What options do I have for creating OLAP cubes with Postgres and making it accessible via .net webservices/wcf?

Mr Shoubs picture Mr Shoubs · Oct 26, 2010 · Viewed 15.7k times · Source

We have a large POSTGRESQL transactional database (around 70 million rows in all), and have previously created a data warehouse from this (updated daily) to run reports off of.

To make this more flexible (as lots of different users require different reports and aren't very good at specifying what they want) we would like to create a multi dimensional OLAP cube and expose this via web services to our customers and possibly outsource report creation.

We program in .NET (mainly vb.net) and I believe this can be achieved by using XMLA for the webservice (or WCF) layer, but after a bit of research (everything seems propietory - either SSAS and SQL server, or Jasper Server and Jasper Analysis etc), I'm unsure of the following and wondered if anyone else out there has any experience they can share:

  1. How do (various) front ends integrate with this? - we don't want to tie users to a particular front end.
  2. what front ends are available?
  3. What can I use to build the OLAP cube?
  4. Are there any alternative that I haven't found other than this XMLA approach?

Scalability and Performance are huge factors for us, along with quick development time and an interface that is usable by users who only just learned how to use a mouse :p

Note: ideally this solution be OpenSource and Free or less than £1k (most enterprise solutions are silly money)

Answer

picmate 涅 picture picmate 涅 · Jan 24, 2011

We came across a similar type of an issue and used Modrian (Pentahoe BI) as the OLAP server. This is free and open source. It should provide you access to the OLAP cube from your VB code since Mondrian is also based on XMLA. Front end, as far as browsing the cube and running live mdx queries are concerned, you will need to come up with some solution (actually we are currently developing and integrating the capability to our own BI tool). But we were successful using Mondrian on a web based solution.