Kamaji actually allows to run a shared datastore using different MySQL and PostgreSQL schemas per Tenant. This can help in overcoming the etcd limitation regarding scalability and cluster size, as well with HA and replication.

Kamaji additional CLI flags

Kamaji read the data store configuration from a cluster-scoped resource named DataStore, containing all tha required details to secure a connection using a specific driver.

Once the datastore is running, and the DataStore has been created with the required details, we need to provide information about it to Kamaji by using the following flag and pointing to the resource name:

--datastore={.metadata.name}

Drivers

Further details on the setup for each driver are available here: