Alex Rousskov wrote:
>
> On Tue, 15 Jun 1999, gyan prakash wrote:
>
> > It is able to handle around 100 - 120req/sec just for
> > 10 minutes not more than that.
>
> What response times are you seeing? If your response time is X sec and
> you are doing Y req/sec, then your OS will have to handle at least:
> (X+60)*Y
> concurrent connections in various states. Squid will have to deal with
> (X*Y) connections. You probably should add about 10-20% for workload
> peaks.
>
> >From my experience, things get exponentially worse on a FreeBSD with
> number of connections beyond ~2-3K. Do not know if that is true for
> Linux.
>
> Alex.
Apparently. And with Digital Unix it's like hitting a wall.
D
Received on Tue Jun 15 1999 - 22:31:45 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:46:53 MST