The keynote today was delivered by Ted Kummert, Corporate VP, Data and Storage Platform Division at Microsoft.
Before Ted started, Wayne Snyder showed up on stage, sitting on a hog and surrounded by smoke or dry ice or something. He talked about how PASS has grown over the past year in several areas:
He also talked about the launch of the new sqlpass.org web site, and the fact that membership is now free.
When Wayne was introducing him, he mentioned that Ted once worked at Apple, and that it was a big deal that Microsoft forgave him for that. It struck me as funny, because Apple is not the competitor that Microsoft should fear the most, IMHO.
Ted came on and gave kudos to PASS. He explained how product team members are here for us, and this is much easier to do obviously when the conference is here in Seattle.
How SQL Server 2008 Has "Delivered"
Then he started talking about how SQL Server 2008 has delivered on its four "pillars" in the company's vision for SQL Server and its surrounding technologies.
Enterprise Data Platform
- transparent data encryption
- policy-based management
- spatial / geography
- entity framework
Pervasive Insight – data warehouse scale
- query processor (star joins)
- data compression
- partition table parallelism
- change data capture
- resource governor
Ayad Shommout, Lead Technical DBA of CareGoup Healthcare Systems in Boston, talked about his environment (2000, 2005 and 2008). He explained that he intends to have all 400+ production databases on SQL Server 2008 in the short term, and was an early adopter (went live December 2007). By upgrading alone, without changing any code, they experienced a 25% performance increase. Top three features for Ayad:
Ted talked about ICE, an internal 30+TB data warehouse representing a central repository of all events from sources like Exchange, SQL Server, etc.
He also mentioned that the ad center team has what they believe to be two of the largest analysis services cubes in the world (1.8TB & 1.4TB).
Some of the highlights for SQL Server 2008:
For the future, he talked about three release vehicles that we will see before the next major release of SQL Server (expected in 2011). These are:
This is the Datallegro acquisition you may have read about; expected release H12010. There are significant advances in data warehouse capabilities and scale. Jesse Fountain, Program Manager, talked about "massively parallel processing" (MPP), where multiple instances look to the end user like a single instance, allowing the processing power to be distributed across several database servers (kind of sounded like RAC). The fact data was distributed across multiple instances, and dimension tables were replicated to all. This allows each node to compute its local fact data and then the results are merged at a higher level. The demo showed reports pulling data from a 150+TB data warehouse (over 1 trillion rows), utilizing 24 instances of SQL Server (8 cores, 192 total) on Dell 2950s with EMC2 CX4 and 10GB fiber.
SQL Server "Killimanjaro"
These are new manageability capabilities built around SQL Server, tentatively scheduled for H12010. Dan Jones talked about "down payments" in PBM and data collector that enabled the development of what he called the SQL Server "fabric" – a fabric control point in front of multiple environments, providing a bird's eye view of the environment that represents the fabric (servers/instances/applications). He introduced what they are calling the DAC (data application component, not to be confused with dedicated administrator connection), for developing, deploying and managing database applications. In the demo he showed a new "Fabric Explorer" view built into Management Studio, with a report showing instance / DAC health, over- and under-utilized servers (consolidation or load balancing candidates), and storage. The DAC itself is a new deployment package (similar to MSIs) to deploy schema through Visual Studio. It can be created by pointing to an existing database (2000 and up) via a wizard. Once you have created a DAC, you can import it into the fabric, and along with creating the schema and objects it validates that the fabric's policies are not being violated. His final point was that this architecture allows us to manage at scale, improve integration between the developer and the DBA, and make policy monitoring and enforcement more powerful.
Finally, he talked about "managed self-service for BI" with a pretty funny fairy tale. The moral is that IT can't meet all demands, and that end users should be able to make more decisions. So the intent is to empower them to ask more powerful questions from SQL Server using their existing tools (e.g. Excel). He mentioned Gemini, which includes analytics, reporting, and collaboration.
Donald Farmer, complete with angel wings and a magic wand, came to the stage to demo Gemini using Excel. Gemini is an add-in to Excel, and appears as its own ribbon. He mashed up video store data from the data warehouse (10 million rows!) and box office sales data he downloaded from the web, all on the client machine, and the performance of sorting and filtering was astounding. He mashed the data up, and demonstrated new "slicers" that act like visual filters to slice data up showing sales by geography, month, etc. The report he generated was published to SharePoint, and he showed off some of the "social services for data" features, such as voting and stats on the reports themselves. It seems very powerful for the end user.
SQL Data Services
Somewhere in there, Ted explained that the growing team has been working on "the cloud" for two years. He briefly mentioned SQL data services for application developers, and the new Azure Services Platform which was hyped at PDC. The major announcement here was that the public CTP of Microsoft SQL Data Services is now available – also see this link.
Next Version of SQL Server
Unfortunately, Ted was very vague about the next version of SQL Server, only stating that it should be delivered by August 2011.