Significance of benchmarking?

[eluser]xwero[/eluser]
This and this topic have benchmarking results but in my topic, the second one, i asked the question if the results really matter?

I've did 1, 10, 100 and 1000 iterations but what it means in my topic is that there are that much queries. I think you have to develop complex pages to get 100 queries per page, or not? BorisK added the results of 10m iterations to his thread but in which case do you need to call a helper function 10m times?

The other question is are x iterations the same as x users clicking at the same time on the same link starting the rendering of the page? If that is the case isn't it reduced to a non existing problem when you cache your pages?

I read a lot of opinions about the subject this weekend but i couldn't find a clear line in it. Maybe high traffic website developers could shine some light on this?

[eluser]Sean Murphy[/eluser]
[quote author="xwero" date="1208779601"]
The other question is are x iterations the same as x users clicking at the same time on the same link starting the rendering of the page? If that is the case isn't it reduced to a non existing problem when you cache your pages?[/quote]

No, X iterations are not the same as X concurrent user requesting the same page. Concurrency generally cause issues with the web server (max_connections) rather than a (well designed) web application.

Yes, caching goes a very long way to reduce load and yield better performance. However, it doesn't necessarily fix performance problems, but rather hides them. It's like a band-aid.

Regarding error suppression being slow: it may be slow, but if there isn't a way to make it any faster, then it's a necessary evil.