right 
direction. It is simple and effective for this issue, although it consumes 
much more nicknames (I assume that there are no so much RBriges to use up 
all the nicknames in a Data Center).

As for the issue that switches are interconnected, I think the STP should 
be employed to break the link loops out of the TRILL campus. Even if 
without RBridge Aggregation, CE swithches-internnecting issue, given by 
Sunny, also exists.

Thanks,
Zhai Hongjun
""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
 Protocol Development Dept.VI, Central R&D Institute, ZTE Corporation
 No. 68, Zijinghua Road, Yuhuatai District, Nanjing, P.R.China, 210012

 Zhai Hongjun

 Tel: +86-25-52877345
 Email: zhai.hongjun@zte.com.cn
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""

Mingui Zhang <zhangmingui@huawei.com> 
发件人:  trill-bounces@ietf.org
2012-08-16 16:08

收件人
Sunny Rajagopalan <sunny.rajagopalan@us.ibm.com>
抄送
Erik Nordmark <nordmark@acm.org>, "trill-bounces@ietf.org" 
<trill-bounces@ietf.org>, "trill@ietf.org" <trill@ietf.org>
主题
Re: [trill] Any comments on Coordinated Multicast Trees (CMT) for TRILL - 
draft-ietf-trill-cmt-00

Hi Sunny, 

The scenarios that links fail or CE switches are interconnected are 
interesting and make sense. The expensive (may lead to nickname 
exhaustion) solution you had pointed out is also the right direction I can 
think of right now. The only issue with this solution is its scalability.

Having said that, I think the failure recovery issue for an RBridge 
Aggregation is not that simple. It at least needs a control plane 
coordination among the aggregated members. Think about that a member 
joins/leaves the aggregation. A fast failure recovery mechanism for the 
RBridge Aggregation can be developed based on that coordination. When I 
suggested to  purchase this topic in a separate ID, I am on behalf of 
myself. I’d like to hear opinions from other guys.

Thanks,
Mingui

From: Sunny Rajagopalan [mailto:sunny.rajagopalan@us.ibm.com] 
Sent: Thursday, August 16, 2012 9:36 AM
To: Mingui Zhang
Cc: Erik Nordmark; trill-bounces@ietf.org; trill@ietf.org
Subject: Re: [trill] Any comments on Coordinated Multicast Trees (CMT) for 
TRILL - draft-ietf-trill-cmt-00

Mingui,
This solution cannot be evaluated or deployed without knowing it's 
approach towards link failures. 
IMO, the draft is incomplete without it, and a solution should not be 
deferred to another draft.
--
Sunny

On Aug 15, 2012, at 5:54 PM, "Mingui Zhang" <zhangmingui@huawei.com> 
wrote:
I had read this draft. 

IMO, the updated version focus on the specific RPFC issue in active-active 
multi-homing. The failure recovery from links fails  can be regarded as 
another common issue of active-active multi-homing, which may  be handled 
in a separate ID. 

Thanks,
Mingui

From: trill-bounces@ietf.org [mailto:trill-bounces@ietf.org] On Behalf Of 
Sunny Rajagopalan
Sent: Thursday, August 16, 2012 6:28 AM
To: Erik Nordmark
Cc: trill-bounces@ietf.org; trill@ietf.org
Subject: Re: [trill] Any comments on Coordinated Multicast Trees (CMT) for 
TRILL - draft-ietf-trill-cmt-00

I had sent out comments on this a while back, and none of them have been 
addressed. 
To summarize the issues: 
1) What happens when one of the CE-rbridge links fails? The draft doesn't 
talk about this scenario. I listed the possible solutions in my earlier 
email; are the authors planning to adopt one of them, or do they have 
another fix in mind? 

2) The solution doesn't work if the CE switches are interconnected, but 
this is but this is not documented in the draft. What happens if they get 
interconnected accidentally? What is the process for recovery? This needs 
to be discussed in the draft. 
More details here: 
http://www.ietf.org/mail-archive/web/trill/current/msg04984.html 

From:        Erik Nordmark <nordmark@acm.org> 
To:        "trill@ietf.org" <trill@ietf.org> 
Date:        08/15/2012 03:14 PM 
Subject:        [trill] Any comments on Coordinated Multicast Trees (CMT) 
for TRILL        - draft-ietf-trill-cmt-00 
Sent by:        trill-bounces@ietf.org 

We are trying to understand if this document is getting close to WG last 
call, but we don't know how many have read the document.

If you have reviewed the document, please let us know.
If not, this might be a good time to review the document and provide 
comments.

Thanks,
   Erik and Donald
_______________________________________________
trill mailing list
trill@ietf.org
https://www.ietf.org/mailman/listinfo/trill

_______________________________________________
trill mailing list
trill@ietf.org
https://www.ietf.org/mailman/listinfo/trill