From: Cagri Yucel (cyucel@gmail.com)
Date: Mon May 07 2007 - 13:21:26 ART
Naturally you don't need to do that check for every single route. For
example, in you scenario there are 4 types of routes, EIGRP native, OSPF
native + 2 (ones redistributed into this two). So to make sure 4 iteration
will be sufficient. By and by you will start skipping some of those, for
example in your scenario there is no need to worry about native ones.
Type to find/read edge-core routing concepts of Caslov, it helps
Cagri
On 5/7/07, Gregory Gombas <ggombas@gmail.com> wrote:
>
> Hi Cagri,
>
> You are right on the money.
>
> I forgot to mention that there are EIGRP routers on both sides that
> are redistributing connected interfaces. Therefore routers connecting
> to the OSPF domain were preferring OSPF over the external EIGRP.
>
> So is it safe to say that whenever you have EIGRP external routes you
> may need to tweak adminstrative distances on the edge routers
> connecting to other routing domains?
>
> It must be very time consuming to check the path of every single route...
>
> I wish I could just look at a redistribution scenario and immediately
> know - ah this needs AD tweak, or route tag, or route filter, etc...
>
> Thanks again,
> Greg
>
> On 5/7/07, Cagri Yucel <cyucel@gmail.com> wrote:
> > Hi Greg, To be honest in your scenario you don't need to do anything,
> since
> > EIGRP has its own prevention mechanism against this. AD 90 vs 170. So
> you
> > will be safe.
> >
> > A question arises if you have more protocols being advertised into
> EIGRP, so
> > other EIGRP router will prefer the OSPF route with 110 against this 170
> > root. This is not a loop but suboptimal routing. Best approach is to
> tweak
> > distances for the networks redistributed into EIGRP.
> >
> > I don't have a rule of thumb, I just put the diagram on the table and
> check
> > what happens to every route travelling through the infrastructure.
> >
> > HTH
> >
> >
> > On 5/7/07, Gregory Gombas <ggombas@gmail.com> wrote:
> > >
> > > Hello Group,
> > >
> > > I was wondering if someone can post some guidelines or suggest
> > > recommended reading regarding route redistribution and preventing
> > > route loops?
> > >
> > > I understand there are two main techniques used to preventing route
> > > feedback:
> > > 1. Use route maps to deny tagged routes from being fed back into the
> > > same routing domain.
> > > 2. Change the administrative distance on selected routes to prefer
> > > routes learned from one source over another.
> > >
> > > I'm not always sure which method to use for a given scenario.
> > >
> > > For instance I was recently working on IE Lab 2 where you have the
> > > same EIGRP AS connecting to an OSPF backbone like so:
> > >
> > > EIGRP
> > > | \
> > > OSPF |
> > > | /
> > > EIGRP
> > >
> > > Normally when I have more than one point of redistribution I
> > > immediately configure route maps to tag and deny routes from being fed
> > > back. In this scenario, however, tagging the EIGRP routes on one end
> > > causes them to be denied on the other end (as is the case when the
> > > link connecting the EIGRP networks goes down and its forced through
> > > the OSPF backbone). I had to tweak the administrative distance on a
> > > few routers to get this working.
> > >
> > > So my questions is, is there a general rule of thumb regarding what
> > > type of scenario requires AD tweaking versus route tagging?
> > >
> > > Thanks,
> > > Greg
> > >
> > >
> _______________________________________________________________________
> > > Subscription information may be found at:
> > > http://www.groupstudy.com/list/CCIELab.html
> > >
> >
> >
> >
> > --
> > -cagri
> >
> > _______________________________________________________________________
> > Subscription information may be found at:
> > http://www.groupstudy.com/list/CCIELab.html
> >
>
-- -cagri
This archive was generated by hypermail 2.1.4 : Fri Jun 01 2007 - 06:55:20 ART