Hoping that someone out there using OS clusters or IP clusters can offer some info.
I have been asked to better our Domino failover.
We currently run 2 Domino NT5 servers in a DOmino cluster. Failover, as you all know, is far from seamless and quite cumbersome if the user has a document in progress when the server crashes.
I would guess that we are currently at less than 8 hours of actual downtime each year…but our execs. perceive failover inconvenience as downtime.
They are looking for a guaranteed 100% uptime and seamless failover solution. We have approx 1400 users. We also have a hub/spoke WAN with 26 Domino servers. This solution is only for a single location (Headquarters) at this time. But we would need to be able to scale this (I assume).
A local consulting firm recommended another Domino cluster on an IBM i520 box.
This will not alleviate the clumsy Domino failover ‘feature’.
Anyone done anything with 2 physical servers on an IP cluster or OS cluster?
Basically 2 seperate boxes both viewed by clients as a single Notes server.
All info much appreciated.
Although - I think they will freak when/if they receive a real solution quote.
I am not sure what the solution the local consulting firm is offering. Most (including Microsoft based) IP and OS clusters will have the same problems as the domino failover. They dont actually run two machines in lock-step, but transfer the ip address and launch the service on the backup machine when the primary machine fails. As both machines share a common disk array, you are back up and running quickly. The down side is that all conversations on progress (ie open documents) are lost, the same as domino failover.
You might consider coding the domino application to check for the presesnt of the server before it issues a document save. If the server is not found have the application copy the document to a backup server.
I couldn’t agree more with the clumbsy and far from seemless failover. One issue I have experienced is that some clients failover and others don’t and this appears to be on an intermittent basis, despite presence of cluster.ncf. Although I can’t offer any solutions I do have an ESR open (PMR 18522 215 616) since Nov 9 last year regarding failover issues and IBM have still failed to resolve this. Maybe if they ever do resolve it i will post it on this forum.