-- Rob Allen <[EMAIL PROTECTED]> wrote
(on Thursday, 16 October 2008, 09:44 AM +0100):
>
> On 14 Oct 2008, at 20:52, Wil Sinclair wrote:
>
>> It should also be mentioned that this is something that was identified
>> in the initial performance tests. Matthew will be doing much more
>> performance testing in the next few weeks, so we will hopefully
>> deliver
>> more performance enhancements with the 1.7 production release.
>>
>
> Are the performance tests and the results of them written up anywhere?
>
> I'd be interested to see how they correlate with testing of my
> applications, so the testing methodology used would be useful.
I'll be writing them up following the 1.7.0 release. Some of the tools
we're using are not publically released yet, so any writeup I do now
will not be immediately reproducible.
The basic idea, however, is along these lines:
* Write a series of requests to execute
* Benchmark that series of requests
* Profile each request, and examine what pieces take longest
* Tweak the code
* Benchmark again
* Compare original and secondary benchmark to see if noticable
improvements were made
* Lather, rinse, and repeat
The benchmarking and profiling tools are custom, and those are the
reason I cannot do a meaningful writeup yet.
The requests I'm executing are against the bug application I wrote for
the Dutch PHP Conference workshop I did. Currently, I have it visit a
landing page, the login form, submit credentials, and view a bug listing
page. This allows me to test a typical user visit to the site, and also
allows me to test several expensive operations. I will be writing
several more request profiles to ensure we cover things such as search,
insert/update of databases, translation, etc.
--
Matthew Weier O'Phinney
Software Architect | [EMAIL PROTECTED]
Zend Framework | http://framework.zend.com/
