I'd be very interested in seeing a comparison between storage engines like rocks, wiredtiger, leveldb, sqlite4-lsm, heck even kyotocabinet and berkeleydb. The comparison would describe what workload each storage engine optimizes for, what tradeoffs were made, and a basic feature matrix (multiple process? transactions / nested transactions? etc).
There are a lot of options out there but beyond pure speed or latency measures, its hard to get a high level understanding of what distinguishes one storage engine from the next.
There are a lot of options out there but beyond pure speed or latency measures, its hard to get a high level understanding of what distinguishes one storage engine from the next.