How to Save on Scaling by Moving from Cassandra to Scylla DB
summary
Describes migrating a transactional ML inference service from a 12-node Cassandra cluster (10 cores, 32 GB RAM per node) to ScyllaDB to reduce scaling costs. Covers the operational experience and concrete results of replacing Cassandra with ScyllaDB for that workload.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 1:12Cassandra problems
- 5:52Reasons for Scylla
- 6:35Testing and comparison
- 11:21Data compaction issues
- 13:29Scylla test scenarios
- 17:07Migration results
- 19:40Q&A
related talks
41mPerformance & Fault Tolerance: Building a Modern Database
44mScyllaDB where Cassandra meets Redis!
26mA deep dive and comparison of Python drivers for Cassandra and Scylla
01h 02mBreak Out of Redis
25mCache? Where we're going, we don't need cache!
30mOvercoming Distributed Databases Scaling Challenges with Tablets
26mDesigning a Query Queue for ScyllaDB
27mSparking our way to YugabyteDB: Our journey from ScyllaDB to