Re: [trill] trill ecmp utilization

Hi,

On Wed, May 23, 2012 at 4:13 AM, <dai.xuehui@zte.com.cn> wrote:
>
> Hi Sam，
>
> Thanks for your reply.
>
> with ecmp utilization of all ecmp paths, you may find that a hash
> polarization happens, then what will you do about this?

Polarization is a very well known potential problem. I suppose if you
find hash polarization is happening, you should tell the implementors
of TRILL for your RBridges to fix their implementations.

The ECMP hashing should include some quantity that is unique per
RBridge in a campus, such as the IS-IS System ID or a nickname.
Perhaps the base protocol specification should have said this, but it
didn't, so perhaps it should be included in the Clarifications and
Corrections draft.

Thanks,
Donald
=============================
 Donald E. Eastlake 3rd   +1-508-333-2270 (cell)
 155 Beaver Street, Milford, MA 01757 USA
 d3e3e3@gmail.com

> Best regards,
> Xuehui
>
>
>
> Sam Aldrin <aldrin.ietf@gmail.com>
> 发件人:  trill-bounces@ietf.org
>
> 2012-05-23 13:32
>
> 收件人
> "dai.xuehui@zte.com.cn" <dai.xuehui@zte.com.cn>
> 抄送
> "trill@ietf.org" <trill@ietf.org>
> 主题
> Re: [trill] trill ecmp utilization
>
>
>
> It is per path, not node.
>
> Sam
>
> Sent from my iPhone
>
> On May 22, 2012, at 8:33 PM, dai.xuehui@zte.com.cn wrote:
>
>
> Hi,
>
> I have a problem on the definition of ECMP  utilization in TRILL OAM Req
> draft, which said that:  "OAM MAY provide functionality to monitor
> utilization of ECMP paths to specified destination RBridges. Utilization is
> defined as distribution of traffic over different ECMP members."
>
>
> Given the figure below, RB1 is ingress Rbridge and RB9 is egress Rbridge,
> RB2 has three downstream RBs(RB3, RB4 and RB5) that can reach RB9, and RB6
> has two downstream RBs（RB7 and RB8）to RB9, so there are all 6 ECMP paths
> from RB1 to RB9.
>
> Then, my question is：
> In the current text, does "ECMP member" mean one specific path from RB1 to
> RB9(such as RB1-RB2-RB3-RB6-RB7-RB9)? or one specific node(such as RB2)?
>
> <mime-attachment.gif>
>
>
> Please correct me if my understanding is not correct.
>
>
> Best Regard,
> Xuehui
>
>
>
>
>
>
> _______________________________________________
> 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
>