“Plone is too slow” – you hear that a lot. Everyone that has used Plone has been frustrated by it at some point. When you want to start up an instance it can seem instant or it can seem like it takes an eternity. The same is true for rendering pages. More recently seeming instant has become more and more common.
That’s because work has explicitly gone into making Plone 4 faster. My friends at netsight ran the Bristol Performance Sprint about year ago explicitly aiming to improve performance. Plone 4 runs on Python 2.6, has support for BLOBs, and various other changes that all improve performance.
But has it worked?
I have been setting up this site in my spare time and pretty soon it was obvious that I needed a HTTP cache, but it wasn’t the same realisation as usual: I wasn’t waiting for Plone when I realised. The slowest bit of the site was the blog. Now, I don’t claim to be a wordpress expert, so it was an out of the box installation. However, the Plone site is untuned too and that wasn’t as much of an issue. At the request of Alex Limi I did some quick benchmarks of the uncached site:
I used apachebench to access the Plone and Wordpress sections of the site 150 times and measured how many requests each could serve. I repeated this both with my theming layer in place and with the standard skin. For both, the differences between the raw and themed versions were so small to be discounted, but the difference between the two platforms is pronounced.
So, the next time somebody complains Plone is too slow, remember Plone 4 is just around the corner.
Tags: performance, Plone, speed, wordpress

someone will say “but Wordpress does more!”
Note that ApacheBench measures the number of requests for a single resource. If you want to benchmark how long it takes for the complete page with all its resources to load you should use Funkload, jmeter or just Firebug for a quick informal test.
What kind of test server used? 2.5 req/sec seems little slow even for Plone/Wordpress – I assume you used highly parallel setup?
I also suggest checking a write performance. Wordpress page save vs. Plone page save. Namely anything involving read can be made as fast as possible (just use caches), but writes are the thing separating winners from losers.
Seriously. How can WordPress behind Deliverance be faster?
Guys…
Daniel: it can be faster if the difference is just experimental error, computers are a ccomplex thing, and although I did a few runs it could easily be that googlebot interfered.
Mikko: bottom of the line slicehost server, bog standard installs. I’m not surprised at the range I got, as you get what you pay for. Once caching is configured on Plone I’m confident the number willbe orders of magnitude higher.
Roche: Yeah, I know, I didn’t want css and images confusing the results, it would have made the raw/deliv distinction meaningless too
[...] Plone 4 faster than WordPress [...]