In 2017, we shared our journey from using MongoDB to migrating to Cassandra for storing billions of messages due to the need for scalability and fault tolerance. However, our Cassandra cluster faced serious performance issues due to hot partitions and maintenance troubles. After experimenting with ScyllaDB and observing improvements, we migrated all our databases to it by 2020, except for cassandra-messages. We struggled with migrating trillions of messages but eventually succeeded, leading to better performance, reduced latencies, and unlocked new product use cases. Thanks to our Rust-based data services and ScyllaDB, we can handle massive traffic without breaking a sweat. Join our team if this excites you!
https://discord.com/blog/how-discord-stores-trillions-of-messages