SQL Saturday april 2nd 2011 in Zeist (Netherlands)

I joined SQL Saturday on april 2nd in Zeist (Netherlands) and i would like to share some interesting stuff that i've learned in some of the sessions. There was a keynote by Gert Drapers, several sessions on SQL Server and sponsor sessions. I joined the following sessions:

• Keynote (Gert Drapers).
• SSIS in Denali (Peter ter Braake).
• BI in Denali (Marcel Westra).
• TopBI session (Marc Ducardus).
• Generating SSIS packages (Wolter Zigterman).

Keynote (Gert Drapers)
An interesting presentation! I've seen one presentation of Gert earlier on the first SQL saturday (i think) and now it was a better presentation than the earlier one. Perhaps it's because he presented now in English rather than in Dutch (he is residing about 30 years in the US). And he brought up some interesting stuff about sharding patterns and the CAP theorem (http://www.julianbrowne.com/) which i will investigate further in the future. Gert explained that the explosion of data will be a big issue for database builders like Microsoft and Oracle. Why has the Nosql movement such a big succes? Because it's very scalable and transactions are less important for the Nosql databases. Also event processing (push based) will be more important.

SSIS in Denali (Peter ter Braake)
SSIS in Denali was presented by Peter ter Braake and he showed some features of the CTP1 version of Denali. He showed "new for the developer" features, SSISDB Catalog and the dependency analyzer. To be honest it was all familiar stuff because i've investigated it already by myself. But, it was good to see that Peter did the same steps like i did. You can find more about SSIS in Denali on my blog.

BI in Denali (Marcel Westra)
In a fast speed all kind of features of Denali passed the session. Microsoft calls it "BI Pervasive insight" because BI is a broad and fuzzy definition. What is business intelligence? Good question. He also explained the different points of view of IT (control, scalable and robust) and the end user (creative, fast and agile).

Again, he stressed that Microsoft will support UDM and BISM in the future but if you build a new implementation of a analysis solution please use BISM. So in the end UDM will be less and less implemented.

Generating SSIS packages (Wolter Zigterman)
Wolter showed how you can generate SSIS packages. In contrast on what i believed on how to generate SSIS packages he didn't use the SSIS API and then generate the SSIS packages with C#.  No, he created SSIS template packages and adjust the XML with VB code and it showed very effective. The output of this process are generated SSIS packages. He uses a couple of xls sheets for defining the source and the datavault tables. On my blog here you can find some investigations myself on building SSIS packages with C#.  It is very difficult and a lot of work and time to create. So Wolters approach is very pragmatic and effective. On the other hand if a C# programmer takes the time and work in coorporation with a SSIS developer a profound solution can be developed.
  
An interesting remark of Wolter is that the problem of building SSIS packages with C# is that you need to set both design time and run time properties and this is very ineffective for generating dynamically SSIS packages. This is a very very fundamental issue. Perhaps you don't want to set design time properties when you want to generate SSIS packages? Only run time properties! Something to think about?!

So that's it and until next time..

Greetz,

Hennie