Re: Query performance on merge tables after upgrade

[Bjorn Julander]

While I don't know about specific changed between 4.X and 5.0,
in general merge-tables and optimising of queries don't always go
hand in hand.

As I understand it, each column contain some statistics, like max/min
etc., which is used by the optimier.
For merge tables the 'unified' view of each column also has max/min,
but this will be max/min for all underlying. The effect is that some
columns seem to be vary 'wide' or similar effecting the optimiser.

We use systems with many, many large underlying tables using merge, and
we're often forced to suggest/force index to get the correct order of
execution.

I know too little about your system, but partitioning on some good
column might be a very good idea here??

//Bjorn