Twenty years ago, practically every service provider was asked, "What's your Internet strategy?" A decade ago, it was, "What's your mobile strategy?"
Today it's, "What's your cloud strategy?" Telco Transformation recently spoke with Sorabh Saxena, AT&T Inc. (NYSE: T)'s senior vice president for software development and engineering, about why his company and so many of its peers are migrating to cloud-native networks. By disaggregating systems into smaller, independent functions, we can scale up only the function that is needed rather than, in the past, the entire app, which leaves precious capital stranded. By designing applications in this manner, AT&T and other operators are more nimble and capital efficient.
AT&T has been taking an application view rather than an infrastructure view of cloud computing for several years. We are, and have been, building re-usable business logic and data services that are easily able, via a single API platform, to quickly deliver needed business workflows. We view the recent shift to cloud-native designs in the marketplace as validation of our past efforts.
TT: What does this trend mean for operators' IT support systems?
SS: Cloud-native computing delivers on the promise of cloud computing by decoupling software applications from underlying hardware infrastructure. It also enables scalability of specific business and technical functionality instead of forced scaling of entire systems when only one element becomes constrained.
As a result, IT support systems must pivot from managing infrastructure elements to managing units of software-enabled functionality. Bringing legacy applications and technology in the cloud is a multi-phase journey. Not all applications will travel the same path or realize the same level of benefit. The embedded investment must be connected to the new approach via integration via APIs to allow for incremental progress along the journey. APIs allow for change to occur in isolation, allowing for incremental progress without requiring a big-bang approach.
Establishing software infrastructure to virtualize and manage hardware platforms and the building, deployment, execution and management of container-based applications at scale are key priorities for IT support systems. Open, standards-based components are vital. AT&T is driving these standards in multiple open source and special interest communities. We are shifting our focus from traditional paper-based standards to working code-based standards, placing application and infrastructure management closer to the keyboard and accelerating time to implementation.
This effort will require not only system transformation, but also human skills transformation.
TT: What does this trend mean for the companies that operators work with?
SS: From a customer standpoint, by developing cloud-native applications, AT&T can more quickly provide new services by re-using discrete micro-services. Customers will also benefit from the operators' ability to more efficiently scale to meet customer demands.
From the perspective of a company providing systems and network components to operations, it is imperative that these providers adopt cloud-native designs. For example, in our SDN effort, we want network vendors to provide discrete, independent cloud-native functions that we can scale independently and efficiently as needed. If these providers simply port from firmware to software, operators will not reap the SDN or cloud-native benefits.
We need our system and network component vendors to partner with us in open source communities. Their business model needs to shift to more service integration and hardened distribution or open source components. Their expertise needs to be applied in an open rather than a closed environment.
To advance this goal, AT&T is participating in and contributing to many open source communities. Cloud Native Computing Foundation, Open Network Function Virtualization, OpenFog (Internet of Things), Open Daylight and OpenStack are just a few.
TT: What is the relationship between cloud native and popular new technologies such as SDN, NFV, containers and virtualization?
SS: Simply put, these four terms refer to a "what" and a "how." SDN and NFV refer to specific communication and network functionality. The SDN is personified by flexibility and a departure from expensive and complex physical equipment. VNFs refer to the building blocks [that] are now implemented in software and run on generic equipment, [with] specific examples being virtual switches, routers and firewalls.
Containers and virtualization are part of an evolving methodology and approach, jointly referred to as cloud-native computing. The act of virtualization refers specifically to the ability to establish virtual representations of physical hardware. Virtualization is a specific reference to the ability to abstract a piece of hardware to software applications and their developers. Virtualization requires the addition of a software layer on top of physical hardware to emulate multiple hardware environments.
While virtualization applies to infrastructure, containers enable individual business and technical functionality to be run and scaled independently of others. In addition, containerized functionality is described one time by the developer and can run across multiple environments and technologies. The act of containerizing an application component requires a set of standards and supporting runtime environment to function.
The hierarchy of acts begins with the virtualization of hardware infrastructure. A container management framework is established across the virtualized infrastructure, creating a cloud-native environment. Once established, VNFs may be deployed using containerized application functionality.
Lastly, SDN is established by orchestrating and managing complex, integrated network services. Separation of concerns across the layers of functionality assure appropriate span of control, while a specialty network platform like ECOMP assures all components work together to achieve performance and reliability expectations. Industry standards organizations, which AT&T actively supports, are establishing open standards for each of these integrated tiers of functionality.
— Tim Kridel, Contributing Writer, Telco Transformation