How will data centers hold up in an edge-computing future?

How will the rise of edge computing impact data centers?

That’s a huge question, considering how organizations are embracing edge computing as a way to more quickly collect, process, and analyze data for mission-critical processes like AI. Kevin Sheu, VP of solutions and strategies at Versa, a network and cybersecurity company, told IT Brew that, with the rise of AI, the tech industry is seeing “more and more architectures” like AI systems that are pushed out of data centers altogether.

“More and more of AI inference can start to be pushed to the edge of campus and branch locations; and what we might start to see in the future as well, is more and more the training workloads start to be pushed to the edge,” Sheu said.

Ben Brillat, a distinguished engineer for Kyndryl, said that the edge is “definitely an important part of where we see future data, storage and compute and processing happening.” Brillat added that professionals will rely on edge computing to fulfill their organizations’ compute and analytics needs.

“We are going to see an increased need to have that compute available and doing a useful job,” Brillat said. “It is an [absolutely] essential part of an efficiently deployed end-to-end application environment to have that compute there, to do a job locally, and I think that means that we’re going to see growth in that area.”

Why edge computing could be the future. Pankaj Malhotra, head of enterprise networking and cybersecurity business at Ericsson, sees more devices utilized for edge computing as organizations’ rising data demands put more pressure on data centers.

Sending data back to the data center, he added, means an organization “may miss that window in which that action needs to be taken, and that’s where your edge compute comes in.” On-device processing and analyzing data can take place quickly, he added.

Malhotra said that edge computing is really about application workloads shifting to “microplaces” and allowing devices at the network’s edge to take immediate action.

As more devices come onto the network, Malhotra added, the edge will continue to see expansion of applications helping with compute load. The central data center can “never be out of sync” with micro data centers, which means predictable connectivity becomes increasingly important.

Upcoming evolution. There’s no absolute path for organizations to take when it comes to edge computing. However, Sheu said it’s likely that compute will become more distributed, so professionals will ask more of edge locations, increasing the burden on both devices and AI models.

“It can be an opportunity to drive the business forward if we think about our architecture right,” Sheu said. “What you’re seeing now is an opportunity for the data center teams, as well. If you think about how you architect this and how you build it well, it can be a thing that becomes an enabler of the business relative to competitors.”

Sheu said this approach can differentiate a business relative to its competitors, pointing to how the emergence of the cloud allowed some early-moving organizations to speed through their digital transformation faster than rivals.

“We are going to see an increased need to have that compute available and doing a useful job,” Brillat said. “I won’t predict the future to say what the useful job is, but that it is an essential part of an efficiently deployed, end-to-end application environment to have that compute growth there, to do a job locally.”