hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Bharath Mundlapudi (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HDFS-1061) Memory footprint optimization for INodeFile object.
Date Wed, 24 Mar 2010 18:09:27 GMT

    [ https://issues.apache.org/jira/browse/HDFS-1061?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12849324#action_12849324

Bharath Mundlapudi commented on HDFS-1061:

Storing exponents is a good suggestion and i thought about this but didn't suggest for the
following reasons
1. storing and retrieving need more cpu cycles for the computation. 
2. didn't want to assume power of n.
3. unless we are saving another 8 bytes for byte alignment or else even with 1 byte we will
be forced to use 8 byte on 64-bit JVM.
4. I believe, ext3 and ext4 file systems uses 32 bit and 64 bit block addressing respectively
so i thought 40 bit (1TB) probably good enough for HDFS block size. Note that for my proposal,
we could go for 48 bit preferred block size and 16 bit replication which may not be required
and also won't give any extra bits for storing meta data in future if we want to.  

> Memory footprint optimization for INodeFile object. 
> ----------------------------------------------------
>                 Key: HDFS-1061
>                 URL: https://issues.apache.org/jira/browse/HDFS-1061
>             Project: Hadoop HDFS
>          Issue Type: Improvement
>          Components: name-node
>    Affects Versions: 0.22.0
>            Reporter: Bharath Mundlapudi
>            Priority: Minor
>             Fix For: 0.22.0
> I am proposing a footprint optimization to merge blockReplication and preferredBlockSize
fields into one 'long header' field in INodeFile class. This saves 8 bytes per INodeFile object
on a 64 bit JVM. This memory optimization is transparent and changes are very minimal.

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message