Unlike what one may find in the entity relational world (e.g., ERD modeling), there is no such thing as an enterprise model, or one-size-fits-all data cube, such as what may exist in the two-dimensional world of relational theory.  An OLAP cube will have many levels of dimensions and data points; furthermore there may be tremendous variance of data frequency, calculations and data values at each level.  By nature, each OLAP cube will have a proprietary navigation path that is determined, in part, at load and pre-calculation stages.  A few more OLAP pointers:

• Your OLAP reporting tools should be able to handle both relational and multidimensional data sources.  Multidimensional OLAP (MOLAP) and ROLAP analyses should be supported, whether or not the data represents highly aggregated information or more granular transaction-oriented records.  (ROLAP is essentially relational OLAP and conforms to the SQL query approach of retrieving, calculating and storing data in predefined or ad-hoc tables or views.  A ROLAP architecture assumes interaction with a relational database and computes information on the fly, as opposed to leveraging pre-computed dimensions and data points, most commonly associated with MOLAP.) 
• The process control features of OLAP architecture must not be overlooked.  Capable data architects will understand the best way to implement the automated just-in-time distribution of multidimensional data sets to various user communities.  My preference is to utilize out-of-the-box scheduling components found in many OLAP toolsets (as opposed to existing - often homegrown - scheduling applications.)
• The best data modeling tools handle both relational and multidimensional data models.  In this day and age, no IT departments should be investing in “flat-earth” data and process modeling products that do not offer support for a wide range of information modeling techniques and methodologies.
• Any knowledge generated from or stored in an OLAP/multidimensional cube must easily integrate with Microsoft Excel.  Spreadsheet navigation of cube data has always been a key driver of OLAP success.  This will continue to be the case for the foreseeable future.

There are a multitude of OLAP architectures that offer a decent mix of performance, maintainability and scalability.  However, developing and maintaining an OLAP architecture that will best suit your company’s immediate reporting and business intelligence needs, while scaling far into the future, will prove to be a strenuous mission. Service-level agreements invariably will become strained as more users request and take delivery of reports. The growth of OLAP cube sizes will negatively affect service-levels agreements, as large chunks of multidimensional data are pushed and pulled across corporate networks and storage devices—the most common latency issues associated with business intelligence and KPI dashboards will be amplified.

Still, there are proven approaches to combating cube latency and bloat, starting with smaller OLAP data cubes based on more specific, less complicated data models.  Such an approach has the potential to make multidimensional reporting and analysis more iterative and repeatable, resulting in a marked increase in data transparency and governance.  My general feeling is that vendors of OLAP reporting and business intelligence solutions should be doing a better job of educating their customers on OLAP best practices, including a frank discussion of the most common pitfalls and perils of OLAP, starting with data explosion.

About the Author

A Contributing Editor for Dashboard Insight, William Laurent is one of the world's leading experts in information strategy and governance. For more than 18 years, he has advised numerous companies and governments on technology strategy, methodologies, and best practices. William currently serves on the faculty of Baruch College, runs an independent consulting company that bears his name, and lectures frequently on various technology and business topics worldwide. He would enjoy your comments at wlaurent@williamlaurent.com

Copyright 2010 - Dashboard Insight - All Rights Reserved.