[henrikschroder]

The one great thing with Cassandra is how easy it is to expand your cluster. You just start a new server up, point it to the existing cluster, and it automatically joins it, streams the sharded data it should have to itself, and start serving requests.

Balancing your cluster requires a little bit more handholding, and if something goes wrong or you fuck it up, it can be pretty challenging. But most of the time it's pretty painless.

There are a lot of other warts though, the data model is slightly weird, the secondary indexing is slow, and eventual consistency is hard to wrap your head around, but it doesn't require much effort to run and operate a large cluster, and if that's important to you and your application, you should check it out.

The NoSQL space is pretty interesting, but there is no clear winner, each of the competing solutions have their own niche, their own specialities, so it's impossible to give general recommendations right now.