As federal agencies move to a hybrid, multi-cloud model of computing, they are moving past virtualization, in search of more elastic methods to manage workloads. The incorporation of edge computing can augment clouds and data centers with lightweight and fully functioning applications.

When the DoD acquires systems in a fixed environment, like a limited-space data center, with limited power and cooling, the lead time is long. And a one-size-fits-all approach should be eschewed.

“The problem with essentially copying and pasting into a cloud environment is you may find that you are estimating and therefore being provisioned resources that you will never use,” said Paul Puckett, Director of the Army CIO’s Enterprise Cloud Management Agency, on Federal Monthly Insights – Network and Application Modernization.

“This pivot is critical for when it comes to how we buy cloud as a utility. The NDAA called this out, we need to stop buying computing infrastructure, storage infrastructure, like this fixed thing in a data center,” Puckett said. “We need to start designing systems and services that are intended as a distributed architecture that are intended to leverage elastic scalability.

Puckett, a member of the Senior Executive Service, who has been in his current post since 2019, calls elastic scalability one of the core components of what differentiates cloud computing from typical on-premise compute.

“I would push back if someone said, ‘Well, hey, the problem with this model is I have an issue with estimating precisely what the future looks like.’ You’re right, you do and I don’t want you to do that. I want you to come with a design that leverages every single capability of the cloud, to enhance the way that my mission is executed,” Puckett said to Federal News Network’s Jason Miller.

The evolution for federal agencies now allows for a more economical way of buying cloud infrastructure.

“If it was a firm, fixed-price capability, all that investment would be upfront. And whether we were effective or not, with how we consume infrastructure, was almost irrelevant,” Puckett said.

“I would change the actual metrics of success, we should not be dictating, you need 10 virtual machines. We should be dictating, hey, here are the service-level objectives, the service-level agreements. Here are the actual SLAs. Here’s the actual availability and a percentage of these critical services. Here’s the time that it needs to be for me to be able to process some data and then execute some decision. By starting to put out parameters of how we want to function and operate what our objectives of a mission are, we actually create space in the room for the technical design that will deliver that service,” Puckett said.

Copyright © 2024 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.