[ https://issues.apache.org/jira/browse/CASSANDRA-5972?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13756667#comment-13756667 ]
Jacek Lewandowski edited comment on CASSANDRA-5972 at 9/3/13 2:47 PM: ---------------------------------------------------------------------- Yes, that does assume a bunch of things. I know that it is far more complicated than the current solution. And, I agree that a huge advantage of the solution proposed in https://issues.apache.org/jira/browse/CASSANDRA-3200 is its rather straightforward implementation, especially in comparison to the one described here. However, I've found it difficult to prove that the my solution is worse :) For example, in a situation where one of n+1 replicas has outdated sub-range R (and the rest are in sync), the procedure from C*-3200 will initiate n+1 data transfers. With the algorithm described here, we would require only one data transfer with a probability (n-1)/n, or 2 data transfers with a probability 1/n. It depends on which node starts the sequence. was (Author: jlewandowski): Yes, that does assume a bunch of things. I know that it is far more complicated than the current solution. And, I agree that a huge advantage of the solution proposed in https://issues.apache.org/jira/browse/CASSANDRA-3200 is its rather straightforward implementation, especially in comparison to the one described here. However, I've found it difficult to prove that the my solution is worse :) For example, in a situation where one of n+1 replicas has outdated sub-range R, the procedure from C*-3200 will initiate n+1 data transfers. With the algorithm described here, we would require only one data transfer with a probability (n-1)/n, or 2 data transfers with a probability 1/n. It depends on which node starts the sequence. > Reduce the amount of data to be transferred during repair > --------------------------------------------------------- > > Key: CASSANDRA-5972 > URL: https://issues.apache.org/jira/browse/CASSANDRA-5972 > Project: Cassandra > Issue Type: Improvement > Components: Core > Reporter: Jacek Lewandowski > Priority: Minor > > Currently, when a validator finds a token range different in n replicas, data > streams are initiated simultaneously between each possible pair of these n > nodes, in both directions. It yields n*(n-1) data stream in total. > It can be done in a sequence - Replica[1] -> R[2], R[2] -> R[3], ... , R[n-1] > -> R[n]. After this process, the data in R[n] are up to date. Then, we > continue: R[n] -> R[1], R[1] -> R[2], ... , R[n-2] -> R[n-1]. The active > repair is done after 2*(n-1) data transfers performed sequentially in 2*(n-1) > steps. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira