[
https://issues.apache.org/jira/browse/HDFS-17807?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18007688#comment-18007688
]
ASF GitHub Bot commented on HDFS-17807:
---------------------------------------
rdhabalia opened a new pull request, #7808:
URL: https://github.com/apache/hadoop/pull/7808
<!--
Thanks for sending a pull request!
1. If this is your first time, please read our contributor guidelines:
https://cwiki.apache.org/confluence/display/HADOOP/How+To+Contribute
2. Make sure your PR title starts with JIRA issue id, e.g.,
'HADOOP-17799. Your PR title ...'.
-->
### Description of PR
The current getBlockLocations method returns replica locations without
considering the real-time load on individual DataNodes. This can lead to
clients repeatedly selecting the same DataNode, particularly when network
distance or rack locality remains unchanged. Since data locality isn’t always
reliable due to various operational factors, the existing network
topology–based ordering has not fully optimized performance and may
inadvertently create load hotspots. so, replica ordering based on dynamic
weight metric and specifically current read rate or read load will create more
balanced read across the dataodes in a cluster.
Adding `dfs.read.replica.uniform.distribution.enabled` to enable this
feature. The policy is disabled by default and can be enabled to replace the
current ordering mechanism, offering a more adaptive and performance-aware read
path strategy.
### How was this patch tested?
Unit test & mvn clean install
### For code changes:
- [ ] Does the title or this PR starts with the corresponding JIRA issue id
(e.g. 'HADOOP-17799. Your PR title ...')?
- [ ] Object storage: have the integration tests been executed and the
endpoint declared according to the connector-specific documentation?
- [ ] If adding new dependencies to the code, are these dependencies
licensed in a way that is compatible for inclusion under [ASF
2.0](http://www.apache.org/legal/resolved.html#category-a)?
- [ ] If applicable, have you updated the `LICENSE`, `LICENSE-binary`,
`NOTICE-binary` files?
> Support balanced read distribution ordering in getBlockLocations()
> ------------------------------------------------------------------
>
> Key: HDFS-17807
> URL: https://issues.apache.org/jira/browse/HDFS-17807
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: namenode
> Affects Versions: 2.10.0, 3.4.1
> Reporter: Rajan Dhabalia
> Priority: Major
> Labels: pull-request-available
>
> The current getBlockLocations method returns replica locations without
> considering the real-time load on individual DataNodes. This can lead to
> clients repeatedly selecting the same DataNode, particularly when network
> distance or rack locality remains unchanged. Since data locality isn’t always
> reliable due to various operational factors, the existing network
> topology–based ordering has not fully optimized performance and may
> inadvertently create load hotspots. so, replica ordering based on dynamic
> weight metric and specifically current read rate or read load will create
> more balanced read across the dataodes in a cluster.
> The policy is disabled by default and can be enabled to replace the current
> ordering mechanism, offering a more adaptive and performance-aware read path
> strategy.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
