Originally posted by: jtswoody

I have a prospect who is hesitant to use PowerHA unless they can manually control failover. They are concerned PowerHA may prematurely failover during day-to-day operation. Is it possible to manually control failover with PowerHA, after the software has been activated?

Originally posted by: SystemAdmin

I am not exactly sure what you mean by 'manually control' in the context of your message. PowerHA does allow you to manually move a resource group. I am also not sure what you mean by 'prematurely failover' either. There are specific conditions that will cause a failover. In general, PowerHA only controls and reacts to whatever resources you configure within PowerHA. You can also control how those resources failover.

If you could provide a more specific example someone on the forum should be able to help you out.

--Brandon

Originally posted by: jtswoody

I agree the customer is looking for something which is mutually exclusive - PowerHA is meant to automatically failover to a secondary system in the event of the loss of a primary system's ability to run an application. However, the customer does not want to let PowerHA determine when a disaster has occurred, and they don't want software such as PowerHA to automatically handle failover.

Considering our customer's concern, I believe they only want to replicate their data to a secondary storage system. When they eventually have a DR site, they can make it a warm site, with a secondary server running, connected to the secondary storage system, ready to access the replicated data in the event of a disaster as determined by the customer.

The customer's RTO is 8-12 hours. This is plenty of time to declare a disaster, to switch over to using the server at the DR site, to start up their Oracle database and applications on this secondary server, and to switch their users over to access their data at the DR site. This allows the customer to manually failover to the server at the secondary site, without using any automatic failover, such as that provided by PowerHA.

Originally posted by: Casey_B

Ahhh...So that is the missing detail.

They are concerned about the disaster recovery piece of things.

It is hard, or impossible to really add manual control into a local HA cluster.

The recent software levels for HA did add a feature for manual control between
geographically seperated sites in a disaster.

This is only able to be used with PowerHA-eXtended Distance.

There is some information in this redbook, and should be some information in the manuals:

http://www.redbooks.ibm.com/redbooks/pdfs/sg247841.pdf

However, like you mention it might be possible to fulfill the customer's expectations without installing
HA.

Good luck, Hope this helps,
Casey

Originally posted by: j.gann

I agree that the very purpose of powerha is to automate failover.

If manual failover is all that is ever needed you can consider setting up identical lpars on different systems using the same set of san disks (alternately not concurrently of course).

The concept is simple but has some caveats. It's been titled "hypervisor ha" in an ibm presentation I recall. I'd call this "standby lpars".

Regards
Joachim Gann