You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Vladimir Ozerov (JIRA)" <ji...@apache.org> on 2016/03/21 08:25:25 UTC
[jira] [Resolved] (IGNITE-2836) IGFS: Investigate why
BinaryMarshaller is slower than OptimizedMarshaller in meta operations.
[ https://issues.apache.org/jira/browse/IGNITE-2836?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Vladimir Ozerov resolved IGNITE-2836.
-------------------------------------
Resolution: Fixed
> IGFS: Investigate why BinaryMarshaller is slower than OptimizedMarshaller in meta operations.
> ---------------------------------------------------------------------------------------------
>
> Key: IGNITE-2836
> URL: https://issues.apache.org/jira/browse/IGNITE-2836
> Project: Ignite
> Issue Type: Task
> Components: IGFS
> Affects Versions: 1.5.0.final
> Reporter: Vladimir Ozerov
> Assignee: Vladimir Ozerov
> Fix For: 1.6
>
>
> *Problem*
> Running complex Hadoop tasks with default node settings typically shows worse performance than with explicitly set {{OptimziedMarshaller}}. Most probably it is caused by the fact that all our IGFS meta classes are Externalizable and hence cannot be handled by {{BinaryMarshaller}} out-of-the box.
> *Solution*
> 1) Add all participating IGFS classes (IgniteUuid, IgfsFileInfo, IgfsBlockKey, IgfsListingEntry + all processors) to exclusions in static initializer of {{BinaryContext}}.
> 2) Mark all these classes as {{Binarylizable}} and implement writes/reads in raw mode.
> Then check performance.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)