Q: Why can't we have data marts which have the same structure but at different locations? For example, consider a telecom provider which operated from multiple countries. Can't we build three data marts with same structure in the three countries? What will these data marts be called?

Joe Oates' Answer:

For a geographically dispersed BI capability, it is normal to have data marts located in different countries rather than pay for high-speed lines from those countries to the headquarters. It certainly helps to have the same structures for dimensions and facts in all locations. However, there is more to it and then merely having the same structures.

The headquarters location will surely wish to provide analyses and reports based on all of the "satellite" data marts and consolidate them into some sort of central data warehouse. One of the issues that will arise is that different countries will have different products or versions of the same product stored in dimension tables. The primary key values of rows for corresponding products will more than likely be different in each country's data mart.

So, even if all that is to be done is to store aggregate fact tables at the center location, it will still require ETL to bring data in from each of the data marts and store it in the central data warehouse. In effect, you'll need to treat each of the data marts as "legacy systems." Other issues include common products having different names due to local language, local sales currencies these that have to be translated into the headquarters accounting consolidation currencies, etc.

Adrienne Tannenbaum's Answer:

Same structure, different data populated is a simple distributed database architecture. Do they have any common values (such as country code)? Things to be dealt with include synchronization and access/security - should each location also have access to the other location's data - same types of access ... Will they all be sourced identically?

Usually one of the data marts is the master/parent/primary part of the overall distributed architecture and serves as the controlling start point in the overall data distribution scenario.

Tom Haughey's Answer:

There is no reason that you cannot have identical structures in multiple locations. Many warehouses do. As a case in point, the financial DW for a large computer manufacturer does exactly this. Figure 1 is an example of it.

Figure 1

The four regional data marts are identical in structure and in software. The only difference is the values each stores. The four identical data marts are, say, North America, South American, Europe-Middle East-Africa, and Americas Far East. In this case, each regional data mart contains only the financial transactions for that region. Each region in turn feeds data to the enterprise warehouse. From the enterprise warehouse, you can run enterprise queries or regional queries. From the regions, you can only run queries for that one region. Another case in point, a consumer products company has 12 business units. Each BU has an identical database, but only contains data for that BU.

Name them what you want. Perhaps, append the region name to the mart in each case if you cannot use the same name.

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