----------------------------------------------------
Please read the FAQ at <http://java.apache.org/faq/>
It has a search feature and all the answers!
----------------------------------------------------

On Wed, 13 Oct 1999, Kevin Macclay wrote:

> Suppose you call ab with -n 60 -c 60 to simulate a rush of 60 simulatenous
> hits on a particular page.  With the (c-1) + 1 extra executions, you'll get
> 120 executions instead of 60 -- which makes sense from what you were saying.
> I know that ab will report 60 "complete requests" and not count the other
> requests in it's stats.  My question is, won't these 60 extra executions
> affect the outcome of the results significantly because the server will be
> getting hit with twice as many executions?

If you want to see how a server, sitting there doing nothing, reacts
to getting 60 simultaneous requests all at once and then no further
requests, yes.

However, that isn't typically what you want.  You typically would want
to see how a server that has a more sustained load of 60 requests would
handle them.  In that case, to prevent trailing off at the tail end, you
need to keep issuing requests to keep the load up.

There are lots of ways to do benchmarking, and lots of different
particulars.  The key is that your particulars mesh with what you are 
trying to test and that all your tests are consistent.



--
--------------------------------------------------------------
Please read the FAQ! <http://java.apache.org/faq/>
To subscribe:        [EMAIL PROTECTED]
To unsubscribe:      [EMAIL PROTECTED]
Archives and Other:  <http://java.apache.org/main/mail.html>
Problems?:           [EMAIL PROTECTED]

Reply via email to