IO POV

IO.POV presents IO executives’ views on subjects that have been condensed and edited. This first post covers evolving economic requirements of deployment, wasted capital, the real goal of IT, provisioning in the future, and moving beyond one-size-fits-all infrastructure.

IT has to deliver more compute capacity at a lower operating cost every single day.

The enterprise customer is demanding a scale methodology for delivering applications to their users.

Industrial compute capacity is being transformed from industrial use to enterprise use. This is not optional, it’s a necessity. If you want to compete economically with your peer group, you’re going to have to adopt some type of industrial style of deploying compute capacity to run your environment and drive your applications.

Industrial Compute (1)
Stuck in the ‘Teens

The data tells us that infrastructure is woefully underutilized. The silos of capacity that we’ve created, despite the compression driven through virtualization and other technologies – if you look at the entire stack, you’re talking about full utilization rates of the total asset base in the ‘teens. And that’s when it’s good. So this is an enormous amount of wasted capital.

Industrial Compute (3)
Connected – For Less

Everyone forgets what the goal of IT is, and that’s to keep the user in a persistent or non-persistent connection to an application. And our goal is to deliver that in the most efficient and most economical way possible. Tied out to what the service level needs are. That’s it.

And what that means is that everything below that connection is going to be commoditized, it’s going to be delivered in a commercial and standardized way, and it’s going to be delivered using industrial scale infrastructure.

You know how many license fees you’ll pay when you spin up a virtual machine on an open cloud platform? None. Not one. Not one software license fee, not one hardware license fee. You won’t have to pay for the privilege of using something with a brand and plastic on it. And the crappy support that comes along with it.

The barriers to this are coming down.

We should keep our eyes on the prize.

Industrial Compute (4)
No More Cardboard Boxes

The days of compute being shipped to data centers in cardboard boxes, then being unboxed, put in racks, custom-configured, custom-cabled, custom-managed, are coming to an end.

Provisioning infrastructure – from the IT hardware down to the physical layer – will be an app on a smart phone. Users will stand up a virtual machine, select an operating system, select a production or dev network, in a zone with the appropriate level of resiliency in a data center. All on a smart phone, across an LTE network from anywhere on the globe. In less than three minutes.

Infrastructure is going to be provisioned with clicks.

That’s what the world of compute deployment is going to be. If people don’t see this coming, they’re going to get run over by it.

Industrial Compute (5)
The Application Makes the Call

The data center should be an application level decision. The decision needs to go from an enterprise level decision, one data center fits all approach, to a call by an application.

Ultimately, the data center should be configured based on the application call. The capability to provision infrastructure across the data center based on what the individual application call needs at a given time. Is it reliability, is it security, is it sustainability – what does it need when it enters that infrastructure? And then we give it what it needs when it needs it.

This will be the next mile marker, when we’ve enabled application level decisions to be made about the data center. In a hybrid approach.

And it has to be a hybrid approach. Some applications will be suitable for public cloud. But you’ll have to be able to support legacy infrastructure for certain apps. So you’ll have to have a continuum available to you.

This is what enterprise compute is going to be like in the near future.