Cassandra 1.2 Startup time improvements.

datastax.com

32 pointsxedin3 comments

This is nice, but the main reason I gave up on Cassandra was the difficulty of operating it. It's a far cry from, say, MySQL in terms of devops-friendliness. Curious how much work is being done in that respect.

There's a certain amount of complexity that comes with the territory when you move from a single-machine design to a cluster, but Cassandra does as much as possible to simplify things, starting with a fully-distributed "every node is the same" design. (As opposed to having master + router + config + slave nodes, for instance, each of which needs a different kind of troubleshooting.)

Additionally, DataStax makes the "DataStax Community" distribution of Cassandra available for free, including for production use, which includes the OpsCenter management and monitoring tool: http://www.datastax.com/products/community

When did you give it a try? Around .7 or so it became fairly download and run friendly for single node development. Setting up a cluster is fairly painless just tell it about other nodes in the cluster. Running it in production is about as easy as running any other java app in production, mostly a lot of memory tuning. I would say it is much easier than setting up a Mysql cluster.

Reply on news.ycombinator.com