I am thrilled! Yesterday at TechEd, Microsoft announced that somewhere in 2014 the new SQL Server version will be released. WOW! So soon? We are working with the 2012 versions like a year, year and a half? But trust me, the 2014 release will have some real neat things! They announced that somewhere around the upcomming Build Event (right here) the trial version of SQL Server 2014 will be released. If you want to know what is in it, look at this link : http://www.microsoft.com/en-us/sqlserver/sql-server-2014.aspx. A small recap of what you can read in those papers is right here!
“Hekaton”
One of the coolest things, if you ask me, is the part that goes by the codename “Hekaton”. This is an in-memory database where you can get a performance boost of about 10 to 50 times. The company Bwin did a very succesfull implementation. For that succes story look at this video on youtube : http://www.youtube.com/watch?v=nO200qJ_i-Y. As you can see in that movie they got from 15.000 request/sec to 250.000 request/sec. That’s like.. WOW! And that all thanks to some real nice optimizations. T-SQL code for example, is compiled to native machine code, which is quite faster to execute. Hotspots in your data are loaded into fast memory optimized structures. It is also highly scalable where some locks and latched do NOT exist anymore. That’s a little against the grain for me, but for a solution like this you must leave those things out for performance reasons.
If you want to know more about hekaton, look right here : http://research.microsoft.com/en-us/news/features/hekaton-122012.aspx
More hybrid functionality towards Azure
A lot of things are introduced towards the cloud solution, Microsofts Azure.
- In SQL Server 2012 we have got the AlwaysOn functionality, which was quite cool if you are used to the old school 2005/2008 mirroring functionality. In SQL Server 2014 you are able to use a secondary replica in a Azure VM! With that in mind, think about reporting capabilities, HA-DR, etc, etc.
- Deployment to the cloud should be easier now.
- Cloud storage. Use the clouds storage solution on-premise. So your data and/or log files will reside on the Azure storage, while your RDBMS is on-premise. That’s real cool, but then i still think about io latencies. We have to wait for the trial versions to know exactly what this means.
- Backup scenario’s to the cloud. You can backup and restore directly into a Windows Azure Blob
Performance
Besides the above mentioned “Hekaton”, there are a lot of performance related new things.
- Improvement of the Buffer pool. More and more servers are given SSD’s instead of the spindles. Since these storage devices can handle a lot more IOPS, it is quite evident that memory pressure shouldn’t be an issue. With the improvement of the Buffer Pool we are save now.
- Parallel bulk loading of data. With the improvement of the SELECT INTO command we now can insert data parallel. Finally!
- Improvements to the cardinality estimator and statistics.
Conclusion
Well, as you can see, a lot of new things are coming our way somewhere soon. As soon as the trials are available we will know a lot more. But only the performance boosts, Hekaton, new/improved BI functionality and of course more cloud is making me really looking towards the end of June for the first preview.
