Continue in 2 seconds

What are the pros and cons of having a data mart as a separate instance instead of with a data warehouse?

  • Joe Oates, Chuck Kelley
  • December 07 2006, 1:00am EST

Q: Please provide some advice on the pros and cons of having a data mart in separate instance instead of with a data warehouse.

Chuck Kelley's Answer:

Let me first say that it depends on your architected environment. If you have an ETL tool vs. using SQL would change these pros and cons. Mine will be based on using an ETL tool with multiple instances. Pros: 1) You can put the data closer to the user community; 2) You can build the data mart into the appropriate data structure (relational vs. multidimensional); 3) You can tune the two database environments better than one since the workloads will be different.

Cons: 1) You need sufficient network bandwidth to move data; 2) It may take longer to populate (because of having to move more data); 3) More system resources will be required to run multiple instances.

Joe Oates' Answer:

The advantages of having separate data marts includes: initial low cost and implementation per data mart; lack of complexity, relatively speaking, because data from a single business area are a subpart of the business area is involved; the level of technical expertise is lower than that required for an integrated data warehouse; as well as the lack of interdepartmental politics. In my experience, as these data marts proliferate, management often perceives them as a problem rather than an asset.

When data marts proliferate, there are inevitably overlaps of data as well as timing differences. Too often, a manager will get reports from different data marts supposedly about the same subject, but with different answers. There are other disadvantages as well including: since the same source data is shared amongst multiple data marts, the ETL costs become very high; multiple teams of analysts and developers are required to maintain the multiple data marts; inevitably, the users of one data mart need something that is contained in another data mart, thus requiring another ETL project to get the required data.

In my experience, management wants integrated information, that is to say, a single source of the whole truth. This is truly a situation in which the whole exceeds the sum of its parts. There are insights possible from integrated data based on enterprise-wide standard definitions that cannot be obtained from multiple data marts. I have also observed that even though an integrated data warehouse takes more time, costs more and requires a higher level of technical expertise and experience than a single data mart, the total cost of ownership over several years is much higher for multiple data marts.

Undertaking the development of an integrated data warehouse is a large task. If your company does not have people who have had successful experience building an integrated data warehouse, you would be well served to obtain the required expertise from a third party. Additionally, an integrated data warehouse will affect the way that management thinks about doing business. Not everyone will be comfortable with change.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access