[
https://issues.apache.org/jira/browse/HADOOP-10410?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13944384#comment-13944384
]
Liang Xie commented on HADOOP-10410:
------------------------------------
bq. Do you have s hacked up hbase patch that makes use of these nativeio
additions?
The modifications on datanode/dfs client/hbase sides all are minor, i have a
plan to impl/test next week.
bq. The patch is for your particular use case on a specific platform of a
specific setting (HBase on Linux). Though valuable to make the use case work,
the HDFS APIs, however, have to carefully designed to support other use cases
(e.g., MR) and other platforms
My current thought is we can pass the "hint" from dfs client, means if a upper
applications like HBase want this feature, just pass the hint to dfs client
layer, then datanode layer. the other applications like MR have not influence
by default.
bq. Why not get some experimental results, and then decide whether to commit it?
and
bq. Doing our own IO scheduling, however, is a much larger project, which to be
effective may also require other deep changes like switching to O_DIRECT IO,
etc.
Yeh, totally agreeeeeeeeed !!! :)
> Support ioprio_set in NativeIO
> ------------------------------
>
> Key: HADOOP-10410
> URL: https://issues.apache.org/jira/browse/HADOOP-10410
> Project: Hadoop Common
> Issue Type: New Feature
> Components: native
> Affects Versions: 3.0.0, 2.4.0
> Reporter: Liang Xie
> Assignee: Liang Xie
> Attachments: HADOOP-10410.txt
>
>
> It would be better to HBase application if HDFS layer provide a fine-grained
> IO request priority. Most of modern kernel should support ioprio_set system
> call now.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
