From: Howard C. Berkowitz (hcb@gettcomm.com)
Date: Mon Jul 05 2004 - 18:42:44 GMT-3
At 3:33 PM -0400 7/5/04, James wrote:
>Hi,
>
>> next-hops are what the IGP is all about in bgp networks as I'm sure you
>> know. They'll always be there. However, your IGP next hop is a derivative
>> of your BGP next-hop, which in this case is controlled by lpref. If you
>> have an IGP path to a local destination that transits an external network,
>> your network is broken ;-)
>
>I agree with you completely. Unfortunately, CCIE labs often require
>non-realistic, soo wrong and broken setup of BGP / transit configurations that
>it rather scares me. But I believe they are doing this for a reason -- make it
>as complicated as possible to ensure that the student understands
>the protocols,
>which is good challenge.
For some value of "understand". IGPs and BGP are optimized
differently for their intended use. RFC 1930 defines an AS as a
collection of routes and routers, under one or more administrations,
that presents a common routing policy to the Internet. If the two BGP
AS have different views and policies, using a common AS might be
vaguely legitimate.
One of the parts of understanding is to know the performance
implications and the appropriate tool for the job. I honestly don't
believe that pure rote knowledge of odd IOS knobs, plus the misuse of
protocols, adds to understanding.
Of course, Parkhurst has said the goal of the CCIE program is not education.
Maybe silly things like this might be done in an enterprise, but I
really would hope the SP specialization might do better.
Unfortunately, SP seems focused on access devices and L1/L2 stuff,
rather than SP routing--one reason why a CCIE tends not to open doors
in ISPs.
This archive was generated by hypermail 2.1.4 : Sun Aug 01 2004 - 10:11:47 GMT-3