Erlang thrift interface to Cassandra
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


Erlang thrift interface to Cassandra

  • This is just a tiny shim on the complete list of (current) cassandra thrift commands, as listed here

    • Yes, the returned data has all sorts of records in it. Sorry. Like I said, its a shim
  • Keyspaces correspond to poolboy pools which are automagically started for you when you set them. These pools are clobbered when the keyspace is deleted (or the application is restarted).

  • CQL commands can be sent using the execute_cql_command, prepare_cql_command, and execute_prepared_cql_command methods. Note that if you are using prepared cql commands, you will need to expliclty start a cql_pool and reference it (these have one and only one worker). Check out start_cql_pool/1 for more info.

Liberally takes ideas from Louis-Philippe Gauthier's truly excellent cassanderl, but with a more recent version of thrift (19.36.0), and poolboy instead of dispcount


  • Make sure you have Cassandra running.
  • Start up erlang_cassandra
erlang_cassandra@paglierino)1> erlang_cassandra:start().
  • Create a keyspace
(erlang_cassandra@paglierino)2> erlang_cassandra:system_add_keyspace(erlang_cassandra:keyspace_definition(<<"foo">>)).
  • Set the keyspace for future commands (note that this also starts up a pool of thrift worker-bees)
(erlang_cassandra@paglierino)3> erlang_cassandra:set_keyspace(<<"foo">>).
  • Do something
(erlang_cassandra@paglierino)4> erlang_cassandra:describe_ring(<<"foo">>).


  • There is a one-to-one mapping between keyspaces and pools. If you want something other than the default number of workers for a given keyspace, use start_pool first with the keyspace_name as an argument, and then access the keyspace. e.g.
(erlang_cassandra@paglierino)5> erlang_cassandra:start_pool(<<"bar">>, [{max_size, 10}]).
(erlang_cassandra@paglierino)6> erlang_cassandra:set_keyspace(<<"bar">>).
  • Take a look at the exports in erlang_cassara.erl

    • In general, whenever you reference a Keyspace (e.g. insert(Keyspace, RowKey, ColumnParent, Column, ConsistencyLevel)), you can use a Destination instead of a Keyspace.
    • A Destination is simply a triplet, i.e., {Host, Port, Keyspace} (this is relevant for when you might be running multiple Cassandra instances, though lord knows why you would do that).
    • As such, you could also use insert({Host, Port, Keyspace}, RowKey, ColumnParent, Column, ConsistencyLevel)
    • If you are relying on Destination, always pass in the Destination in your commands. e.g.
      • instead of erlang_cassandra:set_keyspace(<<"bar">>)., use
      • erlang_cassandra:set_keyspace({Host, Port, <<"bar">>}, <<"bar">>).
  • For all the thrift commands that don't take a Keyspace (e.g. system_update_keyspace) you can either let the application deduce the Keyspace from the command, or you can explicitly specific it yourself. e.g. any of the following are acceptable

    • system_update_keyspace(KeyspaceDefinition)
    • system_update_keyspace(Keyspace, KeyspaceDefinition)
    • system_update_keyspace({Host, Port, Keyspace}, KeyspaceDefinition)