I recently saw an article in Allaire's Knowledge Base "ColdFusion
Performance Debugging"
http://www.allaire.com/Handlers/index.cfm?ID=8627&Method=Full
They gave several procedures for increasing the performance of your CF
server including simultaneous requests. If I remember correctly they had a
loose guide/formula for determining this number.
The article is a bit outdated, but there is alot of general info on cf
server performance and debugging methods. Hope this helps.
Shane Witbeck
Webmaster
-----Original Message-----
From: Dave Watts [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, August 15, 2000 10:55 PM
To: [EMAIL PROTECTED]
Cc: [EMAIL PROTECTED]
Subject: RE: number of simulataneous page requests
> I was wondering if anyone had any suggestions for that CF
> server parameter?
>
> We did testing with 2 CF servers (and separated DB and File
> Server), an Equalizer load balancer and the WAS load generating
> tool from Microsoft (with 4 client machines). We still found the
> best performance to be with 8-10 per second (10 is the default),
> even though the servers have 512 MB and are dual PIII 650's. I
> would have assumed that they could process more simultaneously...
> Basically, all of our settings are still default for the CF
> servers.
I'd just like to mention a couple of points.
First, the number of simultaneous requests isn't a per-second measurement.
It's the number of requests that CF can handle at any one time. It
corresponds with the number of threads CF uses to service requests. CF,
unlike some other application servers, creates a number of threads to
service requests at startup, then doesn't create or destroy any threads over
the lifetime of the service.
Second, you're not necessarily looking for a high number here. What you're
looking for is the number that will give your site the shortest overall
response time. So, if you set it to a high number, let's say 12, then your
server could service 12 requests simultaneously. However, it might be the
case that additional requests would spend longer in the queue as a result of
the overhead of having 12 threads servicing requests. Optimally, you want
the minimum request time, which includes how long a request sits in the
queue before it's serviced. A higher setting for simultaneous requests
lessens the likelihood of "thread starvation", which may occur when threads
run into problems for various reasons, but it doesn't guarantee better
response time performance.
Dave Watts, CTO, Fig Leaf Software
http://www.figleaf.com/
voice: (202) 797-5496
fax: (202) 797-5444
----------------------------------------------------------------------------
--
To unsubscribe, send a message to [EMAIL PROTECTED] with
'unsubscribe' in the body or visit the list page at www.houseoffusion.com
------------------------------------------------------------------------------
To unsubscribe, send a message to [EMAIL PROTECTED] with
'unsubscribe' in the body or visit the list page at www.houseoffusion.com
