Monitoring Dockerised Cassandra with Prometheus
As part of the move away from a proprietary in memory database madb
,
last.fm is moving to ApacheCassandra.
Cassandra is far more easier to deploy to a public cloud (in our case GCE), scale
up/down and query. madb
, being completely in memory does have pretty good
performance characteristics, but scaling requires machines with more ram,
sharding by user and in has some weird and interesting bugs.