November 2006 - Posts

Data Mining Addin for Excel

My experience with the term "add-in" was that you have a piece of software put on top of a bigger software, just to enable some little functionality. Today at SQL Pass 2006 I've seen how this term has to be redefined.

The Data Mining Add-ins for Office 2007 (CTP Nov 06 downloadable here) transforms your Excel 2007 and Visio 2007 products in full featured Data Mining clients. It requires Analysis Services 2005, but no work is required, the client can be used both to browse existing models or to create whole new models, starting from data contained in Excel or accessible from Data Sources defined into Analysis Services 2005 databases. One limitation of the current CTP (Nov. 06) is that it requires the Beta 2 Technical Refresh version of Excel 2007; a new add-in will be released shortly with a full support for Office 2007 RTM.

This "add-in" changes the accessibility of the Data Mining features of Analysis Services 2005. Yesterday Microsoft only had the server, today Microsoft has also the client, that fully leverages the power of Excel 2007. Guys, the demo was really impressive for the easy of use and the power that this add-in gives to end-user, not to mathematicians.

If you attend SQL Pass 2006

I am here in Seattle attending SQL Pass 2006.

If you are around, feel free to stop me and say hi: I really like to meet blog readers and people who read my many-to-many paper.

Let's contact me if you want to set a meeting place.

The vardecimal type for the fact table

I still hadn't time to install the SP2 CTP but I just read this post and I think it's a very interesting thing! The VARDECIMAL could be an amazing feature!

I encountered several times huges fact tables that have many NUMERIC measures. If the VARDECIMAL is optimized enough, it would save a lot of space. I will try on some real-world scenario after SQL Pass, but please if someone do the same test before, let's contact me!

SQL Server 2005 SP2 - CTP available

Here is available the SP2 CTP for SQL Server 2005.

The what's new document explain that there is an improvement in many-to-many dimension relationships calculations. I will try to evaluate the performance improvement, but unfortunately I'm in the middle of a busy timeframe (next week I will be in Seattle for SQL Pass - if someone want to exchange some opinion face-to-face here, let's contact me directly).

If someone get hands-on experience on many-to-many performance changes, please let me know.