Comments on: Performance Tuning

By: Alan

Alan — Thu, 15 Jan 2009 15:03:33 +0000

I think you’ve misremembered the story. The scenarios I was optimising were running much more that 2 times too slow (i.e. 10+ times slower than the existing product). This was following the replacement of a dummy persistence layer (used during development) with a persistance layer that moved inactive objects out to a database.

The “obvious” place to look was the queries and indexes in the database or faulty logic that decided when to save or recover objects.

As you relate, however, the problem proved to be something else entirely. Which is the main point of the story.

By: John

John — Thu, 08 Jan 2009 11:22:41 +0000

I’ll stick the bill in the post

By: MarkR

MarkR — Wed, 07 Jan 2009 20:45:35 +0000

I think to be honest it’s semantics rather than pedantic (sorry). I would argue that if software is not meeting its performance criteria then there is a problem to be solved. However I agree with you re there being no point optimising code that isn’t working (in the functional sense), and I think this is a point I have so far made by implication, but failed to make explicit.

Thanks for this John, as I think you’ve just handed me an idea for another article

By: John

John — Wed, 07 Jan 2009 20:27:15 +0000

Pedantic perhaps but you’re really looking to identify a bottleneck / which bit of code is used most (and is therefore worth optimising) rather than a “problem”. After all there’s no point optimising code that doesn’t actually work correctly.