You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@zookeeper.apache.org by "Fangmin Lv (JIRA)" <ji...@apache.org> on 2019/07/28 18:52:00 UTC

[jira] [Created] (ZOOKEEPER-3473) Improving successful TLS handshake throughput with concurrent control

Fangmin Lv created ZOOKEEPER-3473:
-------------------------------------

             Summary: Improving successful TLS handshake throughput with concurrent control
                 Key: ZOOKEEPER-3473
                 URL: https://issues.apache.org/jira/browse/ZOOKEEPER-3473
             Project: ZooKeeper
          Issue Type: Improvement
          Components: server
            Reporter: Fangmin Lv
            Assignee: Fangmin Lv
             Fix For: 3.6.0


When there are lots of clients trying to re-establish sessions, there might be lots of half finished handshake timed out, and those failed ones keep reconnecting to another server and restarting the handshake from beginning again, which caused herd effect.
 
And the number of total ZK sessions could be supported within session timeout are impacted a lot after enabling TLS.
 
To improve the throughput, we added the TLS concurrent control to reduce the herd effect, and from out benchmark this doubled the sessions we could support within session timeout.



--
This message was sent by Atlassian JIRA
(v7.6.14#76016)