Is the rise and fall of OLAP really the end of it?

I recently saw a blog post by Cedric Chin on the Holistics Blog titled The Rise and Fall of the OLAP Cube. In the post Mr. Chin gives a good description of what OLAP is and how it arose. It is a good read for those who want to learm more about OLAP.  Mr. Chin then argues that with new technology, cheap memory, columnar databases and  parallel processing OLAP cubes are dying. While for some, I might agree, but what I believe is missing from the article is the reason OLAP cubes appeared in the first place.

First, Analysts wanted to get away from depending on IT to provide data for them. It was a constant wait for IT resources to work on the backlog of data requests to provide reporting the business wanted. OLAP for the most part is business run and business centered.  We have seen IT departments shrinking and their load increase. Adding more for them to do brings us back to the 90s waiting on them for data.

Second, and what I feel is the more important point, is OLAP is easy. Mr. Chen talks about the technology. Parallel processing, Columnar databases, etc. I agree those things are great and will change the way data is stored, but, with that technology, currently there is a cost. That cost is having more technical skills. knowing SQL and how to use it properly. Do I really want to write a new SQL query if  I want a different slice of data?  Most users are non-technical. With OLAP, they ask a question and get a response without having to understand complex languages.  Too often in my history, have I seen bad SQL joins and poorly created queries that provide the wrong answers. Users of OLAP point and click  in Excel or BI and get the answers.

Third, OLAP empowers users to go beyond what the structured data gives. Creation of unique hierarchies and ragged hierarchies, layering in simplistic and complex formulas and procedural calculations add to the users ability to analyze the data. Add to that, the ability to bring multiple data sources together (be it multiple relational sources, or manual user input, or both) make most OLAP instances data agnostic. While all of this could be done in in a tied in data source, it would require the data movement that Mr. Chen argues against and in many cases would be more difficult to accomplish that combining them in the OLAP target.  

I am an Oracle centric consultant and know Oracle is working on integrating OLAP into their Autonomous Data Warehouse (ADW) so OLAP will sit on top of it. I'm sure other companies are doing the same. Does this mean OLAP is going away? I don't think so, it is just the net incarnation of it. OLAP will be around. Why? because for the user, is it easy.