[EMAIL PROTECTED] wrote:

> I'm absolutely a novice in this area, but I think it could be
> improve (it's a good critic)
>
> - Firstly of all, I add some other benchmarks:
>     - CPU and memory usage

  If the network supports the load, it is likely it will be 100%,
  which makes perfect sense: The server will try to process as many
  requests as possible, and hence it will consume all the processor
  resources; a different behaviour would be considered a bug.

>     - how faster a request is done (in average) (how much seconds
>       cherokee spends for serving a page)

  In the benchmarks you have seen, the vertical axis represents how
  many connection the servers replied in one second, so you can figure
  it out just dividing 1 by that number.

>     - relation of size of page and req/s

  Yeah, this is a good point.  We should benchmark different kinds of
  contents: common HTML files, images, relatively big tarballs, may be
  ISO images, etc.  I will keep this in mind :-)

> - Can you compar with lighttpd?. In this page, its creator says that
> it's fast web server. I would like to know how fast
>
> - Finall, can you say me how to generate my own benchmark?. I want
> make my "pinitos" in this area ;-).

  I guess the easier choice is to use "ab". Usually it is included in
  the Apache-tools package.

> And is there any program for make automatical benchmark: for any
> version of CVS, it automatically runs benchmark and publish in
> webpage. And so we have a current (and the most important) and
> automatically a report

  You could write down a script or something to try to do scheduled
  benchmark with the repository code. I guess it is possible..

--
Greetings, alo.
http://www.alobbs.com
_______________________________________________
Cherokee mailing list
[email protected]
http://www.alobbs.com/cgi-bin/mailman/listinfo/cherokee

Reply via email to