Never had any performance issues with limesurvey or phpsurveyor. It is simply not an issue. What is much more difficult is to find people who like to take part in a survey.

The main requirement for a system running limesurvey is low energy consumption since it will probably be running 24/7. For this reason I've been running phpsurveyor during some years on a pentium 200 MHZ box. It consumed a decent 45 watts. Performance was great. A nice choice now is to use a mac mini, eventually running linux if you like, since these machines are very energy efficient.

If you are running Windows you might see a less good performance due to the activity of the virus scanner.

Thanks for sharing your experience with us!

Performance might of course be a problem for some that have many respondents that repond in a short time period. For others this is not an issue, as some has stated. Not only LimeSurvey might be the bottleneck I can imagine. The database engine may also have its shortcomings.

Has anyone experience of using LimeSurvey together with MySQL vs PostgreSQL?

A former colleague had a huge performance problem with a database that had extremely many concurrent users. Switching from MySQL to PostgreSQL solved the problem for him, but these two databases has different strengths and weaknessess so I am interested in hearing if anyone has experience with these two in conjunction with LimeSurvey.

I may share some experience in this thread too. I had some real serious trouble with a survey that did heavily depend on conditions (LS-version 1.53). Switching from one group to the next group caused a processor load of 100% for a few seconds on a pentium 3/700. The program causing the load was the apache server. During this time the server was completely locked.
After I reduced the number of conditions by using a smarter survey layout the load got better. Yet I had to move to a stronger server to get an acceptable loadtime.
But as I said: The survey in question had about 600 questions, more than 10 groups and at least two third of the questions depended on one or more conditions.
As at the time postgres support was not yet available I can not say whether this would have made any difference. But I doubt it because conditions have an impact on the group and page and therefore the page building process of the apache. Though the database may affect the loadtime if many users work at the same time on a survey.

Ciao
Joe