>Hi,
>
>I just wanted to know your opinion as a expert of router.
>When I choose a core/edge router (72xx/75xx/12xx, Mx series) in what part I
>must considered carefully.
>Probably, in forwarding capacity, cpu/routing engine utilization, how can
>the router manage router flappings, how can I push the router over his limit
>...injecting 10.000.000 bgp routes and see what would be happened in that
>router....:)
>Just let me know your opinion ...please
>Thanks for you opinion
>
>Regards
>Sipitung
Well, speaking as an expert, I first try to come up with a fairly
specific definition of the way the router is going to be used. Don't
feel bad about this -- it's an exercise I constantly go through with
my internal product groups, pointing out to them that not all BGP
routers are the same. A customer site multihoming router, a POP
aggregation router, a POP edge router to the intraprovider core, an
intraprovider core router, and an interprovider border router all
have different needs. See appendix A of
http://www.ietf.org/internet-drafts/draft-berkowitz-bgpcon-01.txt.
Knowing the ultimate breakdown limits of a router isn't necessarily
going to help tell you how the router will behave in your
application. Take your example of 10,000,000 routes.
What if you are evaluating a POP aggregation router with 1000
customer interfaces? On most of these, you will receive less than 10
prefixes, and it's good operational design to use prefix-limit to be
sure you don't receive more than 100 routes per peer.
In the other direction, towards the customer, only a small fraction
will want full routes. Many will want just your customer routes,
which certainly is smaller than the default-free table. So the POP
router may not even need a full table, but the 30-50% of it that
comprises customer routes.
10 million routes? Even with the growth rate of the Internet going
exponential again, I wouldn't see that happening for several years
yet. By then, we will have new router generations.
What are you trying to understand about what happens when the router
goes over its limits? It may be possible to understand that simply
from detailed understanding of the protocols and other mechanisms
involved, without hands-on tests.
>
>
>-----Original Message-----
>From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of
>Howard C. Berkowitz
>Sent: Saturday, April 21, 2001 10:16 PM
>To: [EMAIL PROTECTED]
>Subject: Re: Router Tester [7:1479]
>
>
>>Hi groups,
>>
>>Does anyone have clues,comments about the best router tester which can
>>simulates traffics, routes, route damping and so on.
>>I have heard Agillent, Smartbits, GN Nettest but still can't decided which
>>one the best.
>>Any helps, clues, suggestion will be appreciated.
>>
>>
>>Regards
>>sipitung
>
>What problem are you trying to solve in your router testing?
>
>Are you trying to measure the forwarding capacity? The convergence
>characteristics with and without policy? (see
>http://www.ietf.org/internet-drafts/draft-berkowitz-bgpcon-01.txt)
>
>Are you trying to verify the correctness of the protocol implementation?
>
>How do you define "best?" If a given tester cost twice as much as
>another, but could impose 95% of line rate rather than 75% of line
>reate, is this important?
>
>What kind of reports do you expect the tester to give you? How much
>programming are you willing to do to analyze test results?
Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=1600&t=1479
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
