Re: Automatic (performance) testing a sharepoint website
- From: James Dennett <jdennett@xxxxxxx>
- Date: Sun, 13 May 2007 09:28:49 -0700
Phlip wrote:
Michael Bolton wrote:
I anticipate that, in looking for response time, you're getting a
simple answer to a question that's more complex than you might think.
"How long does it take for this response to come back?" may not be as
interesting a question as "Why is it taking so long for a response to
come back?" or "Where might there be a problem that results in
perceived slow performance?" You'll need more sophisticated tools
(and questions) to identify those kinds of problems.
If developers run these tests after every few edits of the source code, any
test that fails its time limit must be caused by the most recent edits.
Generally. So you might not need advanced profiling.
(The developers might also find themselves debugging less..;)
One problem with this is that performance tests, by their
nature, often take a significant length of time to run.
If the time taken to run tests becomes significant, it
becomes an impediment to productivity if it must be done
every time code is compiled, losing one of the benefits
of having fast tests integrated with every build.
Continuous integration can address this; so long as the
tests are run often enough to stop buggy code being added
to the main builds, the tests have done most of their
job, without becoming an annoyance to developers.
Profiling is still worthwhile; just because the latest
edit exposes a performance problem really does not imply
that it caused it. Profiling can establish efficiently
where the performance bottleneck is; it may be in code
which is older but hasn't been tested for scalability
in the past, for example. Knowing which addition exposes
a problem gives a "likely candidate" for where the cause
lies, but it's misleading to think that the best course
of action is *always* to simply revert that last change.
-- James
.
- References:
- Re: Automatic (performance) testing a sharepoint website
- From: Michael Bolton
- Re: Automatic (performance) testing a sharepoint website
- From: Phlip
- Re: Automatic (performance) testing a sharepoint website
- Prev by Date: Re: Automatic (performance) testing a sharepoint website
- Next by Date: Bugzilla supprt on third party access..
- Previous by thread: Re: Automatic (performance) testing a sharepoint website
- Next by thread: Re: Automatic (performance) testing a sharepoint website
- Index(es):
Relevant Pages
|