Originally posted on VMware Blogs by Stephen Spellicy

Are Yesterday’s Networks Holding Back Tomorrow’s Applications?

If you follow the enterprise technology space, it’s hard not to get excited by the coming wave of digital transformation. Automated factories, drone-based delivery, cashier-less stores, and other applications promise to reinvent businesses, even whole industries. These innovations really are coming, and modern computing architectures will play a key role in enabling them. But the true secret sauce fueling digital transformation: we’re finally building programmable networks. And it’s about time.

Go back to the early days of modern computing. The revolutionary change that launched the Digital Age wasn’t just smaller and more capable computer processors, though that certainly helped. It was the emergence of modern programming—the idea that anyone should be able to write applications to instruct processors how to behave to achieve different outcomes. That shift spurred the PC revolution, the Internet, and everything else in today’s technology landscape.

With the birth of microservices-based architectures, we’ve entered a new era of programmability. Applications can now distribute resources across traditional data centers, cloud, and edge, and call on external services on demand instead of recreating them locally. Yet, when we run distributed architectures over legacy networks, we find this freedom tightly constrained. The problem: we’re still operating under a model where the network defines how applications behave, instead of the other way around.

Now, as Communication Service Providers (CSPs) look to implement end-to-end automation and Radio Access Networks (RAN) we’re about to change that story. For the first time, CSP networks will be truly programmable—by applications, even CSP customers—to deliver different characteristics for different use cases. Businesses will be able to fire up computing resources wherever they’re needed, as a service. And they’ll provision custom connectivity on demand as easily as instantiating cloud infrastructure. It’s a hugely consequential change. And VMware is playing a key role in making it happen.

Envisioning Programmable Networks

Why have legacy networks held back innovation? Because they’re designed for centralized workloads and one-size-fits-all connectivity. This makes them fundamentally incompatible with the vision of cloud-native applications, where logic might be segmented into front-end, mid-tier, and back-end components distributed across many locations. Until now though, we’ve had no choice but to build applications for the network we had—limiting their freedom to scale efficiently, or to tap resources from whichever location could best meet their needs.

Conventional CSP networks support limited traffic engineering through mechanisms like Quality-of-Service Class Identifier (QCI). But these are highly complicated, manual, and difficult to scale. 5G changes the game, enabling network slices that deliver custom connectivity to meet different application requirements, for different customers and use cases, under service-level agreements (SLAs).

For the first time, CSPs will be able to program the network—and allow applications and customers to program the network—from access to aggregation to core, with dynamic workload placement and automation. They’ll engineer networks to gracefully direct different types of traffic across cloud, edge, and on-premises infrastructures, according to policy. With modern approaches like zero touch provisioning, they’ll automate service delivery to reduce the need for site visits and reduce truck rolls. And they’ll be able to guarantee specific paths and quality standards for different traffic types, and monetize those capabilities under SLAs.

VMware Leads the Way

The industry is still in the early days of this transformation, but VMware is already helping our customers take advantage of programmable networks. Through Telco Cloud and RAN innovations, we’re helping CSPs implement fully programmable, policy-driven communications paths across their environments. We’re making it easier to adopt microservices architectures that distribute network resources across data centers, customer premises locations, edge, and cloud. And we’re providing tools to help CSPs program the network effectively at each of those layers. VMware innovations include:

• RAN Intelligent Controller (RIC): Part of the O-RAN Alliance specification, the RIC inserts an open software platform into the heart of the radio network to program how the RAN behaves. This in itself is revolutionary, as previously, only RAN vendors could access the radio control and management plane. With VMware RIC (among the industry’s first), CSPs, and their customers can now run internal and third-party applications to extract real-time telemetry, apply policy and application logic, and shape traffic right at the point of access to the user plane. They can launch new capabilities like real-time optimizations to improve spectrum utilization, or smart resource allocation, such as guaranteeing minimum throughput for every user connecting to the network under SLAs. The RIC provides a cornerstone for network slicing, enabling CSPs to automatically instantiate slices from core to RAN to meet SLAs. It’s important to note that some CSPs are even using our centralized RIC in 4G/LTE networks to observe and optimize the radio network—even without Open RAN.
• Dynamic infrastructure policies: One major challenge with microservices-based architectures is ensuring that containers are configured to meet the requirements of the heterogeneous vendor and virtualized RAN (vRAN) workloads that will run on them. That’s because Kubernetes assumes an operator knows a cluster’s purpose before deploying and that all nodes in that cluster will operate similarly. What if that’s not the case—if you need to customize a node for a specific Containerized Network Function (CNF)? You’re looking at a highly complex manual effort that can take hours, with high risk of errors. VMware has solved this issue with the ability to set dynamic infrastructure policies (previously referred to as “late binding”) part of the automated Container-as-a-Service (CaaS) capabilities we provide with VMware Telco Cloud Automation. The dynamic infrastructure policy capabilities automatically configures cloud resources on demand based on CNF requirements, eliminating the need for most manual configuration and reducing CNF customization time by up to 75%.
• Automated lifecycle management: VMware’s Telco Cloud Platform also provides fully programmable lifecycle management of CSP cloud and network resources. Our customers can automate processes like deployment, provisioning, tear-down, and redeployment across their multi-cloud, multi-vendor environments. This greatly simplifies service deployments, streamlining operations and eliminating manual effort across the network. CSPs can use a single, programmable platform to automate infrastructure and services and make the best use of available resources. Since a founding principle of VMware’s portfolio is providing options to transforming CSPs, we also just announced that VMware Bare Metal Automation will enable instantiation, provisioning and configuration of bare metal servers—offering CSPs holistic lifecycle management of network operations. The result will deliver end-to-end network automation. And with our Ready for Telco Cloud ecosystem, they can add new capabilities and launch new services more quickly, using pre-validated solutions.

Together, these VMware innovations provide a versatile platform to enable on-demand, customizable network services. Just as important, they help CSPs innovate in a sustainable way—with streamlined and unified operations, optimized resource utilization, and built-in support for revenue-generating SLAs. It’s the future of network programmability. And with solutions like the VMware Telco Cloud Automation and VMware RIC, CSPs can start capitalizing on that future today.