I wish to develop a web-application much like Google-Statistics, by which I collect statistics on my small customers' finish-customers, and show my clients analysis according to that data.

Qualities:

  • High scalability, handle large volume
  • Compartmentalized - Queries always run on one customer's data
  • Support analytical queries (drill-lower, slices, etc.)

Because of the analytical need, I am thinking about to make use of an OLAP/BI suite, but I am unsure it's intended for this scale. NoSQL database? Simple RDBMS would do?

These things i am using at the office inside a production environnement and delay pills work like no bodies business.

I copled three things

PostgreSQL + LucidDB + Mondrian (More usually the whole Pentaho BI suite components)

  • PostgreSQL : I'm not likely to describe postgresql, really strong free RDBMS enables you to do - certainly - all you need. I personally use it to keep my operational data.

  • LucidDB : LucidDB is definitely an Free column-store database. Highly scalable and can give a really gain of processing time rival PostgreSQL for locating a lot of data. It's not enhanced for transaction processing however for intensive reads. This really is my Datawarehouse database

  • Mondrian : Mondrian is definitely an Free R-OLAP cube. LucidDB managed to get simple to connect individuals two programs together.

I would suggest you to check out the entire Pentaho BI Suite, it worthwhile, you might like to apply certain of there components.

Hope I possibly could help,

I'd state that getting set up OLAP analysis is definitely nice then has great possibility of sophisticated data analysis using MDX.

  • Exactly what do you mean by large volume ?
  • Where are the customer user information?
  • What type of front-finish and confirming will you use?

Cheers.

Disclaimer : I'll have publicity for my very own solution - take a look to www.icCube.com and phone me for additional particulars