Cassandra vs ScyllaDB at scale: why Discord switched
Published 2026-10-19 · Built only from first-party engineering write-ups in our database. Every number below comes from the linked source.
The clearest real-world comparison is Discord. It moved messages from MongoDB to Cassandra in 2017, then moved its message store from Cassandra to ScyllaDB and put Rust data services in front that coalesce requests for hot partitions, at trillions of messages. Discord's post describes ScyllaDB as Cassandra-compatible and written in C++, with no garbage collector, after years of JVM garbage-collection pauses on Cassandra.
Cassandra vs ScyllaDB: side by side
| Cassandra | ScyllaDB | |
|---|---|---|
| Stories in our database | 3 | 2 |
| Companies | Discord and Netflix | Discord |
| Published | 2017–2026 | 2023–2026 |
| Techniques alongside it | database migration (2), backups as cold tier (1), compaction (1), cost optimization (1) | data services layer (1), database migration (1), database operations automation (1), idempotent tasks (1) |
| Scale numbers stated in sources | 15+ PB compressed data in cold storage (Netflix); trillions of messages (Discord); billions of messages (Discord) | dozens of clusters, hundreds of database nodes (Discord); trillions of messages (Discord) |
| Direct comparison story | Discord: How Discord Stores Trillions of Messages | (same story) |
Verdict
Discord's own caveat: a new database alone doesn't remove hot partitions, so it also added a request-coalescing layer. If your pain is GC pauses and repair toil, Discord's story is the one to read.
Counts are stories in How They Scaled tagged with each side, see the Cassandra and ScyllaDB pages. They describe what these teams wrote about, not market share or benchmark results.