Part one of this series (November 1999) focused on the business characteristics of your organization and how they influence the decision of whether to build the operational data store (ODS) or the data warehouse (DW) first. (Note: for information on the differences and purposes for these two constructs, please read the five ODS articles appearing in DM Review's July/August 1998 issue.)
This column focuses on the more technological or project-oriented reasons for deciding which way you should jump. Let's set the stage. Suppose you find yourself in the rare and enviable position in which your organization has stated that both of these environments are critical and there is no clear winner. The business community and corporate politicos have determined that both are important, realize that only one can be built at a time and have left the ultimate decision up to you. Well, here are some suggested topics to consider before making your decision.
All right, let's look at the technological nuts and bolts of building ODSs and DWs. Without a doubt, there are a multitude of design tools, databases, construction tools, management tools and tools to access the data warehouse and data marts. Over the past decade, the vendors in this area have been very insightful and prolific in creating solid tools to enhance your productivity. Today we have much to choose from in terms of ETL, meta data, access, administration and underlying infrastructure all aimed at data warehousing.
What about the ODS? Since it is relatively new (the first book published on ODSs was Building the Operational Data Store published in 1996), one cannot expect the technology to be quite as mature as it is for the DW environment. We are seeing the start of some very promising technology in the ODS area. Message brokering and middleware products are showing great promise. However, if the ODS is your choice, be prepared to do a lot by hand.
There are several well-documented commercial DW methodologies available today. These have been honed over the years through trial and error in the industry. The sheer tonnage of articles, books, white papers, etc., is almost overwhelming. Project plans are available, roles and responsibilities have been defined, critical success factors have been identified, example scope documents and scope document outlines are available (see our Web site for a scope document outline www.intelsols.com) and there are any number of conferences dedicated specifically to warehousing.
For the ODS, a methodology and project plan are available in the book just referenced as well as its second edition. However, there are no commercially available methodologies available today but it won't be long! Until then, you will be a pathfinder forging your own way, which is exciting to some and frightening to others.
Both the ODS and the DW are built in an iterative fashion a small piece at a time. However, the time frames for each individual iteration are significantly different. A piece of the strategic decision support environment (part of the DW and one data mart) can be built in a very short time frame. Your business community begins generating business intelligence almost immediately and, therefore, the ROI from this environment can be calculated and appreciated quickly.
Because each iteration in the ODS environment is usually large in scope, it takes anywhere from nine to eighteen months before the business community reaps benefit. Therefore, the enterprise must have greater patience (and faith!) that the resulting structure will show a positive ROI. Also, because the ODS takes longer, it will cost more than the first iteration of the DW, causing the ROI to take longer to be realized. However, the critical and tactical nature of the ODS may cause its ROI to be significantly larger than that for the DW.
Requirements for both these structures are rarely straightforward. Both require significant input from the business community and tremendous dedication in integrating the enterprise's data. The DW project, because a smaller scope can be carved out for it, may have an easier time getting a stable set of requirements from the users.
Another interesting phenomenon that will happen, I guarantee, is that you will generate most of the requirements for your first ODS iteration from the DW project! It seems that getting people to concentrate on integration and strategic decision making stimulates their thought processes and clarifies the needs for tactical decision making. In fact, I have found that we get requirements for an ODS that would not even be thought of had we not done the warehouse first. This one factor may tilt the decision in favor of building the data warehouse first.
The Corporate Information Factory is a proven architecture that can be built one piece at a time. The business community can immediately begin reaping benefit from either the strategic or tactical piece you build, giving you the political capital to continue constructing the full environment. This column gives you tips and topics to consider from a technological or project-oriented point of view in deciding which part of the CIF you build first the operational data store or the data warehouse. These should steer you in the right direction to properly satisfy your business community.
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
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access