Unlocking Efficiency: Sharded Synchronization in High-Frequency Write-Heavy Systems

Abstract

In multi-threaded environments with frequent write operations, traditional synchronization methods often encounter performance bottlenecks, which limit scalability and efficiency. This talk explores sharding, a synchronization strategy designed to allow more parallel processing by implementing finer-grained locking.

The presentation will cover:

• The practical implementation of sharding in C++, with real-world examples.
• A comparative analysis of sharding versus traditional mutexes, shared mutexes, and lock-free container implementations.
• Key insights into how sharding fits into the broader landscape of synchronization techniques, empowering attendees to make informed decisions for their systems.

By the end of the session, participants will gain a clear understanding of when and how to use sharding to optimize performance in write-heavy, multi-threaded systems.

Aleksander’s expertise in designing scalable, high-performance systems makes him uniquely qualified to discuss this topic. His ability to translate complex technical ideas into actionable insights ensures that the audience will leave with valuable takeaways applicable to their work in DevOps.