Ideally you want net new writes landing in the FC tier then you have fewer issues with destaging cache out to slower disk. Once they land let AO decide where they need to go based on access density. This doesn't mean that you can't pin things in Nearline just keep in mind it's highly likely you'll run out of I/O in the nearline tier well before you run out of capacity so don't let a pinned workload drown the nearline.
Unfortunately there is no best method other than keep it simple initially, that way you can get a baseline before thinking about more complex configurations as it takes a bit of experimentation. AO is not a panacea and doesn't understand your own personal preference for a particular applications importance, everyone's workloads, I/O density, acceptable response times and peak periods differ.
AO, virtual volumes and user CPGs
Re: AO, virtual volumes and user CPGs
after a rethink we are going to keep things simple and use just the 1 AO policy. to limit how much data ends up NL initially we will use a growth limit, the idea of this to prevent data being pushed too low.
we'll leave it on analyze for now but can you still see where data would have been moved? and where is this located?
Thanks
we'll leave it on analyze for now but can you still see where data would have been moved? and where is this located?
Thanks