Cisco12000 and per packet load balancing...

Discussion in 'Cisco' started by mikester, Feb 28, 2004.

  1. mikester

    mikester Guest

    X no-archive: yes
    In my experience it is hard to get ANY CPU utilization of note on a
    12K's GRP. It isn't a powerhouse CPU to begin with but the
    archetecture of the GSR is distributed. I've used a great many 12k's
    in the past and never had to consider CPU utilization only line card
    capacity (or ports not utilization).

    It is a very robust platform.
    mikester, Feb 28, 2004
    1. Advertisements

  2. mikester

    Captain Guest

    So why do you think all the admins at this ISP so worried about
    over working their 12K?
    Captain, Feb 28, 2004
    1. Advertisements

  3. It's probably a combination of paranoid prevention and ignorance about the
    platform. We've got a few 12000s, and I have never seen any issues with CPU
    load even creeping above 1%. The 12000 is also extremely distributed; i.e.
    every blade is a separate router. I suppose it's possible to overload if
    everyone did per packet, but I don't see that as even remotely feasable.

    Craig Johnson, CCIE #6965
    Craig Johnson, Feb 29, 2004
  4. Doubtful even then, as the per packet/per destination load balancing is
    still part of CEF and as such farmed out to the blades as part of dCEF.
    The CPU isn't involved either way.


    M.C. van den Bovenkamp, Feb 29, 2004
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.