Data store
The data store for march is a classic RDBMS (developed against PostgreSQL).
Indexing
Lucene via Solr.
Why not Ferret? Because Ferret broke repeatedly (corrupt indexes); and when you have 500000 documents to index and no repair mechanism...
