central catalog for all resources

2. propagate resource states

3. notify to the catalog if resource is changed
(add, remove, updated, ...)

list all available resources / 
download tables

5. SparkInterpreter can use the zeppelin resource pool  as a data source

1. user executed JDBC paragraph and wants to share via a resource pool.  

then, JDBC interpreter creates `JDBCTableData` to point the table result which want to be shared

Each `ResourcePool` instance in an interpreter consists the distributed resource pool

7. Spark interpreter fetches
the remote table's rows, columns
by using remote `invokeMethod` call

10. get the actual resource by querying the table in the storage

ResourcePool

ResourcePool

6. The table (resource) that spark cluster wants to use might be persisted in a different interpreters' resource pool