hadoop-zookeeper-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Mahadev konar (JIRA)" <j...@apache.org>
Subject [jira] Commented: (ZOOKEEPER-321) optmize session tracking in zookeeper.
Date Thu, 04 Feb 2010 19:35:29 GMT

    [ https://issues.apache.org/jira/browse/ZOOKEEPER-321?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12829719#action_12829719

Mahadev konar commented on ZOOKEEPER-321:

  here is what I think the average use case will be. Lets say we have 2000 clients on each
of the zookeeper server. If we assume that even 100% of these clients are read only, we will
be saving  - <size of long> * 2000 of traffic from a follower to the leader (the session
touch updates that go out every 2 seconds to the leader). That is 8 KB of data from the follower
to the leader. 

A total of 4 * 8KB of data can be saved on the net traffic to the leader from the followers
which is not very high at all. The amount of the network saving that we will make does not
seem like worth the effort that we would be spending in doing whats being proposed (basically
no session tracking for read only/watching clients).

Does this clear things out?  

> optmize session tracking in zookeeper.
> --------------------------------------
>                 Key: ZOOKEEPER-321
>                 URL: https://issues.apache.org/jira/browse/ZOOKEEPER-321
>             Project: Zookeeper
>          Issue Type: New Feature
>          Components: c client, java client, server
>            Reporter: Mahadev konar
>            Assignee: Mahadev konar
>             Fix For: 3.3.0
> sometimes a lot of zookeeper clients are read only. For such clients we do not need the
session tracking in zookeeper. Getting rid of session tracking for such clients will help
us sclae much better.

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message