[ https://issues.apache.org/jira/browse/IGNITE-2523?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15667394#comment-15667394 ]
Konstantin Dudkov commented on IGNITE-2523: ------------------------------------------- Semen, * I tried to get rid of using keys() method, now I revert this code back. * We use it as a common interface for GridNearAtomic*UpdateRequest. It's not possible to use interface, because it must extend abstract class GridCacheMessage. Now I remove unused GridNearAtomicUpdateRequest interface and move all method signatures to abstract class to make code more clear. * I'm trying to investigate this point, but that (and others) test runs locally pretty good - can it be a problem in TC? > Introduce "single put" NEAR update request. > ------------------------------------------- > > Key: IGNITE-2523 > URL: https://issues.apache.org/jira/browse/IGNITE-2523 > Project: Ignite > Issue Type: Task > Components: cache > Affects Versions: 1.5.0.final > Reporter: Vladimir Ozerov > Assignee: Konstantin Dudkov > Labels: performance > Fix For: 1.8 > > > Essentially, in this case we could get rid of all collections and garbage > inside GridNearAtomicUpdateRequest/GridDhtAtomicUpdateRequest. > This should drastically decrease message size and improve performance. -- This message was sent by Atlassian JIRA (v6.3.4#6332)