[ 
https://issues.apache.org/jira/browse/HDFS-15757?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17340768#comment-17340768
 ] 

Xiaoqiao He commented on HDFS-15757:
------------------------------------

Thanks [~fengnanli] for your pings and sorry for late response since I'm on 
vacation last week.
In my production env, I have created another proxyCache replace to 
ConnectionManager. The result is expected and the TP99 cost of each RPC 
dispatch is less than 10ms which is far less than using ConnectionManager. I 
have to state that this improvement is based on our internal branch which is 
different to community branch such as async RPC dispatch etc. And the request 
mode could be very different to common case (ugi count is very large and hit 
rate is very low when using ConnectionManager).
For this improvement, I think it is ready to enter trunk based on our staging 
env benchmark when backport it into my internal branch.
I do not get the original thought to involve the ConnectionManager for Router 
after trace the history, So I wonder if ConnectionManager is effective and is 
it necessary for Router? Please correct me if something I missed. Thanks. cc 
[~elgoiri],[~ayushtkn]

> RBF: Improving Router Connection Management
> -------------------------------------------
>
>                 Key: HDFS-15757
>                 URL: https://issues.apache.org/jira/browse/HDFS-15757
>             Project: Hadoop HDFS
>          Issue Type: Improvement
>          Components: rbf
>            Reporter: Fengnan Li
>            Assignee: Fengnan Li
>            Priority: Major
>              Labels: pull-request-available
>         Attachments: RBF_ Improving Router Connection Management_v2.pdf, RBF_ 
> Improving Router Connection Management_v3.pdf, RBF_ Router Connection 
> Management.pdf
>
>          Time Spent: 3h 10m
>  Remaining Estimate: 0h
>
> We have seen high number of connections from Router to namenodes, leaving 
> namenodes unstable.
> This ticket is trying to reduce connections through some changes. Please take 
> a look at the design and leave comments. 
> Thanks!



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org

Reply via email to