Reduce default msgs per partition about akka-persistence-cassandra HOT 5 CLOSED

This is only the case if an application uses a single PersistentActor. In many applications, there's one PersistentActor per aggregate and persistenceId is the most significant component of the partition key. So writes should distribute well in the cluster. On the other hand, reducing the partition size should be fine here as well, unless replay throughput should be as high as possible.

from akka-persistence-cassandra.

Good point. An additional benefit would be to reduce the impact of tombstones, the smaller the partition the quicker we start using a different one and stop reading over deleted sequence numbers. That gets me thinking we could add an additional static column to indicate the whole partition has been deleted but it was in use and should move to the next partition.

from akka-persistence-cassandra.

An additional benefit would be to reduce the impact of tombstones, the smaller the partition the quicker we start using a different one and stop reading over deleted sequence numbers.

Makes sense, let's reduce the default partition size.

from akka-persistence-cassandra.

Anyone have a preference on the number? 500k? We'll need to think about this when we do a migration tool.

from akka-persistence-cassandra.

500k sounds like a reasonable default.

from akka-persistence-cassandra.