From: Godswill Oletu (oletu@inbox.lv)
Date: Mon May 22 2006 - 08:26:15 ART
Even if the reference is to one Router, I will prefer to err on the side of
caution here and configure the 'auto reference bandwidth' on all routers in
the OSPF routing domain.
Since, OSPF do not rely on gossips, to compute its best path routes; each
OSPF enable router have a complete view of the whole OSPF routing domain and
each router will run the SPF algorithm on all the routes it receive before
arriving at what the best path is. If one change the auto reference
bandwidth on Router A, and when it received all its routes, then run its SPF
algorithm on the routes, then applied the configured auto reference
bandwidth, Router A, from its calculation might deduce that Router C, is the
best path to reach a particular prefix.
On the other hand, Router C, will receive these routes, run the SPF
algorithm on the routes, and by the time it applied its default auto
reference bandwidth to the results, it might independently assume that
Router A is the best path to reach that same prefix; one have created a
routing loop here or at best a suboptimal routing path in the OSPF domain by
not leveling the playing field for all OSPF routers.
In some cases, one might be able to get away with it and will not notice any
problem, but in the CCIE lab, there will be too many things on ones plate,
and you do not want to add to that insanity by creating an unleveled playing
field for all your OSPF routers on how they arrive at what the best route
is. It is possible to get away with it in the lab, in a hub and spoke
topology, where the number of OSPF next hub routers are limited, but in a
multiaccess media like Ethernet, you will see the suboptimal routings and
possible loops if all the OSPF routers on that media do not have the same
auto reference bandwidth value to work with.
If the use of auto reference bandwidth is call for, enabling it on all OSPF
routers, might be a more smarter approach.
HTH
Godswill Oletu
----- Original Message -----
From: "Alexei Monastyrnyi" <alexeim@orcsoftware.com>
To: "Kashif Masood" <kashifmasood27@hotmail.com>
Cc: <ccielab@groupstudy.com>
Sent: Monday, May 22, 2006 4:19 AM
Subject: Re: Few general questions
> Hi.
>
> On the lab they usually mean one router where you have to change
> reference bandwidth, they usually say it explicitly. In the real world
> you should consider the whole OSPF domain to stick with routing
consistency.
>
> "dense mode is not allowed" usually means "sparse-dense is not allowed
> either"
>
> A.
>
> on 22/05/2006 08:28 Kashif Masood wrote:
> > Hi
> >
> > Can you please clarify my doubts. I will appreciate your response
> >
> > 1) If we change auto-cost reference bandwidth on one router, does that
> > means that we have to change the cost in our entire ospf domain or
> > just on this router.
> >
> > 2) Also if they ask us that we cannot use dense mode, does that means
> > we cannot use ip pim sparse-dense mode as well or we can use
> > sparse-dense mode.
> >
> > Thanks for your time
> >
> > Regards
> > Kashif
> >
> > _________________________________________________________________
> > Express yourself instantly with MSN Messenger! Download today - it's
> > FREE! http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/
> >
> > _______________________________________________________________________
> > Subscription information may be found at:
> > http://www.groupstudy.com/list/CCIELab.html
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Thu Jun 01 2006 - 06:33:22 ART