[jboss-jira] [JBoss JIRA] Updated: (JBMESSAGING-1340) Session replication

[Tim Fox (JIRA)]

     [ http://jira.jboss.com/jira/browse/JBMESSAGING-1340?page=all ]

Tim Fox updated JBMESSAGING-1340:
---------------------------------

        Summary: Session replication  (was: In memory session replication)
    Description: 
Implement replication of sessions between live and backup.

It will be possible to replicate sessions to a backup node. Each node will be twinned with a backup. The nodes backup will be configured in the live nodes jbm-configuration.xml. When a node is started it will not become active until its backup is started.

If a node fails, failover will occur on to the backup. It will then be possible to promote the back to a live node, and bring up another backup without downtime.

The replication needs to happen in memory synchronously.

Write to journal on both nodes can be lazy (asynchronous) since on failover we can be sure it is in memory on backup so no single point of failover.

But can persist to disk asynchronously?? I think we can persist to disk asynchrnously on the master node, but needs to be synchronous on the backup node, in case we get a failure of the backup too shortly after failure of the master.

If we can tolerate duplicate messages (or messages thought sent but not store) in a double failure then can do asynchronous storage on both master and slave.

  was:
Implement in memory replication of sessions.

It will be possible to replicate sessions to a backup node.

The replication needs to happen in memory sychronously.

Write to journal on both nodes can be lazy (asynchronous) since on failover we can be sure it is in memory on backup so no single point of failover.

But can persist to disk asynchronously?? I think we can persist to disk asynchrnously on the master node, but needs to be synchronous on the backup node, in case we get a failure of the backup too shortly after failure of the master.

If we can tolerate duplicate messages (or messages thought sent but not store) in a double failure then can do asynchronous storage on both master and slave.


> Session replication
> -------------------
>
>                 Key: JBMESSAGING-1340
>                 URL: http://jira.jboss.com/jira/browse/JBMESSAGING-1340
>             Project: JBoss Messaging
>          Issue Type: Task
>            Reporter: Tim Fox
>         Assigned To: Tim Fox
>            Priority: Critical
>             Fix For: 2.0.0 Beta
>
>
> Implement replication of sessions between live and backup.
> It will be possible to replicate sessions to a backup node. Each node will be twinned with a backup. The nodes backup will be configured in the live nodes jbm-configuration.xml. When a node is started it will not become active until its backup is started.
> If a node fails, failover will occur on to the backup. It will then be possible to promote the back to a live node, and bring up another backup without downtime.
> The replication needs to happen in memory synchronously.
> Write to journal on both nodes can be lazy (asynchronous) since on failover we can be sure it is in memory on backup so no single point of failover.
> But can persist to disk asynchronously?? I think we can persist to disk asynchrnously on the master node, but needs to be synchronous on the backup node, in case we get a failure of the backup too shortly after failure of the master.
> If we can tolerate duplicate messages (or messages thought sent but not store) in a double failure then can do asynchronous storage on both master and slave.

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira