[ 
https://issues.apache.org/jira/browse/HADOOP-15550?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16640251#comment-16640251
 ] 

Jonathan Eagles commented on HADOOP-15550:
------------------------------------------

I have some interest in back-porting this improvement to 2.8 (Mainly to get rid 
of the new ObjectMapper creation for KMSJSONWriter). Seems like a simple 
cherry-pick to 3.1 is possible when JsonSerialization library was added as part 
of HADOOP-13786. Back-porting to earlier lines either means partial 
back-porting HADOOP-13786 or creating a small KMSJSONWriter custom performance 
patch to remove the ObjectMapper creation.

[[email protected]], [~tlipcon] do you have preference on this? It seems the 
most straight forward to 1) back-port this jira to 3.1. and 2) Create a 
separate KMSJSONWriter jira for earlier lines, but I am open to suggestions.

> Avoid static initialization of ObjectMappers
> --------------------------------------------
>
>                 Key: HADOOP-15550
>                 URL: https://issues.apache.org/jira/browse/HADOOP-15550
>             Project: Hadoop Common
>          Issue Type: Bug
>          Components: performance
>    Affects Versions: 3.2.0
>            Reporter: Todd Lipcon
>            Assignee: Todd Lipcon
>            Priority: Minor
>             Fix For: 3.2.0
>
>         Attachments: hadoop-15550.txt, hadoop-15550.txt, hadoop-15550.txt, 
> hadoop-15550.txt, hadoop-15550.txt
>
>
> Various classes statically initialize an ObjectMapper READER instance. This 
> ends up doing a bunch of class-loading of Jackson libraries that can add up 
> to a fair amount of CPU, even if the reader ends up not being used. This is 
> particularly the case with WebHdfsFileSystem, which is class-loaded by a 
> serviceloader even when unused in a particular job. We should lazy-init these 
> members instead of doing so as a static class member.



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

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to