@marutha you have plenty of methods, KVS chains and aggregated filtering, secondary indexes, etc. KVS is a tool, you should have an approach.
KVS is like berkley DB, you have only storing. Schema and Retrieval methods you should design by yourself. Erlang is useful here to provide supervision of keys and tables. For each KVS chain — dedicated Erlang process.
usually you need only kvs:add and kvs:fold
having these two you can write any web social app
for online analytics you still need joins and secondary indexes: SQL or mnesia here is king.