100x faster != 100x cheaper. Anyway, I'm familiar with the codebase of both Cassandra and especially MongoDB and I just don't see room for a 100x performance im

100x faster != 100x cheaper. Anyway, I'm familiar with the codebase of both Cassandra and especially MongoDB and I just don't see room for a 100x performance improvement without sacrificing something. A bit of I/O tweaking certainly isn't going to result in such improvements nor will "Coding in C" which in and by itself does very little for performance at all. That said, you are correct that the proof is in the pudding and everyone is free to do their own benchmarks. I'll do my own and see what's what.