Multi-dimensional data processing alternatives

Question

I am looking for a multi-dimensional data processing but without having to use any OLAP. For example, I have a table of sales transactions. I would like to have the total sales on many dimensions like Product, State, Region, Time (monthly, quarterly) etc. The dimensions on which I need the measure (Total Sales) is dynamic and is known only at run time.

I can accomplish this using OLAP technologies (cubes). But I am looking for simpler alternatives (Open source) like possibility of using key-value databases or some software on top of RDBMS (not an ideal option considering performance needs).

Over to you folks !

Correct me if I'm wrong: you want the feature set of OLAP cubes without actually using OLAP cubes, and you're willing to install software to query your OLTP database in order to get those features? — Mike Sherrill 'Cat Recall', Apr 30 '11 at 10:58
MySQL. But I am supposed to keep it DB agnostic. But to start with, even DB specific solutions are fine. — Sathya, May 03 '11 at 11:17

score 1 · Answer 1 · answered May 03 '11 at 12:00

1

Pentaho is the only open source player I know of in this arena. I'm sure there are others. I don't know whether Pentaho supports MySQL, though.

Greenplum Community Edition is built on PostgreSQL.

answered May 03 '11 at 12:00

Mike Sherrill 'Cat Recall'

91,602
17
122
185

Pentaho comes with its own OLAP right ? ( Mondrian). I am looking for a non-OLAP solution. – Sathya May 04 '11 at 06:47

score 1 · Accepted Answer · answered May 03 '11 at 13:46

1

Try build your internal queries around the "With ROLLUP" option. Several Databases (including MySQL have this option). This will allow you to perform OLAP cube like querying within your database by setting a few flags.

http://dev.mysql.com/doc/refman/5.0/en/group-by-modifiers.html

answered May 03 '11 at 13:46

Ralph Winters

297
1
5

looks promising. Will try and let you know. – Sathya May 04 '11 at 06:48

Multi-dimensional data processing alternatives

2 Answers2