Continue in 2 seconds

Architecture and Technology

Published
  • April 01 2003, 1:00am EST

A data warehouse is an architectural construct, not a technology. A data warehouse contains integrated, historical detailed data that is available to a wide set of sources that need to access and analyze the data. A data warehouse is built on different technologies. To build a data warehouse you need an operating system, a database management system (DBMS) and one or more tools for accessing the data.

In addition, you probably need an extract, transform and load (ETL) tool, a data-cleansing tool, multiple access and analysis tools, and perhaps a data monitor. If your warehouse is large enough, you need some form of alternate storage and a cross-media storage manager. However, in spite of all the technologies required to build a data warehouse, it is still an architectural component, not a technology.

There are several implications to this fact. The first is that you cannot go to the store and buy a data warehouse. There are no vendors that sell data warehouses. At the very least, you must string together several technologies and deal with several vendors; this can be tedious.

Yet this is not all bad news. Because a data warehouse is an architectural component and not a technology, it means that you have choices as a consumer when it comes to technology. If you don't like one hardware vendor, you can select another. If you don't like one software vendor, you can select another. This ability to choose outweighs the hassle of dealing with several vendors.

Why does the world even find the idea of a single vendor appealing? Have we forgotten the perils of proprietary technology? There once was a day when many organizations were held captive by their vendors; the vendors had proprietary technologies and they treated their customers as if they were hostages. This hostage mentality showed up in many ways, including:

  • Pricing policies, especially maintenance rates.
  • New features and other product advances – you got what the vendor felt like building, not what you needed or wanted. Worse yet, you were limited by the vision of the DBMS vendor.
  • Responsiveness to bugs – the DBMS vendors were responsive to crisis situations when and how it suited them.

Anyone wishing to return to a position of dependence on a single DBMS vendor either has a very short memory or simply is unaware of past practices. In short, it is good news that data warehousing is supported by multiple vendors' technologies.
Another implication of data warehousing being an architecture and not a technology is that new product features and advances are not limited by a single vendor's vision of the way data warehousing should proceed. In order to understand the importance of this distinction, consider the history of data warehousing. The concept of a data warehouse was absolutely contrary to the theory on which modern database management systems operated. There once was a day when DBMS vendors only supported transaction processing. When data warehousing made its appearance, the DBMS vendors uniformly trashed the concept. The fact that data warehousing fit on their technologies was a coincidence and even an embarrassment to some of them. It took the demands of their customer base to drive the DBMS vendors to support data warehouses. DBMS vendors and their technologies have been dragged into the world of data warehousing. So is it a surprise that DBMS vendors do not readily support the needs of decision support system processing today? Their vision simply does not extend beyond the boundaries of their existing technology. For example, ETL processing is a natural addition to DBMS technology. Did ETL technology begin with the DBMS vendors? No. As another example, near-line storage (data stored apart from disk storage) is a natural extension of DBMS technology into the world of data warehousing. Did the DBMS vendors pioneer near-line storage? No. As yet another example, OLAP technology is a form of technology that very nicely complements basic DBMS technology. Did the DBMS vendors pioneer OLAP technology? No. A final example is monitoring the usage of data in the warehouse. Did DBMS vendors lead the charge here? Of course they did not. Their reaction was – why would anyone want to do that?

In a word, the DBMS vendors got stuck in a mental time warp. They were successful in an earlier day and simply wanted to stop their technology at their point of maximum success. Any further advances to their technology were made outside the walls of their development staffs. They wanted their world to stop developing once they reached a point of critical mass. All of which means that many technologies today are required to support the architectural construct known as the data warehouse.

Data warehousing is a real marketplace. It is proven. It has a real future, real products, real value propositions and real companies.

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