ActiveCluster

we are doing 1 snap a day and keeping them for 1 week. the performance problem occurs before we take the snap and during a backup window

we have 4 ports per controller so 8 per datacenter

and that is over FC

I got this from the storage team

what layer is the "queue time" for?

I _think_ it's effectively "other". So, queue time in your fabric, or on the computer side (both physical and virtual). But, if you're only taking one snap a day, that's going to be a rather sizable delta to push over the wire. And, if it's a single PG with a lot of volumes in it, that's going to disrupt some things.

but if its a snapshot with activecluster, is there anything to push over the wire?

but in any case wouldn't the performance impact be after the snapshot is taken not before?

Yeah. The snapshot is also getting pushed over the wire.

So, the impact is before the snapshot, but aligns with that snap?