You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@ignite.apache.org by "Andrew Mashenkov (JIRA)" <ji...@apache.org> on 2018/09/25 13:11:00 UTC

[jira] [Created] (IGNITE-9688) MVCC: Implement out-of-order enlist optimization for bulk cache operations.

Andrew Mashenkov created IGNITE-9688:
----------------------------------------

             Summary: MVCC: Implement out-of-order enlist optimization for bulk cache operations.
                 Key: IGNITE-9688
                 URL: https://issues.apache.org/jira/browse/IGNITE-9688
             Project: Ignite
          Issue Type: Improvement
          Components: cache, mvcc
            Reporter: Andrew Mashenkov


For now, we always enlist updates in given order via setting "GridNearTxEnlistFuture.sequential" flag to true.
This flag is always true for query updates as we do not know full data set at a time future has been created.
For putAll (and other batch operations) full update map is known and we can make per-node batches full filled before send.

E.g. we can send batches to nodes one by one regarding primary node order.
This optimization has to be disscussed.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)