RE: CEF Polarization?

You are right about per-desination vs per-packet. What I've seen is that
per-destination is never 50-50 but it is almost there, but the one I am
experincing is like 1 - 2 ratio.

> From: markom_at_ipexpert.com
> Date: Tue, 16 Mar 2010 15:55:52 +0000
> Subject: Re: CEF Polarization?
> To: muzammil_mukhtar_at_hotmail.com
> CC: ccielab_at_groupstudy.com
>
> On Tue, Mar 16, 2010 at 15:43, Muzammil Mukhtar
> <muzammil_mukhtar_at_hotmail.com> wrote:
> > Hello Group,
> >
> > I ve a question about the issue I am experiencing. B I want to load-share
the
> > traffic across two links but it is not being shared properly. CEF is
showing
> > traffic share of 1 and 1 for each path but in actuall traffic is going
out
> > more from link compare to other.
> >
> > Could some one shed some light on what CEF Polization is? In my
understanding
> > CEF polarization happens where the decision is made on each/different
levels.
>
> What you are experiencing is not CEF polarization. CEF polarization is
> "global" event in your network when multiple links and devices using
> the same load sharing hash overutilize set of links.
>
> What you are most likely experiencing is the result of
> "per-destination" load sharing (in most cases, it's per-flow
> actually). When you use his method, load will never be 50:50. If you
> wish 50:50 split, you should consider per-packet load sharing, but
> beware of consequences.
>
> Also, on some platforms you can manipulate what determines a link to
> use. I.e. what components of a flow to utilize. You should investigate
> the options of the platform you use.
>
> --
> Marko Milivojevic - CCIE #18427
> Senior Technical Instructor - IPexpert
>
> Mailto: markom_at_ipexpert.com
> Telephone: +1.810.326.1444
> Fax: +1.810.454.0130
> Web: http://www.ipexpert.com/
>
>
> Blogs and organic groups at http://www.ccie.net
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
>
>
>
>
>
>
>
Received on Tue Mar 16 2010 - 16:01:54 ART