Re: [trill] 答复: active-active solution comparison question

This draft gives an analysis of the current solutions for TRILL 
active-active, I think it is useful for us to learn about the advantages 
and disadvantange of those solutions fastly and to analysize them 
objectively, although there is something required to be perfected. I hope 
it can become much more better in the next version.

Thanks,
Hongjun Zhai
---------------------------------------------
Add: No.68 Zijinghua Road, Yuhuatai District,
Nanjing City, R.P. China, 210012

Tel: +86 025 52877345
Email: zhai.hongjun@zte.com.cn
---------------------------------------------

Haoweiguo <haoweiguo@huawei.com> 
发件人:  "trill" <trill-bounces@ietf.org>
2014-03-11 17:18

收件人
Liyizhou <liyizhou@huawei.com>, "Tissa Senevirathne (tsenevir)" 
<tsenevir@cisco.com>, "windy_1@skyhighway.com" <windy_1@skyhighway.com>, 
"trill@ietf.org" <trill@ietf.org>
抄送

主题
[trill] 答复:  active-active solution comparison question

The draft "draft-hao-trill-analysis-active-active-02" lists all current 
possible solutions to solve TRILL active-active connection problems.  We 
may start from there.
Thanks
weiguo

________________________________________
发件人: Liyizhou [liyizhou@huawei.com]
发送时间: 2014年3月11日 16:44
收件人: Tissa Senevirathne (tsenevir); windy_1@skyhighway.com; 
trill@ietf.org
主题: Re: [trill] active-active solution comparison question

OAM related work started from October 18, 2010 by 
draft-bond-trill-rbridge-oam-00 which turned into 
draft-ietf-trill-rbridge-oam later. Then TRILL WG reached consensus to 
start over from requirement, framework, etc.

Active-Active work also encounters re-writing the problem statement, 
requirement and online/offline multiple solutions debates. It is 
progressing.
There are multiple solutions available and analyzed in the document 
repository.

I agree active-active is very important and should be given the highest 
priority currently in WG.
So we should start on the technical discussion about how to make the whole 
solution work ASAP.

Yizhou

> -----Original Message-----
> From: trill [mailto:trill-bounces@ietf.org] On Behalf Of Tissa
> Senevirathne (tsenevir)
> Sent: Saturday, March 08, 2014 7:24 AM
> To: windy_1@skyhighway.com; trill@ietf.org
> Subject: Re: [trill] active-active solution comparison question
>
> We should not take our eyes away from key points. Like I stated during
> the meeting,
>
> 1. Active-active is one of the most important enhancements to TRILL,
> without that TRILL has no story in data centers 2. We started serious
> discussions of Active-Active in Taipei and now almost 3 years later we
> are still having the same discussion and no acceptable solution, which
> means we are at least 2 years late.
>
> Fine print: OAM discussions also started in Taipei, during this time we
> have delivered, Requirement Document, Frame work document, Completed
> Liaison with IEEE 8021, Fault Management , Loss Delay drafts are on
> Last call and MIB document is a WG document. Active-Active still on
> discussion. (borrowing from John), if you have read this far it is no
> longer fine print, it is big bold letters that "we, as a working group
> have failed to deliver on active-active edge solution". Let's leave
> differences, theoretical discussions a side , let's get the work done.
>
> -----Original Message-----
> From: trill [mailto:trill-bounces@ietf.org] On Behalf Of
> windy_1@skyhighway.com
> Sent: Friday, March 07, 2014 8:58 AM
> To: trill@ietf.org
> Subject: Re: [trill] active-active solution comparison question
>
>
> I agree with Jon. I like this statement from Dragnet "Just the facts
> ma'am, nothing but the facts." And, although this is my opinion, I
> believe that facts and examples are the easiest way to convince other
> people that one has a valid point.
> gayle
>
> <rant>
> In general if anyone in a presentation is going to make subjective
> qualitative statements about X is easier that Y, or Z is less
> complicated than W, or A is faster than B.
>
> Please show your work/logic.
>
> These items are some of the most likely to upset folks, cause ego
> challenges etc.
>
> No where else can you make performance or complexity a compare value
> without giving a data reference. If you did that in front of a customer
> many would just laugh.
>
> And do not to pick this case, as I have seen _much_ worse. But it's a
> good example because I think there was a very real attempt to be fair
> and open.
> I don't think there was a hidden agenda here as there sometimes is. And
> even with that good intention we end up with statements like:
>
> CMT scalability = Medium
> Centralized replication = High
>
> That's marketing not engineering in my opinion.
>
> But most importantly it is inflammatory and not helpful in encouraging
> mature debate.
>
> </rant>
> (If you really read this far, thank you) J
>
> > On Mar 7, 2014, at 2:15 PM, "Tissa Senevirathne (tsenevir)"
> > <tsenevir@cisco.com> wrote:
> > Hi Haoweguo
> > In your presentation you indicated  CMT scalability is Medium and
> > Centralized replicator scalability is high. Can you please explain
> the
> > reasoning behind this ?
> > Thanks
> > Tissa
>
>
> _______________________________________________
> 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
_______________________________________________
trill mailing list
trill@ietf.org
https://www.ietf.org/mailman/listinfo/trill