Many announcements were made for the future release of SQL Server , and in particular:
- Hekaton project unleashed. This is an in-memory storage engine designed for highly-demanding OLTP systems. Removing the need for page latching will improve the performance of existing T-SQL by orders of magnitude (Hekaton is greek for 100x) without modifying the code. Although the secret had leaked previously (see Aaron Bertrand’s post here) and we were expecting a big performance bump, the demos were very impressive.
- Updatable and clustered ColumnStore indexes: the readonly constraint when adding a columnstore index to a table was a big pain, compared to other vendor’s similar technology (I can only think of Sybase’s IQ engine which has been around for about 20 years). Finally the next release of SQL Server will allow read/write. I really can’t wait to see the technical / performance implications on updating such structures.
- PDW v2: new features like updateable columnstore indexes will also be added to the next release of PDW. Insane query execution times showed at the demo against 1 Pb of data ! We have a session on PDW v2 coming up in december at the french SQLRally called ‘Journées SQL’ , in case you’re in Paris you can already register right here. Among other features, the Polybase engine allowing to query both relational and Hadoop data from SSMS. A demo of this was showed during Keynote 2.
- SQL Server 2012 Service Pack 1 available: again an annoucement made at the keynote 1 by Ted Kummert.
In my humble opinion, the first keynote was much better than the second one. During the Polybase demo, multiple bad comments flooded out on twitter . Ironically the subject of the demo was about Twitter and Sentiments, by mining positive and negative feedback from twitter logs. You may read BrentO comments about this here.
As far as I’m concerned, I must admit that I can’t really appreciate what it takes to build a demo like this, but I found that it took a long time to get to the climax (at least what I considered to be the climax) when Julie added a SQL Server replica on Azure at the end of the session. Hopefully, just after the keynote, I watched Adam Machanic‘s session on paralellism and THAT WAS A MOMENT. The greatest session on the live feed so far. I’m still collecting a lot of facts about paralellism, serial zones, MAXOP and CTFP, and this helped a lot. Thanks Adam !
Today don’t miss Dr DeWitt’s session on Big Data, check out the schedule at SQLPASSTV.