Service providers increasingly explore cloud-native deployments

Back in 2012 when telecom service providers worldwide launched the Network Functions Virtualisation (NFV) initiative, their focus was very much on leveraging software, virtualisation, commodity servers and open standards in order to accelerate the deployment of new revenue-generating services while at the same time lowering network operational costs (OPEX). The most efficient way to achieve these goals was seen as replacing existing physical network functions with virtualised instantiations of the same functions, running as VMs under the control of a hypervisor.

Over the past few years, this strategy has resulted in a wide range of functions that were typically delivered as dedicated hardware subsystems becoming available as virtualised software implementations. Traditional equipment manufacturers as well as pure software companies now provide Virtual Network Functions (VNFs) that enable standard telecom functions like Customer Premise Equipment (CPE), Evolved Packet Core (EPC), IP Multimedia Subsystem (IMS), Content Delivery Network (CDN), Radio Access Network (RAN), Session Border Controller (SBC) and others to be deployed as virtualised software applications that mirror the functionality of physical counterparts, now known as Physical Network Functions (PNFs). Telecom folks do like their acronyms.

The deployment of VNFs instead of PNFs has led to definite OPEX reductions, thanks mostly to improved resource allocation and the reduction of operational expenses associated with service installation, commissioning and maintenance.

Over the past couple of years though, service providers have concluded that more drastic advances are needed in order to meet their long-term goals of delivering compelling end-user experiences from highly-efficient operations that are based on modern technologies pioneered by the webscale giants. This has led to an increased focus on ‘cloudification’ as the next strategic initiative beyond virtualisation.

Cloudification is implemented through the deployment of ‘cloud-native’ services which, rather than representing a straightforward replacement of a PNF by a VNF, are functions developed from the ground up for deployment in cloud or edge data centers. Designed for simplicity, flexibility and rapid deployment, cloud-native services are typically implemented as microservices that run within Linux containers orchestrated by Kubernetes, rather than as complex VNFs that run within VMs.

A cloud-native strategy allows service providers to accelerate both the development and deployment of new services, while the ability to rapidly scale up or scale down those services allows for resource utilisation to be optimised in real-time, in response to traffic spikes and one-time events.

In taking this approach to cloudification, the telecom industry is following the lead of webscale companies like Amazon, Facebook and Google, who implement the majority of their applications and services as containers in a cloud environment.

Cloud-native services are also key to ‘Zero-Touch Automation’ (ZTA), which is the term used to describe the automation of IT and data center infrastructure. In the telecom industry, ZTA extends this concept of automation beyond the initial installation phase to cover the entire lifecycle of network operations including planning, delivering, onboarding, monitoring, updating and ultimately, decommissioning of services.

ZTA will move telecom networks from today’s automatic functions to fully-autonomous operations that bring significant top-line revenue improvements as well as sustainable reductions in operational costs.

ZTA will lead to massive improvements in customer experience while service providers will benefit from improvements in operational efficiency and sustainable reductions in OPEX. They will also be better positioned to benefit from the skills of software-oriented millennial engineers and innovative third-party software vendors.

As a leading provider of software infrastructure platforms for telecom networks, Wind River has been collaborating closely with service providers as they refine their strategies for cloud-native deployments. It’s clear that while service providers continue to launch new network services based on VNFs running in VMs, they are also actively starting to explore the deployment of additional cloud-native services that are based on containers.

While many applications will be implemented as containers in line with this cloud-native approach, others will remain as VMs, so it’s important that the underlying virtualisation infrastructure treats both as first-class citizens, co-existing in the telecom cloud.

Recognising this trend, the Wind River Titanium Cloud software infrastructure platform now supports containers in addition to VMs. For maximum flexibility, containers can run either within VMs for maximum tenant isolation and security, or on ‘bare metal’ for minimum footprint. (This bare metal container support is currently characterised as a ‘tech preview’ function as we partner with key customers to confirm exactly what level of functionality will be required for their near-term use cases.)

Future releases of Titanium Cloud will be based on the StarlingX edge compute infrastructure project hosted by the OpenStack Foundation. These releases will leverage the innovative StarlingX roadmap that includes light-weight cloud-native configurations, leading to lower cost and increased agility for edge use cases.

While Titanium Cloud addresses the needs of containers in cloud data centers and edge infrastructure, service providers see additional business advantages in deploying container-based applications in network appliances as well as specific network subsystems such as the Radio Access Network (RAN). Wind River addresses these use cases through containerised solutions based on Wind River Linux, which will be the subject of a future post.

At Wind River, it’s delighted to be contributing to the realisation of the cloud-native concept within telecom infrastructure.

Courtesy of Wind River.