Replicas as a Strength, Not a Burden - Scalable Consensus for OLAP Engines with LSM Storage

Learn how to leverage database replicas as a performance advantage rather than overhead in this conference talk exploring scalable consensus mechanisms for OLAP engines using LSM storage architecture. Discover innovative approaches to distributed database design where replicas enhance system capabilities instead of creating bottlenecks, with specific focus on implementation strategies for analytical processing workloads. Explore the intersection of consensus algorithms and LSM tree storage systems, understanding how to architect distributed OLAP engines that scale effectively across multiple replicas. Gain insights into practical solutions for managing distributed state in analytical databases, with real-world examples and technical deep-dives into consensus protocol optimization for storage-heavy analytical workloads.