[
https://issues.apache.org/jira/browse/HDFS-7454?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14229285#comment-14229285
]
Haohui Mai commented on HDFS-7454:
----------------------------------
Thakns Chris for the clarification. Just took a skim at the patch -- it looks
it is deduplicating the {{AclFeature}} instead of {{AclEntry}}. It makes sense
in Vinay's use case.
As Chris pointed out in HDFS-5620, it requires some complexity to make the
feature really work. My feeling is that the optimization might be unnecessary
if we are able to fit an {{AclEntry}} into a int. On a 64-bit JVM, the current
implementation requires 48 + 8 = 56 bytes per {{AclEntry}} (the size of the
{{AclEntry}} + a reference to the object). Fitting it into a 4-byte {{int}}
would reduce the the memory usage of ACL by a factor of 14.
Consider a large cluster that has 192G heap to store 300M files. If all files
have the default ACLs which contain 3 entries, the scheme can support 300M
files with default ACLs using 4G memory. With these number, the scheme seems a
pretty good thing to have before we really thinking of getting into the mud of
implementing an interner.
> Implement Global ACL Set for memory optimization in NameNode
> ------------------------------------------------------------
>
> Key: HDFS-7454
> URL: https://issues.apache.org/jira/browse/HDFS-7454
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: namenode
> Reporter: Vinayakumar B
> Assignee: Vinayakumar B
> Attachments: HDFS-7454-001.patch
>
>
> HDFS-5620 indicated a GlobalAclSet containing unique {{AclFeature}} can be
> de-duplicated to save the memory in NameNode. However it was not implemented
> at that time.
> This Jira re-proposes same implementation, along with de-duplication of
> unique {{AclEntry}} across all ACLs.
> One simple usecase is:
> A mapreduce user's home directory with the set of default ACLs, under which
> lot of other files/directories could be created when jobs is run. Here all
> the default ACLs of parent directory will be duplicated till the explicit
> delete of those ACLs. With de-duplication,only one object will be in memory
> for the same Entry across all ACLs of all files/directories.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
