Lifting functions previously “baked into” hardware up into software will increasingly assist customers and developers to innovate faster, according to Intel networking and edge computing executive Nick McKeown.
McKeown, in an Intel opinion piece on the eve of its 12th-gen processing announcements at MWC Barcelona, said that moving those functions into software opens it up to a much larger population of developers who can then try out their ideas.
“You have handed over the keys from those who build hardware to those who own and operate big networked systems for a living. Only they know how to operate at such scale; only they can write the software to determine how their systems should work.”
For example, the next “killer app” for 5G would drive new technological developments to benefit many businesses, he said.
Ultimately, Intel’s view is that the network itself should become fully programmable, which should make for more agile transformations that can benefit businesses, he said.
“I’ve had a burning desire for more than 15 years to fix this,” McKeown said. “I mean networking broadly defined: in our homes, in the cellular networks, in Wi-Fi, in enterprises and in the public internet,as well as inside cloud datacentres.”
He said that networking had been locked down and determined by standards and OEMs with “little incentive” to change, because this was thought necessary for performance and efficiency.
But this had changed, he added — pointing to Japan’s Rakuten Mobile, which has a virtualised Intel Xeon/FlexRAN based network.
“Rakuten was able to build a 5G network with software on the same infrastructure it uses to offer its dozens of online services,” he said.
Rakuten Mobile’s network has been rated as offering performance with 40% lower capex and 30% lower opex per site than competitors.
Notions like the “metaverse” will also require much more powerful compute and comms capabilities, accessible at low latencies, distributed across multiple device types.
All of this would become far more achievable with a more composable and programmable infrastructure, McKeown said.
“Our customers are already deploying a lot of AI inferencing in their premises at the edge of the network, where they are analysing video as it streams from cameras, to monitor inventory, measure foot traffic and identify manufacturing anomalies,” he said.
“For this reason, we are seeing rapid growth in our very successful OpenVINO inference platform. Pair this with 5G and we see this as the next killer app,” McKeown said.
Intel currently encouraging developers to sign up and attend the oneAPI summit at the next ISC High Performance computing conference in May-June. Submissions for presentation must be received by 4 March 2022.
( Image © 2022 Gerd Altmann/Pixabay )
