From: Darby Weaver (darbyweaver@yahoo.com)
Date: Sat Feb 24 2007 - 23:02:30 ART
Danny,
It sounds like you have yet to go to the lab.
With that said, I will advise you here and now, that
you should remain very flexible and not get used to
doing it one way or the other.
I used to always use Loopbacks for my BGP peerings and
while I was at NMC-2, I had the fortune to meet a
certain Brad McConnell who gratiously took the time to
explain to me why I might not want to more than was
being asked for a given scenario. He was correct, but
hey I think the guy averaged 98 or 99 on all the
CheckITs that the NMC crew threw at him...
So who can argue with that.
Now for Item number 2 and redistribution, you will
need to read the reuqirements of your lab very
carefully.
Bruce Caslow and Bob Sinclair, beat the concept of
looking for Layer 1 Physical Loops and Layer 2 Loops
into my brain... about this time last year. I left
NMC-1 with this idea firmly implanted in my brain.
Nowadays, I'm going to add another concept to the
picture, one which I have not yet fully examined and
it may not even be that relevant, but perhaps the
concept of watching out for any Layer 2 Paths - not
seeing this being a troublesome topic for me lately,
especially after recent live experiences with Rapid
Spanning Tree at work and that fantastic job the
Brians from InternetworkExpert did in their Class on
Demand, it was very well explored and unless something
can come along better, it stands in my mind as the
definative reference for Spanning-Tree manipulation
options...
So...
Have I answered your questions?
Darby Weaver
--- Danny Cox <dandermanuk@gmail.com> wrote:
> In the lab there will usually be multiple ways to do
> things. I have a
> couple of examples in mind which illustrate which
> tactics to apply:
>
> 1 BGP neighbours - for iBGP especially, in
> practice, I would
> normally build a link between loopbacks, use
> update-source etc.
> Assuming there is nothing in the question which
> states one way or the
> other what should be done, do you good people have a
> view for whether
> you would prefer loopbacks and update-source over
> just using the
> addresses on the physical links?
>
> 2 redistribution between routing domains - it may
> be possible to
> just use one point of redistribution and avoid the
> fun and games of
> route tagging etc. On the other hand, there may be
> more than two
> points where redistribution could be done, which
> would, arguably, make
> the network more resilient once the tagging or
> distribute-lists are
> sorted out. Would you folk consider that there
> would be a principle
> of doing the CCIE lab to prefer reliability over
> simplicity?
>
>
> Simplicity or reslience .. is either preferred
> where no guidance is
> given to the contrary, or is this just an 'ask the
> proctor' job ? I
> imagine asking the proctor would be good, and I
> think the answer from
> them will be to 'do as you choose within the bounds
> of the question',
> but I'd like to know what the group's general
> attitude is.
>
> cheers
> Danny
>
>
This archive was generated by hypermail 2.1.4 : Thu Mar 01 2007 - 07:38:48 ART