The driver triggers schema refreshes on schema change events (see Cluster.Manager#handle) or responses (DefaultResultSetFuture#onSet).
Currently, this is always a full refresh, even when only a single table was created/updated. For keyspace that have lots of tables, querying schema_columnfamilies and schema_columns for all tables can have a non-negligible impact on network usage.
refreshSchemaAndSignal can take a table name as a parameter to target the refresh more precisely, so we should take advantage of that.
Any delivery time estimate for this?