FREE DM Review Site Registration!
Sign-up today and access DM Review on the Web!

Your FREE registration entitles you to:
FREE email newsletters

FREE access to all DM Review content

FREE access to web seminars, resource portals, our white paper library and more!

   

OLAP Schmolap

Reality IT

At my job, we have bought an OLAP tool in order to facilitate some of our business intelligence activities. This is nice, as I like OLAP a lot. Really, I mean it.

OLAP is the abbreviation for online analytical processing, as opposed to the operational OLTP - online transactional processing. OLAP had its commercial breakthrough in the BI market some 10 years ago, just after the breakthrough of the ad hoc query and reporting tools. Why? Well, one reason would be that ad hoc querying is all very nice but often tends to have long response times. Enter OLAP with its multidimensional way of storing and handling data. This allows for short response times that, in turn, encourage further in-depth analyses. With OLAP, you can for example quickly answer questions such as, "Who were our top 10 customers buying chain saws in Colombia last month and did they buy more or less as compared to the month before?" Very useful indeed. You could, of course, answer this with your old-fashioned ad hoc query tool as well but often only if you had the patience to wait for the SQL query to take its time (which it sometimes very much tended to do).

Oh yeah, OLAP has really established itself as one of the pillars of a modern BI package. And it comes in all types and flavors. OLAP is not just OLAP, but can be further specified as ROLAP, MOLAP, HOLAP, LOLAP, DOLAP or WOLAP. And they all mean something! Not that the end user cares, but IT might, as the different abbreviations all have some influence on the performance for the OLAP solution.

ROLAP, for example, stands for relational OLAP, which in practice means that the underlying physical database is a classical relational database that can be accessed via SQL. MOLAP - multidimensional OLAP - is a multidimensional database that uses a propriety format for handling its data as there is no real standard on how to access data in multidimensional databases like there is SQL for relational databases. Then we have got HOLAP, which stands for hybrid OLAP. This simply means that the vendor is trying to get the best out of the advantages with ROLAP (especially its ability to handle large amounts of data) and MOLAP (its quick response times in particular).

With HOLAP, mixing the best of ROLAP and MOLAP, it is of course important to avoid the "Pizzolate" effect. I like pizza and I also like chocolate. However, I never mix it and put chocolate on my pizza, getting something best described as Pizzolate. Few restaurants have this on their menu. In other words, it is important that HOLAP solutions really succeed in getting the right mix of ROLAP and MOLAP.

We also have LOLAP, local OLAP and DOLAP, desktop OLAP. This is the same thing and it means that there is data processing done on the client desktop (or laptop). And then we have got WOLAP, which is Web OLAP, i.e., a Web interface for accessing the OLAP application.

Among all these different OLAPs, I am surprised that I did not see eOLAP or iOLAP as well during the dot-com boom. Or myOLAP or xOLAP or ODOLAP - on-demand OLAP. It will probably come, given the existing OLAP jungle.

Rich as it is in architectural possibilities, how is OLAP actually defined in all this architectural OLAP schmolap? Like its different flavors obviously, i.e., in many different ways. The classical definition is the one by the late Codd, who came up with the 12 rules of OLAP (with another six added later on). Had Codd not done this for an OLAP vendor, these rules would not have been so controversial. But he did and, therefore, these rules have by default not been really accepted. This left the field open to others to make their own definitions of OLAP. And so they did abundantly. So well, in fact, that there is still no widely accepted technical standard on what OLAP is or should look like.

The amazing thing in all this is that most of the intended business users still have no idea what OLAP means. Or they simply do not care. And really, why should they care? The business users want to be able to answer the usual questions, which basically comes down to "Who bought/sold/did what, when and to/with whom?" From a pure business perspective, it matters less if the underlying database is multidimensional or not, as long as the business user is satisfied. It just so happens though, that OLAP is often a very good solution for providing quick answers and in-depth understanding to many business queries.

So, have you "olaped" your data lately? And if so, did you slice and dice the chain saw sales figures and drill further into the customer profile? At my job, I shall certainly enjoy myself with our OLAP tool, taking full advantage of the accompanying concepts of dimensions, hierarchies, drill down and slice and dice. Of course, the underlying data need to be properly cleansed, grouped and otherwise prepared, but once this is done, I shall "olap" my data until I drop!


Gabriel Fuchs is a senior consultant and business intelligence expert. His column Reality IT takes an ironic look at what real-world IT solutions often look like - for better or for worse. The ideas and thoughts expressed in this column are based on Fuchs' own personal experience and imagination and do not reflect the situation at any particular company. His book, Dealing with Nasty Colleagues: The Art of Winning in Office Politics While Still Getting the Job Done, can be ordered at www.amazon.co.uk. He can be reached at sgfuchs@bluewin.ch.

For more information on related topics, visit the following channels:



Industry Vendors