In 2012, we predicted that Communications Service Providers (CSPs) of every size and in every region would embark on a journey to virtualize, and ultimately "cloudify," their networks. Almost all of them have begun that process in the years since.

While that's a victory for the industry, it's also fair to say that most CSPs haven't gone far in their journey. Those that have are taking the best elements of software-defined networking (SDN) and network functions virtualization (NFV) to develop the first generation of what we call the "telco cloud."  The telco cloud is the internal cloud-based infrastructure that will permeate the entire CSP network, providing a flexible and agile service delivery, operations, and management environment.

As CSPs virtualize network control and functions, it's worth asking if merging or, "cloudifying," is worth the effort. Most of the time the answer is yes because doing business in the Internet Economy requires operating with speed at scale. And that, in turn, requires a degree of network flexibility that too many CSPs either don't have or can't offer at scale. To be clear, the telco cloud isn't an adaptation of existing networks — it's a transformation in which network service delivery and operations are made to be cloud-native and operating like any other cloud service. To achieve this, virtualization is not enough. Virtualization is an important first step since it allows infrastructure resource sharing and operational flexibility. However, if the underlying network remains relatively rigid and static, CSPs cannot take true advantage of the flexibility that virtualization enables. The key to realizing these benefits is to combine the flexibility introduced by NFV with the network programmability offered by SDN.

Let's take an example of the Gi-LAN in the mobile network that sends flows through various functions, factoring in policies and users. In the past, I'd manage these requests with physical boxes. Virtualization has helped to accelerate the process to some degree, but it doesn't entirely eliminate the need for over-provisioning. Worse, once you over-provision you create hardwired connections that are tough to change as circumstances and service demands evolve — and that's a problem because changes will always be needed.

Yet, it doesn't have to be this way. Working cooperatively, an SDN fabric programmed to follow policy can control and automate functions for the Gi-LAN. Adding new services becomes as simple as programming a new policy. This sort of service agility and flexibility are what you get when you bring SDN together with NFV. SDN with NFV accelerates the "cloudification" journey, with more and more CSPs recognizing how important SDN is to NFV.

At the start of the industry movement on NFV, it was clear to many of us that open development, alongside SDN and NFV, was the third pillar that would accelerate the journey to the telco cloud by providing the necessary innovation and development scale. It’s encouraging to see that this third pillar is solidifying today.

Open source projects such as Open Daylight, OpenStack, and OPNFV, all of with which HPE are engaged, contain the SDN and NFV ingredients as well as the connective elements that allow for SDN and NFV to work in concert and with greater flexibility.

As an example, the latest release of the OPNFV software framework not only provides the ability to use SDN control together with virtualized network functions (VNFs) but also allows for the choice of one of three different SDN controllers within an overall NFV framework — based on need. That kind of flexibility points the way to the future.

We have not yet achieved the true cloudification of both network control and of network service delivery. This requires exploding the monolithic VNFs of today to extract the common services and move to a service-based (and eventually a microservice-based) infrastructure. It requires moving beyond the simple separation of control and data planes embodied in the SDN of today to a similar microservice architecture of control functions that can be called on by many different applications and services in different ways.

So the end of the cloudification journey is still a bit further up the road, but what matters is that the journey to the telco cloud is well underway, and we don't need to reach the final destination to begin realizing benefits.