Here is some more performance results based on Ruediger's suggestion :
-------------------------------------------------------------------------
test2.html : 500 substitution s/hello/hi/g.
Requests Per Second (RPS) as measured by ab for s/hello/hi/ substitution.
Enabled CPUs : 1
testname Concur MaxReq mod_sed mod_line_edit mod_substitute
mod_substitute(with n flag)
RPS RPS RPS
test2.html 100 100000 1278 404 1130
1531
Summary :
On simple string substitution (s/hello/hi/) mod_sed perform better than
mod_substitue but with flag [n] mod_substitute performed better than
mod_sed.
-------------------------------------------------------------------------
Test case summary : pattern capture
Regular expression used : OutputSed
's/.*\(one\).*\(two\).*\(three\).*\(four\).*\(five\).*/\5 \4 \3 \2 \1/'
: Substitute
"s/.*(one).*(two).*(three).*(four).*(five).*/$5 $4 $3 $2 $1/"
String : hello one 1 two 2 three 3 four 4 five 5.
Enabled CPUs : 4
test12.html : 500 lines containing the above string (Result in 500
substitution).
testname Concur MaxReq mod_sed mod_substitute
RPS RPS
test12.html 125 100000 1322 3230
Summary :
With pattern capture, mod_substitute performed much better than mod_sed.
-------------------------------------------------------------------------
Test case summary : test for simple regular expression.
Regular expression used : OutputSed 's/.*one.*two/1 2/'
: Substitute "s/.*one.*two/1 2/"
String : hello one 1 two 2 three 3 four 4 five 5.
Enabled CPUs : 4
test22.html : 500 lines containing the above string (Result in 500
substitution).
testname Concur MaxReq mod_sed mod_substitute
RPS RPS
test22.html 400 400000 2013 1689
Summary :
With ordinary regular expression, mod_sed performed better than
mod_substitute.
-------------------------------------------------------------------------
mod_substitute (what is available in 2.2.8 is leaking memory) so the above
performance results must be repeated with new mod_substitute from trunk (to
have better comparison).
Regards,
Basant.
On Thu, Apr 10, 2008 at 01:44:53PM +0200, Ruediger Pluem wrote:
> On 10.04.2008 13:27, Jim Jagielski wrote:
>>
>> On Apr 9, 2008, at 5:09 PM, Basant Kukreja wrote:
>>>> just that, too. Any performance comparisons with its competitors?
>>> Some quick performance comparison results between mod_sed, mod_line_edit
>>> and
>>> mod_substitute.
>>>
>>> Search/Replace string : s/hello/hi/
>>> test0.html : No matches, so no substitute.
>>> test.html : 1 comparision/substitute
>>> test2.html : 500 comparision/substitute
>>>
>>> Performance data was generated with 200 clients for 100000 requests (for
>>> worker
>>> mpm).
>>> $ ab -c 200 -n 100000 <http-uri>
>>>
>>> Requests Per Second (RPS) as measured by ab for s/hello/hi/ substitution.
>>> mod_sed mod_line_edit
>>> mod_substitute
>>> test0.html 3263 3206 3129
>>> test.html 3229 3213 3197
>>> test2.html 1278 404 1130
>>>
>>> Note that mod_sed performs better than either mod_line_edit or
>>> mod_substitute.
>>
>> Interesting... What versions of each did you test?
>
> Would be interesting to know what happens speed wise if you use the n flag
> for mod_substitute to use strmatch instead of a regular expression.
> And depending on the version you used it would be interesting what happens
> if you use the q flag (applys only to trunk).
> Furthermore from a first glance of mod_sed I would guess that it runs into
> trouble with large static files. So could you test mod_sed with one of
> the simple cases above and a 3 GB file and monitor httpd's memory consumption
> in comparison to mod_substitute?
>
> Regards
>
> Rüdiger
>
>
