It sounds to me like you are about to fiddle with a poorly understood and barely documented setting. Do you have *any* reason to believe that there is a problem with your site that this might help with?

In particular unless you have a reasonable number of requests (say 5-10 or more) being processed *genuinely* simultaneously over a sustained period I'd be willing to bet money that fiddling with anything to do with threads will have no measurable effect.

I have no data on what might be a normal turn round time for a web request on your site, but let's guess it's about 200mS. So without any parallelism that's 5 per second. Processing 5 in parallel would be 25 per second. On those numbers you probably need to be hitting 1500 requests in your peak minute to make it worth even getting out of bed for.

I'd recommend doing some measurement first. After that - if the disk IO time gave me cause for concern - I'd probably look to cache the template long before looking at threads.

---

Best wishes,
- Richard
Agile Development Consultant, Practitioner, and Trainer
www.dynamisys.co.uk

Fair comment. I *still* say that the first step in optimization is measurement. You need to find out where the time goes. Then you can judge if the time consumed is reasonable and attackable.

It may be that the I/O time is only 1% of the turnaround time. Even if a massive effort made the I/O ten times faster it would probably go completely un-noticed.

Measure first.

---

Best wishes,
- Richard
Agile Development Consultant, Practitioner, and Trainer
www.dynamisys.co.uk