Revolutionizing Database Performance: Deep Dive into SQL InMemory Technology
Regular 50 minute session for SQLBits 2026Thursday - 23 Apr 2026 - 09:00 - 09:50 Room 1BTL; DR
SQL Server still slow after tuning? Learn when In-Memory OLTP truly fixes concurrency and throughput problems, when it doesn’t, and how to avoid costly mistakes through real demos and engine-level insight.
Session Details
You’ve tuned queries, added indexes, and upgraded hardware, yet SQL Server is still struggling under load. In-Memory OLTP promises massive gains, but many teams adopt it blindly and see little benefit or worse, new problems.
In this session, we’ll cut through the hype and focus on when In-Memory OLTP genuinely solves performance bottlenecks, and when it absolutely does not. Using real-world scenarios and live demos, you’ll see how In-Memory OLTP behaves under concurrency, what architectural patterns benefit most, and which pitfalls can make performance worse instead of better.
It’s a decision-making guide for DBAs and engineers who need predictable results.
In this session, we’ll cut through the hype and focus on when In-Memory OLTP genuinely solves performance bottlenecks, and when it absolutely does not. Using real-world scenarios and live demos, you’ll see how In-Memory OLTP behaves under concurrency, what architectural patterns benefit most, and which pitfalls can make performance worse instead of better.
It’s a decision-making guide for DBAs and engineers who need predictable results.
3 things you'll get out of this session
Identify workloads where In-Memory OLTP genuinely improves concurrency and throughput
Understand how SQL Server’s in-memory concurrency and transaction model affects performance
Avoid common design and adoption pitfalls that lead to poor or unpredictable results
Understand how SQL Server’s in-memory concurrency and transaction model affects performance
Avoid common design and adoption pitfalls that lead to poor or unpredictable results
Speakers
Thodoris Katsimanis's other proposed sessions for 2026
Synchronous Availability Groups Don’t Always Mean Zero Data Loss - 2026