Continue in 2 seconds

Dirty Data, Dumb Decisions

Published
  • March 01 1998, 1:00am EST

DM Review readers should be a lot more aware of the perils of dirty data than most IT staffers, thanks to Larry English's columns. However, as I listen to vendors hawk the latest OLAP tools, I feel a chill come over me. Remember the "rounding error" that plagued one of the early versions of Lotus 1- 2-3--and the resultant "off by a penny" problems that had accountants and bookkeepers pulling their hair out?

What I remember about the debacle isn't the temporary setback the snafu gave spreadsheet technology in general, it was the slew of accompanying articles that illustrated how easy it was for end users to get wrong results. Formulas were "hidden" behind the answers or, worse yet, embedded in some hard to figure out macro. The message was clear: use the tools, but use your head as well. When I taught novices how to use Lotus back in those days, I told them they should do "manual" calculations on some sample data to make sure the answers the spreadsheet returned were correct.

The same goes for today's OLAP (on-line analytical processing) tools.

OLAP is hot, because more and more users are familiar with the idea of slicing and dicing data, of performing roll ups and drill downs.

According to Nigel Pendse who, with Richard Creeth, publishes The OLAP Report (www.olapreport.com), there are a good 30 vendors in the space. One might think this plethora of choices would discourage newcomers, but that's not true. Joining the market "leaders" (Oracle, Hyperion, Comshare, Cognos and Arbor) for example, are WhiteLight (www.whitelight.com) with a new "rapid OLAP" server, IBM (www.software.ibm.com/data) who was expected to be reselling a version of Arbor Essbase beginning last month, Influence Software (www.influence.com) whose Aperio delivers OLAP to Java-enabled browsers and Microsoft.

Yes, Microsoft. The software giant's OLAP server product (code-named Plato, based on technology it acquired from Israeli-based Panorama systems) isn't expected to be released commercially until mid 1998--in the same time frame as SQL Server 7.0. In the meantime, however, Microsoft is building a team and working with team members to set up the infrastructure, notably OLE DB for OLAP (code-name Tensor), Microsoft's DCOM-based "universal data access" specification (www.microsoft.com/data).

It's not surprising, then, that the current players are jockeying for position, market share and mind share. Arbor should have shipped its new Essbase 5 OLAP server by the time you read this. In December, Seagate Software (www.seagatesoftware.com) added OLAP functionality to the latest version of its Crystal Info 6 reporting and business intelligence engine. Seagate also announced that Microsoft would be bundling a special version of Crystal Info with Windows NT 4.0 and higher. Brio Technology (www.brio.com) updated its Brio Enterprise Server OLAP package with new versions of both its traditional "fat client" BrioQuery tools (available for Mac and Windows 3.1 or higher) and its Brio.Insight and Brio.Quickview browser plug-ins. In January, MicroStrategy updated its ROLAP line with a Web-server version (DSS Web 5.0, which still requires both DSS Server and at least one copy of DSS Architect) and revamped end-user tools that feature more wizards. Business Objects 4.1 added an OLAP module, too, and several new Java-based client tools, such as dbProbe from Internetivity (www.internetivity.com) joined the fray as well. OLAP market leader Cognos (www.cognos.com) has reinforced its position by moving into the server arena with NT and UNIX server versions of PowerPlay 5.0, and Oracle (www.oracle.com) continues to make its powerful Oracle Express Server family both better integrated with other Oracle products and easier to work with. For example, Oracle now bundles a Relational Access Manager (RAM) ROLAP tool with Oracle Express at no additional cost. RAM has two components: a Relational Access Manager Server (the ROLAP "engine") and a Relational Access Administrator (the GUI tool used to set up a ROLAP connection and map meta data).

Can vendors include "wizards" that warn users of potentially non-sensical operations or results? Although we probably won't see that kind of built-in smarts for a while, vendors can simplify life for us by providing industry "templates" and best practices that encapsulate the experiences of others with some vertical market smarts.

In addition to the top-notch OLAP Report site, I recommend you read Cognos's 130-page Multidimensional Manager (you can request your free copy of this fantastic little gem from www.cognos.com). Another good resource is Erik Thomsen's OLAP Solutions: Building Multidimensional Information Systems from Wiley (ISBN 0-471-14931-4).

I see OLAP and data warehousing moving forward in lockstep this year, with a growing amount of analysis being done one way or another over the Web.

My message is this: In addition to the well-known security issues associated with data out in cyberspace, plan to educate your end users about the potential dangers of performing wrong-headed analyses. It's not just dirty data that can cause problems.

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