Author: David Sliter, Vice President & General Manager of Communications Solutions Business, HPE
The media has been buzzing about the state of network functions virtualization (NFV) in the communications service provider industry—including some criticisms that NFV has not lived up to expectations. It’s true that the industry is still figuring out the best way to capitalize on virtualization. And there are opportunities for all stakeholders—NFV solution vendors, communications service providers (CSPs), developer communities—to learn what works and what doesn’t. This process will unfold over the next few years. But this shouldn’t come as a surprise; it’s just what any major technology transformation looks like.
As we navigate potential roadblocks and learn as we go, let’s be clear: cloud and virtualization represent the future of service provider architectures. The benefits are simply too great, and the push from CSPs themselves too strong. But, because operator environments are so different from enterprise IT environments, this will be a journey, and we should understand what to expect.
Virtualizing CSP Versus Enterprise Environments
Why has the CSPs’ embrace of virtualization been so complicated? Why can’t we just take enterprise cloud architectures, throw them over the wall and into telco environments, and get started? Because service provider environments are a very different animal than enterprise IT.
First, CSP architectures must meet demands for “carrier-grade” resiliency, performance, and scalability that simply don’t exist in the enterprise IT world. After all, most enterprise applications don’t fall into the category of life impacting applications with the same criticality of an E911 service call. Telco services and network functions also tend to be far more distributed than enterprise resources.
Additionally, in enterprise clouds, most applications are effectively self-contained, so inherently well suited to virtualization. They have few if any dependencies on other enterprise functions and run basically the same way regardless of where they’re hosted. In CSP environments, any new software (especially software used to create new virtualized network functions) must address dependencies with a wide range of running services and systems, including legacy OSS/BSS, and interoperate with equipment from many different vendors.
Finally, telco applications are much more mobile. Where enterprise applications can be ported to the nearest cloud and remain there, telco applications constantly migrate in time and geography. This means the network must process huge amounts of state information in real time. This results in CSP cloud and virtualized service configurations that are orders of magnitude more complex than in enterprise environments.