Re: Ignoring MED (BGP question)

From: karim jamali <karim.jamali_at_gmail.com>
Date: Mon, 22 Nov 2010 19:10:20 +0300

Dear Jack,

I believe Marko explained it very nicely, and always note that you cannot
really control traffic coming into your AS due to the fact that the ISP can
set his weight/local preference in a way to override this. Thus if the ISP
is really concerned about the way the traffic will be routed from himself to
reach your network, he will definitely have to tune Local Pref/Weight..

Thanks

On Mon, Nov 22, 2010 at 6:40 PM, Jack Router <pan.router_at_gmail.com> wrote:

> Zack, CCIE lab is NOT REAL WORLD. I am exploring possibilities to uderstand
> how the stuff works.
>
> Many books and workbooks say that "MED is used as a suggestion to an
> external AS regarding the preferred route into
> the AS that is advertising the metric" (Narbik's BGP Lab16). You can also
> google for "bgp med suggestion" and you will find more examples.
>
> If my understanding of English is correct, a suggestion can be accepted or
> ignored. I want to ignore this suggestion. It looks to me that I can not
> just "ignore" it, there is no command med-ignore. In fact I have to to
> perform some action: override incoming "suggested" metric and set it to 0.
>
>
> On 22 November 2010 07:06, Zack Tennant <ccie_at_tnan.net> wrote:
>
> > 1) I am not "setting MED", it's the default behavior because the routes
> > come from an IGP.
> > 2) Mine is a real-world example, and the routes are coming from different
> > AS's.
> > 3) The carrier should be ignoring MED simply because it's different AS's,
> > but they aren't because of the "always-compare-med".
> > 4) If I were injecting the route from two CEs in the same AS into the
> same
> > carrier, I would expect them to honor MED, as that's what MED was
> designed
> > for.
> >
> > In the example you gave, I would default to Marko's answer. There are
> > other
> > attributes to consider before MED; but then your scenario begs the same
> > question of you, that you posed to me. Why set MED if you're then going
> to
> > ignore it?
> >
> > An option for your scenario is to use "bgp bestpath med missing-as-worst"
> > since it sounds as though you are not setting MED on R2-R3.
> >
> > On Sun, Nov 21, 2010 at 21:55, Jack Router <pan.router_at_gmail.com> wrote:
> >
> > > I am not getting it. You set MED on your router and then tell your ISP
> to
> > > ignore it ? What is the point ?
> > >
> > > On 21 November 2010 20:27, Zack Tennant <ccie_at_tnan.net> wrote:
> > >
> > > > MED should only be compared (by default) when it's coming from the
> same
> > > AS.
> > > > Why do you want to ignore it? Why is setting via RM not clean? I
> > just
> > > > told my carrier to ignore my MED (cause they are using
> > > "always-compare-med")
> > > > and I told them to put this on their existing RM
> > > >
> > > > route-map CUSTOMER_X_INBOUND permit 1
> > > > set metric=0
> > > > continue
> > > >
> > > > It's just that easy.
> > >
> > >
> > > Blogs and organic groups at http://www.ccie.net
> > >
> > > _______________________________________________________________________
> > > Subscription information may be found at:
> > > http://www.groupstudy.com/list/CCIELab.html
> >
> >
> > Blogs and organic groups at http://www.ccie.net
> >
> > _______________________________________________________________________
> > Subscription information may be found at:
> > http://www.groupstudy.com/list/CCIELab.html
>
>
> Blogs and organic groups at http://www.ccie.net
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
>
>
>
>
>
>
>
>

-- 
KJ
Blogs and organic groups at http://www.ccie.net
Received on Mon Nov 22 2010 - 19:10:20 ART

This archive was generated by hypermail 2.2.0 : Sun Dec 05 2010 - 22:14:56 ART