At 9:57 PM -0400 4/21/02, Chuck wrote:
>I think I see what you are getting to, Howard, but for the purpose of your
>scenario, are you assuming that the enterprise backbone construction makes
>sense?
>
>for example, in your case, are you assuming something like
>Seattle---Portland----SanFran---SanJose---LosAngeles---SanDiego with the ISP
>connections in SanDiego and Seattle, or better yet Portland and LosAngeles?
>
>What I'm seeing no matter how I try to construct this is that, for example,
>half of Seattle's traffic traverses the entire network to get to the
>LosAngeles egress while at the same time, half of SanDiego's traffic is
>going past LosAngeles, and up to Portland.
I don't think we are communicating.
Let me see if I can do some ASCII art. The external metrics are equal
so you don't need to consider them. Again due to the graphics, I'm
not going to try to show hierarchical structure.
POP 1 POP2
Seattle......Portland.........SF..........SJ.............LA..........SD
10 10 10 10 10
With both POPs up, the exit costs are:
POP1 POP2
Seattle 0 50
Portland 10 40
SF 20 30
SJ 30 20
LA 40 10
SD 50 0
So when both POPs are up,
all traffic from Seattle, Portland, and San Fran exit from Seattle
all traffic from San Jose, LA, and San Diego exit from San Diego.
If either POP fails, all traffic goes back to the one operating POP,
until the failed one starts advertising default again.
This method does scale to multiple POPs and links with different metrics.
>
>Maybe I'm digressing. Maybe this isn't necessarily a good design. OTOH, it
>is a design that saves the company money due to the various pricing issues
>involved, no matter what the transport decision. ( interstate, inter-lata,
>inter-telco, etc )
>
>tell me if I am off topic with regard to your puzzle.
>
>Chuck
Not really -- it does get into an example. If you used E2 routes in
this example, with equal metrics, it would be pretty random where
anything went.
But let's say the ISPs are different, and you want ISP2 (SD) treated
ONLY as a backup. You'd assign a cost of 10 to ISP1/Seattle and a
cost of 20 to ISP2/SD.
With both POPs up,
With both POPs up, the exit costs are:
ISP1 ISP2
Seattle 10 20
Portland 10 20
SF 10 20
SJ 10 20
LA 10 20
SD 10 20
-------------
A metaquestion here: AFAIK, this is very real-world stuff that the
CCIE lab is unlikely to get to. The lab prescribes your methods or
leaves you only one real choice, and also doesn't tend to state WHY
you need to do things one way.
This is the sort of thing I'd do in Internet routing training, and in
fact do in my two Wiley books. I happen to think this sort of
knowledge, if you can demonstrate it, gives you a significant leg up
for working for an ISP, or for an enterprise with extensive Internet
connectivity. But it doesn't have an associated Cisco certification.
Is it something that people/companies find meaningful (i.e., worth
money)?
I ask this, in part, because if I'm going to post methods scenarios
and possibly get into them commercially, is this kind useful, or
should I focus more on things likely to be on the lab?
Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=42198&t=42198
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
