How does one know when it's time to add capacity to a cassandra cluster? Error msgs? Performance? Some other tell-tale sign?
Since capacity planning with Cassandra is a (seemingly) black art knowing where to start and when to grow could be somewhat problematic.
Answer
Without knowing anything about your workload, I'd start looking at:
- IO queue length
- CPU load
- Amount of data read vs written
- GC behavior
- Compaction stats
When any of these start to get "high" (definition depends on your hardware) then it's time to revisit your application, or scale up.
No comments:
Post a Comment