What data store should I use for storing and querying a billion small (~2kb) JSON documents?

I'm trying to figure out which database to use for storing and querying on ~1 billion JSON documents about the size of 2KB each. About 2 TB of data. Nature of JSON documents: * 3 types of documents (structure is loosely defined) * Nested, multi level, sparse Things that I'm looking for are: * High throughput * Secondary indexes support * Queries over JSON Nice to haves: * SQL like query interface Things I'm looking at: * Cassandra * Storing JSON in Postgres * Riak * HBase If more details I missed regarding usage or the data I missed that would help make the decision, please let me know