[
https://issues.apache.org/jira/browse/HDFS-7967?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15815227#comment-15815227
]
Kihwal Lee commented on HDFS-7967:
----------------------------------
I have reviewed the patch and am fine with it. We have used a variant of this
over a year and this is the latest improved version.
We can commit this to branch-2 and 2.8 now, but then we will likely forget
about the remaining work and move on. So I think we need to discuss what we
are going to do for trunk. [~daryn], would you share your thoughts and concerns
on the state of trunk and possible solutions?
> Reduce the performance impact of the balancer
> ---------------------------------------------
>
> Key: HDFS-7967
> URL: https://issues.apache.org/jira/browse/HDFS-7967
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: namenode
> Affects Versions: 2.0.0-alpha
> Reporter: Daryn Sharp
> Assignee: Daryn Sharp
> Priority: Critical
> Attachments: HDFS-7967-branch-2.8.patch, HDFS-7967-branch-2.patch,
> HDFS-7967.branch-2-1.patch, HDFS-7967.branch-2.001.patch,
> HDFS-7967.branch-2.002.patch, HDFS-7967.branch-2.8-1.patch,
> HDFS-7967.branch-2.8.001.patch, HDFS-7967.branch-2.8.002.patch
>
>
> The balancer needs to query for blocks to move from overly full DNs. The
> block lookup is extremely inefficient. An iterator of the node's blocks is
> created from the iterators of its storages' blocks. A random number is
> chosen corresponding to how many blocks will be skipped via the iterator.
> Each skip requires costly scanning of triplets.
> The current design also only considers node imbalances while ignoring
> imbalances within the nodes's storages. A more efficient and intelligent
> design may eliminate the costly skipping of blocks via round-robin selection
> of blocks from the storages based on remaining capacity.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
