Re: More on Bake-Off results ...

On Fri, 15 Oct 1999, Alex Rousskov wrote:

> On Fri, 15 Oct 1999, Marc G. Fournier wrote:
>
> > Just going through the Appendices, and am really curious about something.
> >
> > Squid ran on a PII 333Mhz machine and does 96req/sec ($3960)
> >
> > InfoLibria-L was a Cluster of *4* DynaCache IL-100-7 and did 1600req/sec
> > ($202880)
> >
> > Now, if I had $202,880 to spend, and set up a cluster of 51 Squid servers,
> > I could conceivably see 4896req/sec (51*96req/sec)...no? So, for the same
> > money, I'd be doing 3x better? :)
>
> Not exactly. You have to add the price of network gear and, perhaps,
> software required to put 50 Squids together _and_ make sure that network
> gear does not become a bottleneck.
>
> The 8 port 3Com hub that Squid was using would not scale to 50 Squids. :)

Actually, did a little bit of further figuring out, and if you were to get
a 16port 100baseT Ethernet switch with Squid servers on 15 of those ports,
*assuming* a 1:1 ratio (which isn't reasonable, but making that assumption
anyway)... you would come pretty damn close to the 1600req/sec that the
4Cisco Cluster was doing...

Right now, we've hit a high of 12req/sec through our server, so I figure
we have a little ways to go before it becomes an issue :)

Marc G. Fournier marc.fournier@acadiau.ca
Senior Systems Administrator Acadia University

  "These are my opinions, which are not necessarily shared by my employer"
Received on Fri Oct 15 1999 - 10:36:34 MDT