"Sabi" <Sabi (AT) discussions (DOT) microsoft.com> wrote
Quote:
I understand and thank you for your replies. I found a presentation from
Microsoft which says that there is zero work loss on failover clustering.
PLs check it:
http://download.microsoft.com/downlo...e%20Mitigation |
Sorry, but there is work loss on a failover because Microsoft clustering
does not save the state of the transaction. So, transactions in the fly
(meaning they are processing and have not been written to the transaction
log), will be lost. This also means that your client application will have
to reconnect after the failover so it can connect to the node hosting the
SQL cluster group. The client can be written to re-try if there is a failure
so that they will then reconnect to the cluster and process the transaction
after the failover in the event the transaction is not completed.
What is meant in the white paper is that if the transactions are written to
the transaction log, then there is no work loss on failover.
I hope that helps.
--
Russ Kaufmann,
MCSE: Messaging and Security, MCT, MCITP, MCTS and other stuff
ClusterHelp.com, a Microsoft Certified Gold Partner
Web http://www.clusterhelp.com
Blog http://msmvps.com/clusterhelp
The next ClusterHelp classes are:
July 28-31 in Pittsburgh
Oct 13-16 in New York