Re: Star joins?

[Timour Katchaounov]

Hi Seth,

let me first clarify what I understand under a "star join". To me a star join is
a join where many (usually small) dimension tables are joined with one
(usually big) facts table.
For example:

select * from facts, d1, d2, d3
where facts.c1 = d1.c1 and facts.c2 = d2.c2 and facts.c3 = d3.c3;

If that is what you mean, then there is a current project we are working on
that will speed up considerably the execution of such joins. To be more
precise, the project will add a new join method called hash-join, therefore it
is mostly related to the query execution engine and less to the optimizer itself.

At the moment I cannot give any specific timeframe for it, but the project is
being actively worked on.

Let me know if this answers your question.

As part of our work we develop various tests and benchmars. It would be of
great help to us if you could provide concrete examples of BI queries, scenarios,
etc. where you noticed unsatisfactory performance of the current MySQL (5.0 or
5.1). It would be great if you could give us some numbers - table sizes, indexes,
query execution times.

--
Timour Katchaounov, Software Engineer - Optimizer
MySQL AB, www.mysql.com