Evan Miller

There's too many applications, right?

And we're all trying to consolidate infrastructure into large shared infrastructures rather than hundreds of little dedicated infrastructures.

So a service level solves that problem.

We abstract the service away from the hardware.

We get into a software-defined, software-managed kind of an infrastructure.

Now a service level...

There are three popular ones everybody takes, and it makes sense.

When you go to buy something, you have the choice of small, medium, or large, right?

When you go get an internet service, you have slow, medium, fast.

The human brain thinks of something really cheap or something really good or the compromise in the middle, and that's the way the brain works.

So the most popular service levels across about 240 customers have been

There needs to be a very cost optimized service.

We call it value just to give it a name.

And it's really, really cheap.

It should be a penny a gig or two pennies a gig.

On the opposite extreme, you have that tiny sliver of about 3% of the workloads in an infrastructure that need to go extremely fast, right on the edge, as fast as the application needs.

That's called extreme.

But then the vast majority of the middle we call performance.

That's where all the VMs and databases and that's where all the production and general workload, you know, IT lives.

And so you've got value, performance, and extreme, and the IOPS to go with it.