[
https://issues.apache.org/jira/browse/HDFS-5776?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13871670#comment-13871670
]
Liang Xie commented on HDFS-5776:
---------------------------------
thanks for comments, probably "quorum read" is not accurate, just copy from
fb-20 & 0.89-fb branches' naming.
The core logic is in pread op, send a callable read request against dn1, if no
response until a timeout reach, then chooseDN will pick another DN to send the
similar read request.
This idea is similar with Jeff Dean's : <The tail at scale-dean>, in that
article, Jeff named a "Hedged requests", and a clever defer sending request
optimization to limit the additional load.
> support quorum read in DFSClient
> --------------------------------
>
> Key: HDFS-5776
> URL: https://issues.apache.org/jira/browse/HDFS-5776
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: hdfs-client
> Affects Versions: 3.0.0
> Reporter: Liang Xie
> Assignee: Liang Xie
>
> This is a placeholder of hdfs related stuff backport from
> https://issues.apache.org/jira/browse/HBASE-7509
> The quorum read ability should be helpful especially to optimize read outliers
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
