Re: squidfs type thing & benchmarking

Assuming your code will run on cobalt mips linux, I can give it
a try.

Benchmarking stuff on Cobalt systems has the advantage that the
machines are so slow that you only need one client+server system to
run polygraph (it can be the same machine).

For reference, Squid 2.1.p2 maxes out at about 39 objects/second on
the Bakeoff-1.0 polygraph configuration. This is on a newer 250Mhz
Cobalt system - "CacheRaQ 2", with 256MB memory. Older 150Mhz systems
get about 25 objects/second.

gid

    
    Is there someone who can take some prototype code and give it a good
    benchmarking? I thinking a test identical (or really close) to the Bakeoff
    test, since we can then compare the results to those results.
    
    I've got my own version of squidfs (except I call it COSS...Cache Object
    Storage System) working, but I'd like to get some benchmarks in before I
    go further with it. Unfortunetly I don't currently have the resources to
    give it a real good test myself.
    
    /-----------------------------------------------------------------------/
    / Eric Stern - PacketStorm Technologies - (519) 826-9395 /
    / http://www.packetstorm.on.ca /
    / WebSpeedWare - high performance caching doesn't have to be expensive! /
    /-----------------------------------------------------------------------/
    
    
    
Received on Tue Jul 29 2003 - 13:15:57 MDT