[ 
https://issues.apache.org/jira/browse/HDFS-2006?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13040092#comment-13040092
 ] 

Aaron T. Myers commented on HDFS-2006:
--------------------------------------

One option to not waste memory resources for files/directories which do not 
have extended attributes set on them would be to go the same route as the 
{{INodeDirectory}} vs. {{INodeDirectoryWithQuota}} classes, i.e. have 
{{INodeFile}} and {{InodeFileWithXAttrs}}. Of course, this strategy requires 
that we implement classes for the matrix of all the different possible 
combinations of extra metadata types.

A reasonable question to ask is whether we should concern ourselves all that 
much with the memory consumption of this, since the extended attributes will 
presumably be per-file. Since the NN's memory is dedicated to relatively few 
files and relatively many blocks, it may not matter much if we go with a fairly 
naive approach.

Dhruba, do you have any data as to what percentage of the NN heap in a typical 
cluster goes to per-file metadata versus per-block metadata?

> ability to support storing extended attributes per file
> -------------------------------------------------------
>
>                 Key: HDFS-2006
>                 URL: https://issues.apache.org/jira/browse/HDFS-2006
>             Project: Hadoop HDFS
>          Issue Type: Improvement
>          Components: name-node
>            Reporter: dhruba borthakur
>            Assignee: dhruba borthakur
>
> It would be nice if HDFS provides a feature to store extended attributes for 
> files, similar to the one described here: 
> http://en.wikipedia.org/wiki/Extended_file_attributes. 
> The challenge is that it has to be done in such a way that a site not using 
> this feature does not waste precious memory resources in the namenode.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

Reply via email to