In a co-located, single failover pair, is there a way to roughly estimate the time for the failover agent to take over, based on the number of elements in the DMA or some other metric, assuming no complex automations or orchestrations are in place? just basic, out-of-the-box, monitoring and control.
Hi Jorge,
The time to fail over will indeed depend on the number of elements. The standby DMA is active, but when the failover needs to happen, all the elements on that DMA still need to be activated. And this will roughly be equal to the startup time of the DMA.
Before the elements can be activated, there will first be some checks and the handover over the virtual IP needs to happen first. Then the elements will start and just to roughly give you something, you can maybe use one or two seconds per element. There are of course many variables in this equation, and some heavy elements with a lot of element data coming from a slow and overloaded database, could result in large delays easily... Or small elements, decent compute power allowing a lot of concurrency and good storage, could also go faster as well.
I hope this gives you already some insights on this matter.
Bert