Howdy -

I've got a single 100K file that I'm hitting as part of a stress test to
investigate IIS tunable parameters.

I've got a thousand virtual users doing a GET on this file.

Without compression, I'm getting something on the order of 6 second response
times, and able to get 115 responses per second, and CPU sits around 5-10%.

With compression, I'm getting 1200 responses/sec (yay!) but CPU goes up to
80% and stays there the whole time.

It's my understanding that IIS would only compress this static file once,
and then would cache it and serve it up over and over again - but that's not
what is happening (or some related thing is pushing the CPU up).

Any ideas? Has anyone seen this?

thanks,
jim p.
...in scott$dale, where it's actually muggy.
--
************************
Jim Puckett
Go Daddy Software
Scott$dale, AZ

Re: Static Compression should *not* increase CPU usage, right? by David

David
Thu Apr 06 01:47:46 CDT 2006

What else is configured to execute on requests on this server. Static
compression is kernel cacheable, so it seems like something else is
configured on the server to turn off caching and hence drive up CPU usage.

http://support.microsoft.com/?id=817445

--
//David
IIS
http://blogs.msdn.com/David.Wang
This posting is provided "AS IS" with no warranties, and confers no rights.
//

"Fat Charlie the Archangel" <jpuckett@godaddy.com.dontaddthis> wrote in
message news:5FCE3DF8-9C7D-4C92-BDC9-D40A3F2A2F25@microsoft.com...
>
> Howdy -
>
> I've got a single 100K file that I'm hitting as part of a stress test to
> investigate IIS tunable parameters.
>
> I've got a thousand virtual users doing a GET on this file.
>
> Without compression, I'm getting something on the order of 6 second
> response
> times, and able to get 115 responses per second, and CPU sits around
> 5-10%.
>
> With compression, I'm getting 1200 responses/sec (yay!) but CPU goes up to
> 80% and stays there the whole time.
>
> It's my understanding that IIS would only compress this static file once,
> and then would cache it and serve it up over and over again - but that's
> not
> what is happening (or some related thing is pushing the CPU up).
>
> Any ideas? Has anyone seen this?
>
> thanks,
> jim p.
> ...in scott$dale, where it's actually muggy.
> --
> ************************
> Jim Puckett
> Go Daddy Software
> Scott$dale, AZ



Re: Static Compression should *not* increase CPU usage, right? by jpuckett

jpuckett
Thu Apr 06 13:06:02 CDT 2006


David -

Thanks a heap. It turned out to be that last little entry - trying to load
the default without naming it specifically.

Now I'm churning at 1400 tps and 30% CPU (assuming that that CPU load is
generated by the increased traffic)

thanks again,
jim p.
--
************************
Jim Puckett
Go Daddy Software
Scott$dale, AZ


"David Wang [Msft]" wrote:

> What else is configured to execute on requests on this server. Static
> compression is kernel cacheable, so it seems like something else is
> configured on the server to turn off caching and hence drive up CPU usage.
>
> http://support.microsoft.com/?id=817445
>
> --
> //David
> IIS
> http://blogs.msdn.com/David.Wang
> This posting is provided "AS IS" with no warranties, and confers no rights.
> //
>
> "Fat Charlie the Archangel" <jpuckett@godaddy.com.dontaddthis> wrote in
> message news:5FCE3DF8-9C7D-4C92-BDC9-D40A3F2A2F25@microsoft.com...
> >
> > Howdy -
> >
> > I've got a single 100K file that I'm hitting as part of a stress test to
> > investigate IIS tunable parameters.
> >
> > I've got a thousand virtual users doing a GET on this file.
> >
> > Without compression, I'm getting something on the order of 6 second
> > response
> > times, and able to get 115 responses per second, and CPU sits around
> > 5-10%.
> >
> > With compression, I'm getting 1200 responses/sec (yay!) but CPU goes up to
> > 80% and stays there the whole time.
> >
> > It's my understanding that IIS would only compress this static file once,
> > and then would cache it and serve it up over and over again - but that's
> > not
> > what is happening (or some related thing is pushing the CPU up).
> >
> > Any ideas? Has anyone seen this?
> >
> > thanks,
> > jim p.
> > ...in scott$dale, where it's actually muggy.
> > --
> > ************************
> > Jim Puckett
> > Go Daddy Software
> > Scott$dale, AZ
>
>
>