News for hackers. Stuff that matters.
Re:Cassandra vs MongoDB For Time Series Data by ghc on Wednesday August 21, 2013 @01:57PM (#6252154)
We had some major issues using MongoDB for time series data due to the write volume (real time sensor data). The solution for us was Riak, mainly because we never need to update a vector clock, leaving us without the need for conflict resolution (and last write wins is fine for sensor data).
Re:Cassandra vs MongoDB For Time Series Data by k_bx on Wednesday August 21, 2013 @02:11PM (#6252249)
Why wouldn't generating some hashed _id work? It would then scale for writes easily (or, as in current mongodb, you could use hash-based indexes).

I'm just asking since Riak seemed much slower for me when I tried it.

Re:Cassandra vs MongoDB For Time Series Data by ghc on Wednesday August 21, 2013 @03:21PM (#6252709)
Hashed _id's aren't the bottleneck. Disks are the bottleneck. MongoDB is seriously lacking in the compression department and bogs down our disks constantly.

Riak is not slow as long as you're not running your cluster on VPSs. At our scale Riak's performance has been significantly better than MongoDB's due to our heavy write load, and there are fewer issues with using big disks with somewhat larger seek times.