Intro

In pretty much each major version of Tiki, there is work to improve Tiki performance in real world scenarios. This comes from performance limitations hit on *.tiki.org sites or the various sites of community members. As Tiki is over 20 years old, some Tiki sites have increasingly large data sets. Routinely, databases have been sent to the Performance Team for analysis. Our YSlow and PageSpeed scores have improved significantly.

From Tiki4 to Tiki11, there has been significant investment in performance enhancements. Since Tiki11, we think Tiki is generally just as fast or faster than any other comparable PHP web application out there for real World scenarios.

Some examples

Use this page to test: http://tv.tiki.org/Resources
Tiki can have very good scores when well configured

WebPageTest.org

Performance Challenge Capture 2013102311

GTmetrix.com

Here you see great scores for everything except "Combine images using CSS sprites" and "Optimize images". We looked into a Sprite Generator, but it seemed to be tricky to maintain for site managers. We will re-evaluate after the move to Bootstrap, as we transition to Icon Fonts.

2013 10 23 GTmetrix Tv Tiki Org

The Challenge

If you can demonstrate that Tiki is slower than any other comparable web app for a realistic data set and use case, the Tiki Performance Team, sponsored by https://EvoluData.com, will invest some time to optimize, for the following version of Tiki.

Requirements:

  • You must use the latest stable version of Tiki
  • It has to be for a real-world scenario and not just a theoretical speed test with bogus data
  • It must be for a comparably configured system. You can't compare a fully-loaded Tiki with gigs of data compared to a clean out of the box system.
  • You must also have configured your server properly (ex.: activated OPcache, etc.)
  • Not for an experimental feature
  • You accept alternate ways of achieving similar results. If you hit a performance limitation using PluginTrackerList, our answer may to use PluginList, which relies on Unified Index.
  • For a Linux server, because we don't have the expertise to optimize performance for Windows.
  • You must to be ready to send us a copy of real data and some URLs so we can reproduce and optimize. If you have confidential data, you can use Tiki DB Redactor on your real data and reproduce the issue on that. Or replicate the structure in a fresh Tiki instance, and use Faker to fill it up with fake data.


Here is an example: Around Tiki9, we had received an example of a Tiki forum database which was very slow. A test migration to phpBB showed that phpBB could handle that volume much faster. And thus, the performance team proceeded to some optimizations.

If we fail to improve the speed, we will publicly acknowledge that the other application is faster and credit you.

Tiki already claims to be the:


So maybe in a few years, with all challenges met, we'll be able to claim to be fastest FLOSS Web app 😊


How to launch the challenge:

  1. Report the issue as you would any issue on dev.tiki.org/Make-a-wish
  2. Record a screencast of the slowness and upload that video
  3. Assign that ticket to user marclaporte
  4. Marc will review the ticket and likely ask for a data dump.
    • If it's an obvious performance in Tiki, he'll proceed with that. If he feels it's unrealistic to ask for better performance, he'll ask you to prove that performance is better in another application, with a similar setup and data set.
  5. The performance team will get to work to match or beat the performance of the other application
  6. If we fail, we will publicly acknowledge the superior performance of that application and your contribution to the body of online knowledge 😊