Hacker News new | past | comments | ask | show | jobs | submit login

VoltDB is a real fun platform. I've used it around the v1 and v2 era. Incredibly high tx rate (on few-hundred-dollar servers 5 years ago could get 150K tx/sec.) Being able to use SQL is fantastic. The original research papers (Volt's diverged significantly now I understand) are good reads[1].

It's a good fit any time you're considering storing a bunch of data in-memory for performance reasons. (Like telecom routing info.) Instead of writing a custom daemon, just spit it into VoltDB. Get replication, performance, etc. for free! Very neat.

Sadly the open source version isn't very ACID. They dropped the D from community edition. So you can scale out, but if any node dies, you're toast. There's still some uses, where you're running a transient or easily-rebuildable dataset. Or where you can manually run multiple full nodes (though I guess you'd need to implement cluster failover manually).

I guess it shows that it is hard to make a living off of open source products if they're really great. I've heard this from other open-source companies: the product's fantastic, no one pays. But make a taste as open source, basically a demo/trial, and get them to upgrade to commercial.

1: http://hstore.cs.brown.edu/publications/




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: