Hewlett Packard Enterprise is making networking central to its AI infrastructure strategy, with a vision for self-driving networks that can anticipate problems, automate operations and reduce the complexity facing enterprise IT teams.

AI models and agents are placing new demands on enterprise networking as organizations move larger volumes of data between compute, storage and distributed infrastructure. Without sufficient bandwidth, visibility and automation, network bottlenecks can increase latency, disrupt applications and diminish the user experience.

“The fundamental problem underlying the challenges we have in operating data center networks is … complexity,” said Ben Baker (pictured), director of data center networking at HPE. “You got virtual networks, [VXLAN Tunnel Endpoints], alphabet soup of protocols, configurations of thousands of physical and logical devices … the list goes on. We see operators drowning in data but starved for insights.”

Baker spoke with Bob Laliberte, principal analyst for theCUBE Research, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed HPE’s strategy for AI networking and how the company is building its agentic enterprise. (* Disclosure below.)

Building the self-driving network

Baker identifies three problems facing enterprise networks: limited insights into the applications running on the network, insufficient speed or agility, and poor reliability. These ongoing issues mean that network operators are frequently playing catch-up.

“Everyone is just one hasty change or manual configuration error away from bringing down the entire network with huge reputational and financial impacts to the business,” he said. “If you don’t have fast, robust automated recovery and rollback systems, it could take hours or days of intensive work just to restore services. Data center network operators are … firefighting, constantly dealing with these emergencies instead of focusing on proactive strategic initiatives.”

HPE’s response to the networking crisis is a two-pronged strategy: establishing self-driving networks as a foundation for the agentic enterprise and using GreenLake Intelligence — its agentic AI framework — as an intelligence layer across one unified platform.

The self-driving network not only supports networking for AI but also uses AI to operate the network itself. Part of that vision is HPE’s integration of Apstra Data Center Director, a data center network management tool, with HPE Morpheus, a hybrid cloud management and automation platform.

“We’re starting on second base and everyone else is catching up,” Baker said of HPE’s networking strategy. “Our solution is built on top of the Apstra Data Center Director graph database. This is our secret sauce. Graph databases capture and maintain information about the relationships among the nodes in the network. We understand the correlations among all the disparate data coming in.”

GreenLake Intelligence connects networking with compute, storage, security and cloud infrastructure, creating a common operational model. Under this umbrella, HPE draws on tools from Juniper Networks Inc.’s AI-native networking portfolio, including the Mist AIOps platform, which enables users to address potential networking problems before they cause damage.

“Our Mist Networking Data Center Assurance product sifts through this treasure trove of data,” Baker explained. “The key here is being able to connect this data and the specific types of data that are relevant to a problem in a particular component. One of the key insights that the Mist team had early on was that they linked the people in tech support to the data scientists.”

Bringing networking acquisitions under one umbrella

Since acquiring Juniper, HPE’s goal has been to bring together tools from Juniper and Aruba Networks. Marvis, HPE’s AI-driven networking assistant, has been expanded into Aruba Central, and support for HPE Networking CX switches is being added to the Mist platform.

“I heard some chirping out there that the HPE acquisition is going to slow us down; it’s going to distract us. But I can tell you I’ve seen firsthand that it’s the opposite,” Baker said. “All of these different teams from different business units have been able to come together and execute these portfolio integrations. The net of this is HPE is unique in the industry in being able to bring together compute, storage, networking, hybrid cloud, to deliver these full-stack, cross-domain IT infrastructure solutions.”

HPE continues to invest in liquid-cooled infrastructure. The QFX5250 is a fully liquid-cooled switch with 64 ports supporting speeds of up to 1.6 terabits per second. By advancing its networking hardware, HPE aims to meet the growing bandwidth demands of AI applications.

“That insatiable demand for bandwidth from AI, it’s not going away anytime soon,” Baker said. “HPE is going to continue down this path to the self-driving network. We don’t ever forget what the North Star is. It’s all about reliable networking. People still view the network like a utility, like electricity or water. They don’t want to worry about it and they just expect it to work. The ultimate goal here is always optimizing the user experience.”

Here’s the complete video interview:

(* Disclosure: Hewlett Packard Enterprise sponsored this segment of theCUBE. Neither HPE nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE