[Framework-Team] Re: PLIP load test reports available
me at rpatterson.net
Sun Sep 6 17:02:14 UTC 2009
Eric Steele <ems174 at psu.edu> writes:
> Thanks so much for doing these Ross! These Plone 4 vs previous
> versions comparisons are going to be great for marketing. I'll put out
> a post touting our progress on Tuesday (was going to do it Friday, but
> didn't want to see it swallowed up by the holiday weekend here in the
> One thing we need to keep in mind with the PLIP comparisons is that
> many of them haven't merged in changes to CMFPlone and other packages
> since they did their initial branches so that may be one of the
> factors in their speed differences.
Any chance we can get implementers to do this?
> I'll have to check with the guys at the office, but we may be able to
> spare a VM to do this sort of thing regularly.
FYI, I've been doing these on a m1.large Amazon EC2 instance (2 cores,
8GB). Not ideal from a clean slate, predictable, consistent baseline
perspective, but more so than anything else I have at my disposal.
> On Sep 6, 2009, at 5:03 AM, Ross Patterson wrote:
>> You can find funkload load test benchmark reports and comparisons at:
>> The buildout logs for the individual plips can be found with the same
>> name as the *.cfg file substituting .log for .cfg in:
>> So if a PLIP looks like it has erroneous results take a look at the
>> and let me know if there's anything to be done to get valid results.
>> this soon since the EC2 instance is costing me money so I will take it
>> down soon and I won't want to set the whole thing up again once I take
>> it down.
>> You may also be interested in the following:
>> - Plone 4.0 base against Python 2.4, 2.5, and 2.6
>> - Plone 4.0 against Plone 2.5, 3.0, and 3.3
>> - Plone 3.3 write conflict error comparison
>> http://weblion.psu.edu/static/loadtesting/plone4.0/threads-retries.html> (interesting how well 1 ZEO client with 1 thread does against 2 ZEO
>> clients with 1 thread each)
>> Limi and I are working on adding some explanatory text and legends to
>> these reports but for now this is all you get. :) Click through and
>> around and if there's anything you still can't make sense of, ask and
>> I'll see what I can do.
>> Framework-Team mailing list
>> Framework-Team at lists.plone.org
More information about the Framework-Team