CERN’s Future-Proof Global File Catalog

CERN is exploring scalable solutions for its AliEn global file catalog, a crucial component of their computing infrastructure needed to support the Large Ion Collider Experiment (ALICE) at the Large Hadron Collider. Currently reliant on a MySQL-based system with 12 billion entries, the infrastructure struggles with limited horizontal scalability, taking extensive time to dump and restore data. With future demands expected to increase fivefold, CERN is considering a shift to more scalable databases like Apache Cassandra and ScyllaDB, the latter of which demonstrated significantly higher throughput and resource utilization in tests, achieving 12 times the throughput of Cassandra and better resource saturation. Miguel Martinez Pedreira from CERN highlighted ScyllaDB's enhanced performance in load tests, showing promising results for a future-proof catalog, and expressed interest in continuing to explore ScyllaDB's capabilities.