Nokia is digging its heels into the data center market with a new proof-of-concept (PoC). Unveiled at ONUG 2023, the PoC leverages the vendor's network operating system (NOS) to integrate generative artificial intelligence (genAI) and improve data center network operations.

"The timing is really perfect for us to fully enter and conquer the data center market space," Senad Palislamovic, Nokia enterprise head of technology, told SDxCentral.

The PoC demonstrated the use of ChatGPT at the command line interface (CLI) level, allowing network operators to ask questions in a chat format about configuring a VPN or a BGP for a certain type of interface, for example.

As a new entrant to the data center space, "we are last integrated into the market," Palislamovic said. Existing data centers have likely been using equipment from Cisco, Arista Networks, Juniper Networks or Dell Technologies, meaning "network engineers' familiarity with Nokia's CLI is effectively nonexistent," he said. "This was one of the first problems to be solved."

By integrating genAI with the vendor's NOS, Nokia aims to insert itself into the current data center ecosystem. "ChatGPT comes in handy here," he said. "All of the users that are going to be deploying our data center infrastructure will be leveraging that embedded ChatGPT ability inside of SR Linux to learn to config because they're not familiar with that."

Nokia sees its pursuit of this market as twofold: building data center infrastructure optimized for the AI landscape and using the AI/machine learning (ML) ecosystem itself to "ensure smarter and more streamlined network operations," Palislamovic said. "We eat our own dog food."

As a global company with solutions spanning wireless technology, optical networks, 5G connectivity and data centers, "we ourselves went through the pain of building large clouds and large data centers, and we've learned a lesson on top of that journey," he said. "Our approach to [data center] solutions is really the product of that journey."

To build anything high-performance that keeps the competition on edge, "you have to focus on outcomes and solving the real problems," he said. And cost-effective throughput, while a significant factor in data center connectivity, is not the problem that needs to be solved. "We have chips to deal with that across the entire industry."

From the perspective of AI/ML in the data center, the problem is operationalizing those chips and making them useful, Palislamovic said. Nokia's NOS is designed to be as modular and as open as possible to allow customers to address operational challenges with design, deployment, ongoing maintenance and lifecycle management.

"We brought in a truly open Linux space, ... built with cloud-native tools designed for declarative, intent-based automation that can help enterprises modernize and automate their clients' deployments and day-to-day operations," he said. This cloud-native focus takes advantage of open frameworks and microservices to help enterprises deploy "network-like approaches to increase efficiency at all phases of their [data center] fabric operations," Palislamovic said.