Cassandra snapshot backups

Netflix's Java Cassandra client.

Rackspace's Node CQL client.

Introductory operational information about Cassandra.

Rick's first presentation as a DataStax employee. Good intro to Cassandra, including a short list of things to not do.

It seems that, like me, everyone wants atomic increment in Cassandra. I still don't see it happening.

Arin's concrete walkthrough of the Cassandra data model.

This intrigues me: "unlike most key/value stores, you can query for ordered ranges of keys." In my messing around with Cassandra, it's Thrift Interface seemed to actively hinder doing anything like this.

Worth a second look

Good survey of the key-value storage landscape as of January 2009. Probably wrong by now. His findings mirror my own - memcachedb and perhaps Tokyo Tyrant look the best. Eventual consistency is the big open question for every use case except OpenDNS Stats.

Hammer's talk about data at Facebook. 320 nodes for Hive, 120 nodes for Cassandra. Large data size and a (relatively) low write rate.