http://www.datastax.com/support-forums/topic/is-networktopology-strategy-unbalanced Software, Support, and Training for Apache Cassandra en-US Mon, 20 May 2013 08:45:42 +0000 http://bbpress.org/?v=1.0.3 q http://www.datastax.com/support-forums/search.php http://www.datastax.com/support-forums/topic/is-networktopology-strategy-unbalanced#post-1399 Tue, 20 Mar 2012 12:31:09 +0000 Anonymous 1399@http://www.datastax.com/support-forums/ <p>OK, makes sense now. Thanks... </p> http://www.datastax.com/support-forums/topic/is-networktopology-strategy-unbalanced#post-1395 Tue, 20 Mar 2012 02:00:33 +0000 nickmbailey 1395@http://www.datastax.com/support-forums/ <p>When using the NetworkTopologyStrategy you still need to alternate racks within each datacenter when assigning tokens. Otherwise you can see imbalances yes. The hash of each row key will determine the first replica and the other replicas will be placed around the ring in order while skipping racks that already have a replica.</p> <p>Also see: <a href="http://www.datastax.com/docs/1.0/install/cluster_init#calculating-tokens" rel="nofollow">http://www.datastax.com/docs/1.0/install/cluster_init#calculating-tokens</a> </p> http://www.datastax.com/support-forums/topic/is-networktopology-strategy-unbalanced#post-1391 Mon, 19 Mar 2012 18:08:43 +0000 Anonymous 1391@http://www.datastax.com/support-forums/ <p>Hello,</p> <p>Of course random partitioning distributes the data evenly between the nodes present in a given data center.<br /> But what about the NTS, does it places the replica evenly ?</p> <p>By reading the documentation, I assume that the first node in each rack is elected as a second replica for a all primary replica present in the other rack (assuming 2 racks, 3 nodes per rack and RF=3).<br /> Would not this lead to an unbalanced ring or am I missing something in this strategy?</p> <p>Thanks for your insight.<br /> Nicolas. </p>