Blogging from the PASS Summit : Nov. 7th keynote
Bill Graziano takes the stage at 8:15. He talks about how 3,894 attendees (and 5,611 total registrants) represent 57 countries at the 14th summit. There are over 127,000 members worldwide.
Note that you can watch the keynotes and many sessions through Pass TV.
PASS serves SQL Server community – expertise, support, commitment. He talks about SQLSaturdays, SQL Rally, 24 Hours of PASS, and the Summit. He announces that there will be a third annual SQL Rally Nordic event next November, and that there will have been 543,000 hours of training delivered to the community in 2012.
PASS is delivering its first Business Analytics Conference, April 10-12, 2013, in Chicago.
A lot of people think PASS Summit is about a bunch of sessions. There are also other resources here, such as the SQL Server Clinic (4C3), Developer Chalk Talks (4C4), Hands-On Labs (304), Focus Groups, a Solutions Theater, and even on-site MS Certifications. And don't forget about all of the social interaction and conversations that happen outside of these formal settings.
SQL Server 2012 Service Pack 1
Ted Kummert takes the stage. He talks about SQL Server 2012 and announces the general availability of Service Pack 1. (You can download SP1 here – it's build # 11.0.3000.) There is already a "What's New" article on MSDN.
Hekaton – In-Memory for OLTP
He finally let the cat out of the bag about Hekaton (greek for "100x" – implying 100x performance improvement), an in-memory transactional engine that will ship in the next major version of SQL Server. We've known about this technology for at least 18 months, but were sworn to secrecy (though there were a few slips, e.g. on Connect and this blog post). Shawn Bice demonstrates a tool that identifies tables that are great candidates for in-memory optimization (trading CPU for latches), and helps you migrate them to in-memory tables with no changes to applications.
Anyone remember <code>DBCC PINTABLE</code>? This is NOT that. Paul Randal liked <code>DBCC PINTABLE</code> about as much as he likes shrinking databases.
The tool will also identify stored procedures that can be memory-optimized – essentially re-compiling the procedure so that it runs natively in memory. The big thing here is there is no re-learning and no re-writing of application code to take advantage of this in-memory technology. Very exciting stuff.
Shawn also shows a demo of ColumnStore (which they're going out of their way to label "xVelocity ColumnStore"), which we've all seen, but he talks about two great enhancements in SQL Server vNext: ColumnStore being updatable (finally!), and the clustered index on a table being eligible for ColumnStore.
My take: If you weren't excited about SQL Server 2012, you should certainly be excited about the next version of SQL Server.
SQL Server 2012 Parallel Data Warehouse
The new version, scheduled for H1 2013, will use Windows Server 2012 (Storage Spaces), updateable and clustered xVelocity Columnstore indexes, an updated distributed query processor, and offers totally redundant appliances with up to 40 nodes.
Christian Kleinerman comes up and demonstrates the new admin console and "insanely fast" queries. He does a COUNT(*) query over 294 billion rows (over 1 Petabyte) sub-second, and a much more complex query using the ColumnStore clustered index over the entire table returns in less than 2 seconds. Space savings with ColumnStore clustered index is 5x-15x.
New technology allowing users to query over both relational and Hadoop data through PDW. New syntax: <code>CREATE EXTERNAL TABLE … WITH (LOCATION = 'hdfs://…/whatever.tbl');</code> Then you can join against this Hadoop source just like you would any other table. More info at http://gsl.azurewebsites.net/Projects/Polybase.aspx
Amir Netz comes up and talks about movies. Shocking. He demonstrates the ease of using PowerView, PowerPivot and insightful visualization within Excel. The first demo to get applause was turning the data into a map – a scrolling, zoomable map, with pie charts, tooltips and legends – again entirely inside Excel.
Ted's high-level blog post about these announcements:
Also see Jen Stirrup's great overview post here: