Hello, Forgive me if I have missed anything in the obvious locations, but I am trying to find out if anyone has done an analysis of the gossip protocol as implemented in Cassandra? In particular, I am interested in the the theoretical propagation time (to all nodes) of a change. For example, if a single node makes a change, what is the number of gossip cycles that must elapse before we can be sure (to very high probability obviously, not 100% sure) that everyone has seen it. For those familiar, this is obviously an epidemic analysis, but the Scuttlebutt-style protocol makes it a bit more complex.
If anyone already has this, it would be much appreciated. Thanks! -Bill Katsak Ph.D. Student Department of Computer Science Rutgers University
