Kalen Delaney, course instructor and celebrated author of SQL Server 2012 Internals, will tell you all about how SQL Server 2014’s In-Memory OLTP works.
In-Memory OLTP (Codename “Hekaton”) is a completely new way of managing your SQL Server data, allowing incredible performance improvements. You can choose some or all of your critical tables to be stored completely in memory, all the time, so there is no cost for physical IO’s. In addition, these memory-optimized tables are completely lock-free and latch-free, providing additional performance benefits. Finally, you can choose to contain your application code in natively compiled procedures to avoid any query interpretation during execution, which adds additional cost savings. Current In-Memory OLTP users are seeing up to a 30-fold performance increase for their OLTP workloads.
The target audience for this class is administrators, developers or IT professionals who have experience with a recent version of SQL Server.
Session 1 – Creating a Memory-Optimized Table –
Kalen will show you just how easy it is to create a memory-optimized table, and what you need to do to allow a database to contain these special tables.
Session 2 – Rows and Indexes –
Tables and indexes are stored completed different for memory-optimized tables than they are for disk-based tables. In this session, Kalen will show you the physical storage mechanisms used to keep track of your data. In addition, she will describe the two new kinds of indexes used to organize your data: HASH indexes and RANGE indexes.
Session 03 – Data Operations –
Operations on memory-optimized tables are completely lock-free and latch-free. Kalen will show you how this is possible by describing what happens during data modification of rows in a memory-optimized table as SQL Server creates multiple versions of the rows that are updated. She will show you how your SELECT operations know which versions are relevant and what constructs in your SELECT statements are allowed. Finally, she will tell you how SQL Server manages memory as versions become obsolete and are no longer needed.
Session 04 – Transaction Processing and Concurrency Control –
In this session, Kalen will demonstrate how logging is much more efficient for operations on memory-optimized tables. In addition, she’ll describe the concurrency management for In-Memory OLTP and tell you which isolation levels are supported.
Session 05 – Checkpoint and Recovery –
As memory-optimized tables are completely memory-resident, users often wonder how data durability is possible. What happens when SQL Server is shut down? In this session, Kalen will show you how the checkpoint process writes enough information to disk to allow for data recovery, and how the recovery process allows re-creation of both your tables and indexes. In addition, this session will include information on managing the data files on disk, and some details on backing up your databases with memory-optimized tables
Session 06 – Native Compilation –
In memory storage with no physical reads, no locks and no latches can give big performance improvements, but to go all the way, you can also use natively compiled stored procedures to run your application code accessing memory-optimized tables. This session includes information on what’s required to create a natively compiled procedure, and a performance comparison performing the same operations using regular interpreted code.
Kalen Delaney has been working with SQL Server for over 27 years. She has provided performance consulting services and taught advanced courses on SQL Server to thousands of professionals and organizations. Delaney began her career as a teacher and specialist for Sybase, then in 1992, became an independent trainer and consultant. As such, she worked with both the Microsoft and Sybase companies to develop courses and to do internal training for their technical support staff. She became a Microsoft SQL Server MVP in 1993.
She is the author of SQL Server Internals: In-memory OLTP, primary author of SQL Server 2012 Internals, Inside SQL Server 2005 as well as previous books on SQL Server Internals. Visit her website to find out more.