What are the guidelines on using Database per Agent vs Database per Cluser?
In which scenario do we use each?
Is it possible to switch from Database per Agent to Database per Cluster easily while keeping all the data intact? Does DataMiner automatically transfer all data to the new keyspaces?
Hey Jeroen,
Database per agent typically gets used in small setups (2 or so agents) where there is very little data and the data is only relevant to the agent it belongs to. Another reason to use database per agent is when the system does not meet the requirements for cassandra cluster yet.
Database per Cluster is used when you want to ensure availabilty in from the database, greater tolerance in the event of a failing cassandra node. Bigger data requirements.
It is possible to switch from Database per agent to Database per cluster keeping the data intact, Dataminer will not automatically migrate the data, but a tool has been developed for this.