plaid_draco (11/24/2014)

---

Jeffrey Williams 3188 (1/31/2012)

---

plaid_draco (1/31/2012)

---

Thank you for the quick responses! Very helpful.

I guess my assumption that an active-active cluster would avoid that problem is unfounded. 😀

Second, in SQL Server failover of any instance is a restart of the services on the other node. Any in-process transactions would be disconnected and on startup would be rolled back through the startup procedure. This is done using the transaction log by rolling back any transactions that were not completed.

If you really need that flexibility - you would need to go with either Oracle RAC which is the only one I know of that maintains state on in-process transactions.

It's been nearly three years since I made the initial post and we have set up two successful sql clusters that are spinning away today. However, I wanted to revisit the question of interrupted connections when an instance fails over. We are SQL and won't be switching to anything Oracle per the suggestion above.

My question: Is there now anything relative to SQL server that allows us to fail over while insuring current connections aren't lost?