WHITE PAPER The Five Phases of Virtualization: From Hardware Rigidity to Web-Scale Flexibility It s a harsh reality that today s networks are still burdened by hardware-based rigidity. At the same time, there is an increasing disparity between revenues and costs as well as limited flexibility in the utilization of network resources. It has all hampered the service provider s ability to provide new applications and services to match the sharply increasing demand. That demand is so sharp, in fact, that even by fairly conservative counts, the number of internet-connected devices is predicted to reach more than 20 billion by 2020 i. Providing profitable services on this scale becomes not just a question of technology, but also, and more importantly, one of economics. To shift from the traditional business focus on ARPU to one of efficiently delivering data at the lowest possible cost per bit, operators must focus on future-proof technologies that reduce cost and maximize capex resources. As the industry grapples with the deployment of next-generation communication technologies, operators and enterprises have already realized that virtualization is the key to upgrading their wireless networks. Gone is the reliance on the hardware-based strategies sold by the large, traditional network equipment suppliers. Utilization of NFV (network functions virtualization) and SDN (software defined networks) cloud-based architectures is the new paradigm. Only through virtualization can operators achieve the elasticity, service agility, and ability to slice the network into the many different networks required to serve the myriad new use cases demanded by the next technology turn. In a new survey of global operators by Heavy Reading ii, however, it was shown that, while it is generally acknowledged to be a required transformation, there hasn t yet been much progress made towards that goal. Only 32% of the respondents were actively implementing SDN and Cloud technologies, while a full 62% were still in the early stages of planning and testing. The number of internet-connected devices is predicted to reach more than 20 billion by 2020. 1
Virtualization, however, needn t be an overlay of a completely new and independent network technology. It can, instead, be a process of transition that leverages 4G infrastructures. We have identified five key phases, beginning with software independence and ending with what we call web scale. When complete, the operator will be able to offer a wide variety of profitable applications and services quickly delivered (or terminated as needed) at the truly enormous scales we will soon see. Because it has become possible to virtualize every element of the wireless network (except for the antenna and the radio itself), operators will be able to efficiently deploy networks not only global in scale, but also microsized all the way down to an individual enterprise. White-box, or open-platform hardware with a vepc (virtual evolved packet core) running on x86 computer cores makes this possible. Below, we look at the five-steps required to make a profitable transformation. Figure 1: The Five Phases of Virtualization Phase One: Software Independence Although the first phase of the transformation has already been made in many networks serving mature markets, it is still important to make the distinction. Involving the separation of the hardware from the software, the use of bare-metal servers i.e., hardware appliances become possible. Software functions become discrete elements in their own right, shifting the model from vertically integrated hardware appliances to non-integrated stacks that can run on any compatible hardware appliance. 2
Phase Two: The Virtual Machine The first attempts at virtualization involved taking software that was already running on custom hardware, putting a wrapper around it, and calling it virtualized. In reality, all that really happened was an update to the operating system that was still running on the same custom hardware. Many equipment providers are stuck at this early stage of virtualization, with attempts to move toward the next phase hindered by traditional business models reliant on hardware sales. To make the leap to this phase, each network function must be fully isolated. This means that each piece is packaged in a standardized way, enabling the full portability of each function. In this way, the discreet functions of the network are truly virtualized so that they can share an instance of any operating system running a virtual machine. Isolation of the network functions is the first true step towards enabling full platform independence. Phase Three: Implementing Infrastructure as a Service (IaaS) Moving from platform-independence to a massively scalable architecture, introduces infrastructure as a service (IaaS). Now, the entire instance can run on commercial off-the-shelf (COTS) hardware that is literally plug and play, enabling virtual functions to be rapidly scaled up or scaled down as required by demand and usage. The focus of this stage is on orchestrating the various elements, automation of the scaling, and the healing of monolithic network entities. Phase Four: Moving into the Cloud Once IaaS has been achieved, the natural evolution of the architecture is to move it into the cloud. The key aspect of this stage is the breaking up of monoliths into micro-services, which brings in service agility, and along with containerization, agility at scale. Faster responsiveness in the network s day-to-day operations and much greater flexibility are the main benefits of moving the platform into the cloud. Recent moves by some operators led towards the virtualization of a range of different network resources, components and functions. Micro-service, cloud-native virtualization allows VMs and containers to be spun up anywhere on a network, at any time, with a smaller footprint. Operators can now deploy, configure and manage virtual network functions in the core, edge or in-between, depending on the use case. Management and orchestration (MANO) tools are now critical success factors for operators with a fully cloud-enabled platform. 3
Technological benefits, however, are not the only benefit of cloud-based networks. This phase opens up entirely new business models for the operator as well. For example, should the operator decide to focus on customer service and not technology maintenance, the entire platform can be moved onto a third-party hosted service. Phase Five: Making it to Web-Scale In the final phase, web-scale, fully virtualized service-automation architectures are realized. The programmable network functionalities feature automated scaling and self-healing intelligence, allowing operators to deploy virtualized networks at any scale, and at speed and agility levels currently seen only at a very few companies today. What makes web-scale dramatically different, and exponentially more economical, are the automation and orchestration features. Resilience, performance, scaling, and coordination between all elements of the network even across different network slices is fully automatic. Upgrades and fixes are delivered in real time, requiring fewer maintenance and service resources, while also delivering vastly improved performance. A significant indicator of this performance jump may be seen in the graph below, which highlights the revenues per employee of web-scale companies Google and Facebook against those of progressive carriers and the global average. Figure 2: Web-Scale Companies, Revenue per Employee, Source Deloitte iii, Mavenir While the economics of virtualization are complex, and early-phase deployment does not always translate into immediate return on investment, the long-term cost savings are very clear. It is in this fifth phase that the traditional mobile network economics shift from a telecom model (a limited scale, even with millions of users) to a web-scale model that can serve billions of users. 4
Consumer behaviours and demands will continue to change and evolve and competitors will come from other industries, so it s important for providers to employ advancements that not only support current trends but can also be quickly changed and adjusted for future expectations, which will include exponential increases in demands for mobile data. To escape the bounds of the hardware-defined platforms of the past, CSPs must transform the economics of their mobile networks to dramatically lower opex and minimize capex. Only an end-toend, 100% virtualized network, evolved into a fully web-scale platform can meet that requirement. It may seem daunting initially, but it doesn t have to be, if it s taken in measured phases. Because, to remain competitive in the hyper-connected world to come, it s a transformation that has to be made. About Mavenir Mavenir is purpose-built to redefine mobile network economics for Communication Service Providers (CSPs). Our solutions pave the way to 5G with 100% software-based, end-to-end, Cloud Native network solutions. Leveraging industry-leading firsts in VoLTE, VoWiFi, Advanced Messaging (RCS), Multi-ID, vepc and Cloud RAN, Mavenir accelerates network transformation for more than 250+ CSP customers in over 130 countries, serving over 50% of the world s subscribers. We embrace disruptive, innovative technology architectures and business models that drive service agility, flexibility, and velocity. With solutions that propel NFV evolution to achieve web-scale economics, Mavenir offers solutions to CSPs for revenue generation, cost reduction and revenue protection. i Gartner press release, Gartner Says 8.4 Billion Connected Things Will Be in Use in 2017. ii Heavy Reading 5G Operator Survey, November 2017 iii Monitor Deloitte, To be or not to be, the future of the telco business model, Issue 07/2017 5