I’ve been involved with some heavy data mining, storage and analysis. We’re currently using MongoDB (not my choice) but I’m considering moving to something faster. Queries on large collections are taking a long time, even with indexing. I’m sure I can optimize it further, but it might still be a problem later. Thankfully, Pete Warden has done some legwork and benchmarked a few other key/value stores, namely memcached, Tokyo Tyrant, Redis with MySQL as well.