Jeff Baxter

You could certainly still overload a system.

And in fact, the reason for QS minimums is you can run a system at 115%.

The goal then of a QS minimum is saying, okay, we know that some of your workloads on the system are going to get shanked.

Let's just be honest.

You're running at 115%.

There's no such thing as a free lunch.

Some of your workloads are going to start to get some pretty unacceptable latency.

If you haven't set any sort of QS on the system, that pain is going to be distributed roughly evenly across all of your workloads.

If you have a QS minimums, we're going to provide that, you know,

We're going to provide generally that sub millisecond latency to those workloads at the IOPs and restrict any pain from them.

In fact, the higher the floor you set, the higher the QoS minimum you set, the more weight we're going to give to delivering on that QoS minimum and shift all the pain of overburdening the system onto any of the workloads that are not protected.

Does that make sense at least?

Yeah, so at some point, you're going a little past my understanding of it.

What I do know, and I'll tell you what I do know, is that it is a dynamically generated statistic, basically.

It is a dynamically generated metric.

So

when ONTAP goes in, when ONTAP 9 goes in, it knows what's attached to its system.

It knows what the workloads are on top of it.

And so that number is not some sort of hard-coded thing in there.

Like if you're an A300, your headroom number is X, right?