Conquering Highly Concurrent Workloads

This lecture covers techniques for optimizing data layout and query execution in highly concurrent and selective workloads, focusing on physical design decisions and sharing-aware execution. It explores the challenges of achieving low latency and interactivity in analytical queries, proposing a co-optimization approach. The presentation discusses the importance of materialized views, dynamic programming, and simulated annealing in view selection. Additionally, it delves into partitioning strategies for different workloads and tasks, emphasizing load-aware techniques and adaptive partitioning decisions enabled by online learning.