On Sun, Jan 27, 2002 at 11:22:29AM +0100, KF wrote:
>
> > You should really compare the BigIron 8k with the
> > cat6500 or a cisco7600.
>
> Yes, Sure, but mine current generated traffic wasn't 1gig... but totally 400 Mbit Half duplex.... I did not wanted to compare full
> loaded, that's really apples and oranges (have no test bed and for instance Cat 5500 Supervisor III is a 3.6Gbps crossbar switching
> fabric ... I wanted to compare L3 switching and QoS under 400Mbit Half duplex..... I think that this is "fair"... anyway
If your application is 400 Mbps half-duplex, then this may actually be a fairly
good comparison. Price-wise, used BI8k and Cat5500/SupIII/RSM are about right,
maybe. I'm not sure what you were thinking when you sent an email about this
test to cisco-nsp, but I'd like to help you get geared in the right direction.
It would be much easier for testing and making comparisons if a mission statement
of sorts was put forth. I could put a GSR up against an glass of orange juice
and the GSR would lose (assuming my goal is to quench my thirst).
Knowing "what problem you are trying to solve" is probably a good idea before you
start purchasing network equipment or even before evaluating it in a test environment.
My advice to you: don't be so amateurish about testing hardware. It seems as if your
testing isn't really that great. Using SnifferPro 4.5 isn't considered best-of-breed (or
even close) as far as packet generation goes. If you really want to get some solid results
like you want, you are going to have to look at more than just ftp (unless that's the only
use of these switches, which I doubt). You are going to have to test with different packet
sizes and for certain periods of time, etc. Write out a methodology/approach for doing this
correctly, it shouldn't take much of your time. If you need some help, try searching the
Internet and looking for network equipment testing processes. Some good places to start
might be: http://www.ietf.org/html.charters/bmwg-charter.html, http://www.lightreading.com/testing/,
but there are many others. You might *want* to get Foundry or Cisco involved; they both have
excellent labs and testing engineers that can help you with the process. If you feel you'd
like something more neutral in the testing, you don't trust yourself, and you can afford it,
hire someone outside your company to do this for you. Since it seems like you are set on
doing this yourself, try and do the best you can with the resources you have available.
-dre
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:13:30 EDT