On 07/09/2013 12:25 PM, hajjat wrote:
Hi,

In a quorum read, my understanding is that Cassandra gets a digest of the
object to be read from all nodes, and then chooses the fastest node for
retrieving the data (if the quorum is met).

According to this talk at Cassandra summit, that's incorrect.

http://www.youtube.com/watch?v=VRZk-NhfX18

It gets the real data from one node and digests from others and if there's a mismatch, then it gets the data from the others and repairs them.

Blair

Reply via email to