[jira] [Commented] (ZOOKEEPER-3111) Add socket buffer size option to tune the TCP throughput between leader and learner

["Fangmin Lv (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/ZOOKEEPER-3111?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16604962#comment-16604962 ] 

Fangmin Lv commented on ZOOKEEPER-3111:
---------------------------------------

Commented in the PR, also mention it here for easier reference:

The performance of different buffer size is highly depending on the traffic pattern, and may actually introduce higher latency, so it's hard to give a suggested buffer size.

Internally, we're not customizing the buffer size with this option anymore, instead we're using the system TCP auto-tuning options net.ipv4.tcp_wmem and net.ipv4.tcp_rmem, which are more flexible and will automatically adjusts socket buffer sizes as needed to optimally balance TCP performance and memory usage.

Since it's hard to configure correct, we decided to deprecate this patch.

> Add socket buffer size option to tune the TCP throughput between leader and learner
> -----------------------------------------------------------------------------------
>
>                 Key: ZOOKEEPER-3111
>                 URL: https://issues.apache.org/jira/browse/ZOOKEEPER-3111
>             Project: ZooKeeper
>          Issue Type: Improvement
>          Components: server
>            Reporter: Fangmin Lv
>            Assignee: Fangmin Lv
>            Priority: Minor
>              Labels: pull-request-available
>             Fix For: 3.6.0
>
>          Time Spent: 3h 20m
>  Remaining Estimate: 0h
>
> Add the socket setting to let us able to tune the TCP receive and send window size to improve the throughput during network congestion or transferring large snapshot size during syncing.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)