First of all, I’d like to point to the fact that there is almost always a bottleneck with OLTP database’s performance that is related to disk speed. So in-memory processing can improve the current poor performance with disk-based tables significantly as well as achieving scalability.
In a very simple and quick explanation, Hekaton is new feature coming with SQL Server 2014 to handle in-memory processing for OLTPs. Actually it is something live xVelocity for OLAP. But, how Hekaton can help for improving OLTP performance and scalability? The answer is that SQL Server 2014 is trying to achieve these goals by using
- Optimised algorithms to access memory-resident data
- Eliminating logical locks using a concurrent control method for RDBMS (Relational Database Management Systems) that assumes that multiple transactions can complete without affecting each other, so that, the transactions can commit without locking the source data. This is called Optimistic Concurrency Control (OCC).
- Using lock-free objects to access all data, so threads that perform transactional work do not use locks for concurrency control.
- Using a new concept in SQL Server 2014 called Native Compilation. Native compilation refers converting programming constructs to native code, which consists of processor instructions that can be executed by the CPU without the need for further compilation or interpretation. So, SQL Server can natively compile stored procedures that access memory-optimized tables. Native compilation allows faster data access and more efficient query execution, than traditional, interpreted Transact-SQL.
Microsoft claims that “Use of main memory can result in performance gains from a few percentage points of performance improvement, to 20 times performance improvement.”