>Maybe this is what you�re referring to Rickard, but I think we should go
one
>step further and perform "state-reflection" throughout the cluster of
>servers. In other words, all of the servers share the state of the others
so
>they know who the clients are

Why do they need to know who the clients are? There can be a lot of them...

> and what the state of their transactions were
>prior to the failure of a server-node.

Not necessary. All the recoverable state is in the database.

/Rickard




--
--------------------------------------------------------------
To subscribe:        [EMAIL PROTECTED]
To unsubscribe:      [EMAIL PROTECTED]
Problems?:           [EMAIL PROTECTED]

Reply via email to