I'm starting to move to JBC for Cassandra (away from Hector).    In his strange loop 2012 anti-pattern presentation, Mathew Dennis writes:

"Sometimes people try to restrict clients to a single node. This actually takes work, and causes problems. Donít do it."

Now, I note that the JDBC pooled connection class takes a single  node as an argument creating  a pool of connections to that node.   So, Whats the best way of handling this in a (web) application ?  Should I create multiple pools to each of the nodes in the cluster or is there a better best practice.

BTW
Is there an equivalent  to Hector's getKnownPoolHosts for JDBC ?

Many Thanks

Andy C

The University of Dundee is a Scottish Registered Charity, No. SC015096.