Hi everybody.

I'm using Mondrian combined with Saiku to make some analysis and I'm facing some performance problems.

I build a simple cube using based in a star model. Everything works OK until some points when Mondrian decides split a single MDX query in multiple SQL queries.

Analyzing the problem I have noticed that when I asked for several dimensions Mondrian get the data making several SQL queries asking for a group in one of the dimensions.

With three or four dimensions I got queries similar to this:

SELECT dimension1, dimension2, dimension3, dimension4, sum(metric1)
FROM fact_table1
GROUP BY dimension1, dimension2, dimension3, dimension4

But when I add a new one it becames in several queries (more than 100) similar to this:

SELECT dimension1, dimension2, dimension3, dimension4, dimension5, sum(metric1)
FROM fact_table1
WHERE dimension1 in ('xxx', 'yyy')
GROUP BY dimension1, dimension2, dimension3, dimension4

I've looking for some parametrization to avoid to split queries but I haven't found anything that change the way the queries are generated.

Does somebody face problem like this? Any tips?

Thanks