I have been away from my blog for a while. Although, I try to write something every now and then, I just couldn’t bring my self to write anything this past month. I am hoping I will get into regular writing habit from now on.
Anyhow, you all must have heard about the big news this week about SQL Server. SQL Server 2014 was announced at Microsoft TechEd 2013 keynote. I am super excited and I am sure you all will be after looking at the new features. I am just about getting the hang of SQL Server 2012 and we already have new and exciting stuff to look forward to in SQL Server 2014. Things are really moving fast. SQL Server 2014 will introduce a number of new features. Oh!! and did I mention how relieved I was to see that nothing major has changed in terms of SSAS, SSIS and SSRS!!
At the end of the post you will find links to some good blogs and whitepapers about SQL Server 2014. For now, I want to quickly note features which I am excited about.
Project Hekaton Or SQL Server In-Memory OLTP
Microsoft is introducing In-Memory processing with SQL Server 2014. If I understand correctly, what this entails is keeping high frequency or most accessed tables in memory thereby saving the trip to the disk. The new mechanism also saves enormous amount of effort SQL Server has to spend on locking and latching. Part of SQL Server relational engine have been changed to take into consideration the In-Memory tables. For e.g. Query Optimiser can now take into consideration that the table is now completely in memory and can choose a plan based on this cost rather than making an explicit assumption that all the data is on disk.
There are some restrictions though like you cannot have row size bigger than 8060 Bytes, foreign key constraints are not supported, LOB data types are not supported, the newly introduced natively compiled stored procedures can only access In-Memory tables etc. Despite these, I think, this feature will be quite powerful. Not to mention as it gets new improvements over next version on SQL Server this feature can be really useful.
Although, In-Memory is being touted as OLTP feature, I think it will be useful in DW/ETL scenarios as well. I guess we would have to wait and watch how we can use it.
Updatable Columnstore Indexes
One of the growing the complaints with SQL Server 2012 Columnstore Indexes have been that they cannot be updated. They need to be rebuilt after the data in the underlying table has changed. Well no more!! SQL Server 2014 introduces updatable columnstore indexes.
This is probably going to be the best feature related to BI in SQL Server 2014. Updatable columnstore index with Tabular model, direct query in tabular model or ROLAP in multidimensional model can be the killer combination.
SSD as Cache Store
With SQL Server 2014, you can specify SSD as an extension to memory. This way SQL Server will cache frequently accessed data. Since SSD’s are way faster than disks, the performance will be improved. It’s almost as if the SSD will act as added buffer pool memory for your SQL Server. I think the In-memory and SSD feature will go hand in hand.
Online indexing at partition level and Partition level stats
As the name suggest index can now be built online at partition level and stats are maintained at partition level. I suppose the second part would lead to better execution plan and the first one to lesser downtime when indexing a large table.
Some of the other improvements include
– Unlike SQL Server 2012, SQL Server 2014 supports 8 secondaries in Always On.
– There are some improvements/changes in Resource Governor which, to be honest, I didn’t really grasp
– SQL Server backups can directly be taken to Azure data store and SQL Server files can also be directly stored in Azure storage.
Please find below links to some of the interesting blogs and whitepapers on SQL Server 2014.
SQL Server 2014 Home Page : You can sign up here to get notified when the trial edition is available.
Geek City: SQL Server 2014 In-Memory OLTP (“Hekaton”) Whitepaper: This blog has link to the Hekaton whitepaper. It’s an excellant read to get an idea about how in-memory is going to work. I will highly recommend it.