This presentation will explore both how the performance problem came about in the first place and ways to search for solutions. It is illustrated with examples where several different strategies are used.
How do you make a T-SQL program run a lot faster? Like 100 times faster? It’s difficult. Microsoft tried and got close in the Hekaton project, its in-memory tables feature of SQL 2014. Inspired by that effort, this presentation looks at recent cases where I’ve been able to achieve huge performance improvements in SQL Server using various techniques. We’ll explore how the performance problem was created in the first place and the solution that I eventually found.
The first thing to think about is, "How can I accomplish this task some other way". Sometimes the other way is a better algorithm. Sometimes a different technology is more appropriate to the task. Sometimes, as with Hekaton, it’s a combination of many changes that combine to create the desired outcome much, much faster.
Andy Novick is a SQL Server Developer in the Boston area with 25 years of database and application development experience. His consulting practice focuses on building applications, including software products that use SQL Server's capabilities to the maximum. He has developed both OLTP and data warehouse databases with multi-terabytes scale. He has particular expertise in automating data management for such large databases. Other recent projects have included ETL, Security and SQL Injection prevention. Andy is the originator of the popular "SQL Server Loadfest" event held in Waltham, MA.
Topics: Database Performance, Database Administration, Database Diagnostics Products: