Network as a service is emerging as a critical enabler for enterprises navigating distributed architectures, hybrid cloud environments and the growing demands of AI-driven workloads.

As organizations move applications, data and analytics across cloud platforms, edge locations and data centers, the role of the network shifts from static infrastructure to a dynamic platform that can scale and adapt in real time. Traditional connectivity models, often built decades ago, were designed for predictable traffic patterns and centralized computing. Today’s environments are far more complex, with enterprises requiring flexible connectivity that can rapidly adapt to new workloads, security requirements and business demands.

During a recent conversation with Bill Long (pictured), chief product and strategy officer at Zayo Group, we discussed how enterprises are rethinking connectivity consumption, the growing importance of fiber infrastructure for AI-driven environments and the operational challenges posed by distributed hybrid and multi-cloud architectures. We also talked about new networking models, designed to simplify operations while giving network teams greater agility and control.

The importance of networking in supporting the modern digital economy begins with the physical infrastructure that underpins connectivity. Massive data centers, cloud platforms and enterprise locations all rely on extensive fiber networks to exchange data across regions and platforms, Long noted.

“Zayo is a communications infrastructure provider across North America, and we operate fiber networks,” he said. “So, everyone thinks that the internet’s just magic, but there’s actually physical fiber across the US. We have about 20 million fiber miles, over 147,000 route miles, where we connect data centers, houses, businesses [and] cell towers.”

Network as a service simplifies enterprise connectivity

As enterprise IT environments expand across cloud platforms, data centers and branch locations, managing connectivity between them has become significantly more complex. Many organizations still rely on networking products designed decades ago, even as modern use cases demand far greater flexibility and performance.

While networking has become increasingly important to business success, the technologies used to deliver connectivity have not kept pace with the changing nature of enterprise workloads, according to Long.

“But the network products that the enterprises use to connect those locations are the same products that were really defined around 2000,” Long said. “So, they’re 25-year-old products that really were built for basically passing bits to connect people to people or people to applications.”

In response, newer approaches such as network as a service are designed to rethink both the architecture and operational model of connectivity. Rather than provisioning different services for each location, enterprises can deploy a flexible connectivity framework that allows them to configure and adjust services dynamically as requirements change.

This shift is particularly important in environments where organizations must connect multiple clouds, data centers and branch offices while supporting new data-intensive workloads, Long noted.

“DynamicLink connects all those things,” he said. “So, all locations on the Zayo network and even off-net locations can be connected. And then, using an online portal, you can go and create whatever you want that connection to be.”

These new capabilities are also helping enterprises respond more quickly to emerging technologies such as artificial intelligence. AI workloads often require data to move between multiple environments, including cloud platforms, specialized GPU infrastructure and enterprise applications, creating new demands for bandwidth, security and network performance.

Networking architectures supporting AI will resemble cloud environments but require significantly greater scale, security and performance to meet the demands of modern digital services, according to Long.

“If you think about the overarching network designs are roughly the same, but they’re on steroids, and it requires a rethink of how you do it because your old solutions don’t have the bandwidth, don’t have the security, don’t have the performance and don’t have the control that you’re going to need in the AI world,” he concluded.

Here’s theCUBE’s full interview with Bill Long:

Photo: SiliconANGLE