Advantages of having datamart (ETL/OLAP) instead of only OLAP

[amb3]

Hello,

Can anyone tell me the advantages of having a Datamart instead of an OLAP tools.

The advantages of ETL/OLAP known to me are as follows :

a) Integration to new data source
b) Consolidation of data coming from different sources
c) Acts on the denormalised database rather than on OLTP database thus, not affecting the response time of the live system.
d) Cross system ata availability.

If you know any positive points for having ETL/OLAP tools,instead of only OLAP , then please let me know.

 

[Dreamland]

amb3 -

If you're still monitoring this thread: I can't help but read your question as "What are the advantages of using sound OLAP architecture for OLAP functions instead of bolting on an OLAP-like front-end?" Is there a reason you have left out any mention of the additional hardware, software, loading and maintenance your analytical store would require? If these were not a factor (hard to believe) then there would be little hestiation by most shops to architect separate OLAP structures left and right.

There is always a cost in resources to build any specialized structures and increased data quality and operational risk in any ETL. That's weighed against the benefits of offloading TP stores, consolidating data and having data structures optimized for analytical processing. It's not an either-or decision, nor is it purely a technical consideration. At the end of the day, someone will have to decide "What's this really worth to the business?"

More often than not - at least in many larger organizations I've been in - there is an obvious and justifiable need for more of this architecture *when viewed from the IT side*. Unfortunately, the business side allocates time and money project by project. It's easier to take the limited project resources at the unit or function-level and buy the latest pseudo-OLAP 'tool' to bolt on to the existing OLTP system.

Bad idea? Sure. But if senior leaders from IT or the business side will no step back to consider the broader implications of building (or burdening) the company's architecture project-by-project, then the technical justification is pointless.