Using a data mart for analytics AND as a data source

[min.emerg]

Hi everyone

I am about to start detailed analysis into a data mart for a new client. While they do wish to leverage it to analyse their business processes, they would also like for it to function as a data source for another small application we are to write that will create customer statements for them each month.

I am wary of this request as it might serve to detract from the data mart's ability to provide insight into the business, which is ultimately its primary purpose. We will likely have to include additional information that will not be analysed (at least directly) to satisfy the other application's data needs.

Does leveraging the data mart for this dual purpose sound like a bad idea?

Thanks.

[BoxesAndLines]

Sounds like a terrible idea. Uptime, SLA's, response times, these are all different for an analytical repository than an OLTP repository.

[ngalemmo]

If, by customer statements, you mean some form of account report or financial activity summary, then, as B&L said, it is a terrible idea. Such a statement is clearly an operational report and should be produced from the system of record. It makes everything much more complicated than it needs to be and opens up a host of audit and integrity issues.

[min.emerg]

Thanks for the responses. Very much appreciated.

By statements, I mean "statements of account" that will be sent to customers once a month.

[yuldashev]

Hi,

There are several questions you need to get answered before you make a desicion. Some of them are -
What are the business priorities?
Are there any other requirements for the small application you are going to develop besides being able to provide the monthly reports?
What is the scope of the data mart?
What is the additional information for the application that will not be analysed?
Is it dimensional data or facts?
If it is facts, what data mart they belong to?

To answer your questions if the data mart can be used for analitics and reports at the same time, yes, it can and should. A data warehouse is built not only for analytics but also for reporting purpose.
Of course the reports can be run against the source systems. But if it's a monthly process which has a perfomance impact on the source systems or if there is a complex process should be in place to prepare the reports from different source systems, I would recommend running it against the data warehouse.

By the way, if the Customer dimension is going to be part of the data mart, it should be conformed and have all attributes related to all business processes including the process that generates data for the monthly reports.

Thanks,
Alisher
www.streebo.com

[Sponsored content]