Accelerating Skewed Workloads with Performance Multipliers in TurboDB Distributed Database

Explore a conference talk that delves into TurboDB, a groundbreaking distributed database system designed to tackle performance issues in skewed workloads. Learn how this innovative hybrid architecture combines a single-machine database within a distributed framework to achieve up to ten times better performance than existing solutions. Discover the two key designs introduced by TurboDB: Hybrid Concurrency Control, which ensures process-ordered serializability, and Phalanx Replication, which provides fault tolerance without compromising performance benefits. Gain insights into the implementation of TurboDB using CockroachDB and Cicada, and understand how it significantly enhances performance under skewed workloads compared to traditional distributed databases.

NSDI '24 - Accelerating Skewed Workloads With Performance Multipliers in the TurboDB Distributed...