Cassandra looks very interesting. It does have a very similar data model to BigTable. I wouldn't exactly call it a BigTable clone, though, since it doesn't sit on top of a DFS, provides eventual consistency instead of full consistency, and uses the peer-to-peer mechanism you mentioned. It also currently appears to be targeted at low latency user-facing random-read workloads rather than batch scans, although Jeff H. told me batch scans are on the roadmap. If I understand correctly, it also seamlessly supports operation across multiple data centers right out of the box.
Definitely a project worth following. Chad On 7/15/08 2:02 PM, "Andrew Purtell" <[EMAIL PROTECTED]> wrote: For those who may not have heard about it. http://code.google.com/p/the-cassandra-project/ http://www.slideshare.net/jhammerb/data-presentations-cassandra-sigmod/ Cassandra is another Bigtable clone more or less but eschews an explicit coordinator (Bigtable's Chubby, HBase's HMaster, Hypertable's Hyperspace) for a P2P/DHT approach for data distribution and location and for availability. - Andy
