[Zope] Testing the performance of SQL Methods

[Peter Bengtsson]

I want to find the perfect caching settings on our Z SQL Methods in the
Advanced tab, and to do this I have prepared a little script that loops
through a bunch different setting combinations and calls different little
methods that does the processing.

Let's not worry about the settings or the data analysis for the moment.
What I want is a stupid little function that uses a SQL method and is very
process heavy and should use the internal caching.


So far I have tested a few different little functions:

- One that reads lines from a file, checks if the line exists, adds the
line. (about 10 000 selects and about 10 000 insert statements)

- One that reads 5 lines from a file, check if one of them, adds the new of
the five. More or less like the above but it waits a little bit

- One that selects all records from a table 10 times per setting (10*5000
records selected per setting)


None of these methods have produced any useful results, except from one: No
correlation.

The only deviations I see in process performance is depending on the
activity of my Windows 2000. If Outlook decides to go and check for new
mail, the particular test during that time takes a few milliseconds longer.


Say something that I can code that will surely use the internal caching of Z
SQL Methods.
Is there a way of safely prove that Z SQL Method result caching is actually
being done.


Not a mathematician, yet
Peter