On Wed, 22 Jan 2003, Pawel Jakub Dawidek wrote:
> Is there a chance to configure polygraph when I don't know how many
> req/sec tested cache machine can handle? Or what to do when I want
> to test this quality?
Polygraph robots can be configured to work in best-effort mode. Simply
omit req_rate field when configuring a robot. A best-effort robot will
send requests as fast as the device under test can respond, with no
more than one transaction outstanding at any time (except for embedded
objects that may be fetched in parallel). Please note that best-effort
workloads often yield misleading results because response rate becomes
dependent on response time, unlike in real life. If nothing else, you
should use a [very] large number of best-effort robots to approach
realistic load conditions.
An alternative approach that may be applicable (depending on what your
test goal is), is a so called "boiling frog" workload. Please search
Polygraph Web site for details on that. In short, it lets you find the
peak request rate at which the device under test can no longer
function well. That peak can be used as an optimistic approximation of
sustained performance.
Finally, our new PeakFinder feature will make boiling frog obsolete:
http://www.web-polygraph.org/docs/userman/inflight.html
> When new web-polygraph version will be available?
Development snapshots are available right now (usually for a fee
unless are doing research, etc.). We do not have a precise schedule
for making the development code public, but it might happen in the
next 3-6 months.
HTH,
Alex.
--
| HTTP performance - Web Polygraph benchmark
www.measurement-factory.com | HTTP compliance+ - Co-Advisor test suite
| all of the above - PolyBox appliance
This archive was generated by hypermail 2b29 : Mon Feb 06 2006 - 12:00:25 MST