Data Caching for Enterprise-Grade Petabyte-Scale OLAP

Explore a 20-minute conference talk from USENIX ATC '24 that delves into data caching solutions for enterprise-grade petabyte-scale OLAP systems. Learn about the challenges faced by organizations like Uber and Meta as they shift towards decoupling compute from storage in their massive data platforms. Discover how the Alluxio local (edge) cache addresses operational issues such as read-heavy I/O traffic, potential throttling, and skewed data access patterns. Gain insights into the implementation and effectiveness of this embeddable cache, which leverages local SSD resources to improve data transfer efficiency in petabyte-scale analytics environments. Understand how this solution integrates with OLAP systems like Presto and storage services like HDFS, and hear about real-world deployment experiences at Uber and Meta over a three-year period. Acquire valuable perspectives on managing modern, massive-scale OLAP workloads and optimizing performance in enterprise-grade data analytics environments.