Testing and tuning TYPO3 performance
By Michael Scharkow & Steffen Müller
TYPO3 has a reputation of being a resource hog, and performance on popular websites (such as typo3.org) has often been somewhat flaky in the past. This article addresses some recommendations on performance tuning as well as some testing scenarios with regard to TYPO3.
Intended Audience
Website administrators and people who plan to launch popular and heavily frequented sites with TYPO3.
Some basics about website performance
"Premature optimization is the root of all evil."
C.A.R. Hoare
Ideally, website performance should not be an issue as long as your application is scalable with more and better hardware. Why bother with expensive performance optimization when you can simply throw more hardware at the problem? This principle is successfully applied in basically all really large dynamic websites, such as Wikipedia, Slashdot, LiveJournal and most prominently Google.
TYPO3 is in our opinion basically suitable for large and popular sites. But most websites are probably not that big, so this paper examines performance in a typical one server setup.
Our central question is: What can you do about TYPO3 performance except buying more and better hardware? Which factors are influencing your site's speed and server load? Where should you start optimizing, thus avoiding premature optimization? How can you test your website, or even simulate a slashdotting (which is basically a friendly Denial of Service (DOS) attack)?
Performance tests sometimes come along with unforeseen and nasty consequences. Before you start, make sure that you don't run into troubles with your ISP/ hosting provider. Don't do anything on your production server unless you know what you are doing. You might crash your machine or get locked out from it. In any case, this document comes with absolutely no warranty.
