The future of SQL Server: Project Kilimanjaro, Madison and Gemini

The Microsoft BI Conference 2008 has been the source for quite a few announcements. First of all, now that DATAllegro was purchased by Microsoft, the talk is about SQL Server handling bigger amounts of data. There have been posts not just about terabyte projects, but also petabyte projects. The full integration of DATAllegro's technology into SQL Server is planned for the first half of calendar year 2010 and is code named Project Madison. SQL Server 2010 itself is code named Project Kilimanjaro.

The real surprise? Project Gemini which sounds like Microsoft giving QlickView a run for its money... but I'm ahead of myself. Project Gemini is the code name for Analysis Services using column-oriented and in-memory technologies running inside Excel. This means faster access for all applications working against cubes stored in this brand new way. Microsoft is hoping to empower the end - users even more, as now they will be able to build models by themselves.

I think this is marvelous news and a brilliant move on the part of Microsoft (not that you would expect anything less...). First of all, I saw a presentation of QlickView a while ago. QlickView offers instant, in memory, manipulation of massive datasets. The demo I saw was very impressive. Thing is, for the good and the bad of it, QlickView works against tables rather than cubes and I don't know how much that may limit your capabilities at defining calculations (I would have to stress I haven't tried and can only assume). But in any case, QlickView looks very impressive visually and the response time we saw was excellent, which is also the reason why its share in the BI market is growing rapidly. The QlickView representative who came to give us the presentation said he believes in memory storage is the future and I guess now I see just how right he was... Microsoft has taken QlickView's idea of in memory storage one step further, in the fact that for them it's going to be for the cube itself. This means that you should still be able to use MDX calculations against your cube.

Secondly, this also puts an emphasis on the capabilities of Excel and SharePoint and the way they work with BI. Microsoft is pushing the capabilities of Excel even further, after making Excel 2007 a better cube viewer and enhancing its Data Mining capabilities. This is on top of empowering Excel itself and turning it into Excel Server. Now, Excel should become the staging area for power users to create their own cube models and share them, after publishing them to SharePoint. So in that sense I feel like Microsoft is also leveraging its capabilities in work portals and data collaboration around the organization with Excel and SharePoint. All in all, very exciting news.

For more details about these projects, please refer to the recent posts from Chris Webb, Mosha Pasumansky and Marco Russo.