The Role of Automation in Telcos

Automation is not new for hyperscale players such as Amazon Web Services (AWS) and Microsoft Azure. These companies have API toolsets, software automation, and Artificial Intelligence (AI) and Machine Learning (ML) that monitor and drive a cloud infrastructure that includes millions of compute cores. With cloud platforms, there is no infrastructure heterogeneity and that aids in end-to-end platform observability, and by extension, automation. Furthermore, for software giants such as Google, Amazon, and Microsoft, a decision to launch a new service across their global operations does not require a conversation among multiple companies that need to achieve alignment and agreement among themselves.

Telcos, by contrast, are composed of thousands of Communications Service Providers (CSPs) and hundreds of vendors that compete and move sometimes slightly, sometimes vastly, in different directions. Furthermore, their vendors’ component interfaces and solutions come with their own naming conventions, data formats, and proprietary interfaces. This ecosystem diversity creates complexity that does not lend itself well to Internet-like scalability, agility, and automation. This is particularly relevant for 5G network slicing, a promising technology that warrants End-to-End (E2E) life cycle management and multi-domain monitoring and automation capabilities. But with several cellular network layers and domains, monitoring of granular network metrics and statistics is not a walk in the park for CSPs.

5G networks will be built, and based on, fundamentally different network principles, such as 5G standalone, service-based architecture, and network slicing. 5G standalone architectures are born in the cloud and are, therefore, cloud-native and container-based by design. Containers significantly change the way Network Functions (NFs) are operated and onboarded, and the way applications are deployed and monitored. Further, 5G is inherently a distributed system, one where varied hardware and software components are located in different networks, spatially separated by any distance and in entirely different geographies. This is especially relevant for virtual RAN (vRAN), environments where CSPs’ Data Centers (DCs) increase in number from today’s hundreds to potentially thousands, VMs proliferate from hundreds to thousands, and networks shift from tens in core environments to hundreds at the “edges.”

With that complexity in their operations, CSPs will seek capabilities that can automate and manage the life cycle of heterogeneous network components and aid them to transform their operations towards hyperscale capabilities. In general, such automation capabilities will have to meet the following requirements:

• Specifications and Key Performance Indicators (KPIs): CSPs and the provider(s) of automation solutions should know what to specify—which attributes, interfaces, and integration points are core for existing network operations and which ones are complementary.
• Component Verifiability: That enables CSPs to monitor and measure network capabilities and interfaces so that they can verify that operations specifications are met. With multi-vendor deployments, individual monitoring of granular network metrics and statistics may not come without challenges.
• Interface Predictability: From an automation perspective, this allows CSPs to avoid any poorly understood or unpredictable interdependencies across NFs and underlying infrastructure. That way network operations can run with predictable consequences.

CSPs seek solutions that increase network operations efficiency and maximize the business impact of E2E full value chain automation. In telcos, however, agreement and alignment are required among multiple vendors to establish functional interfaces to do life cycle management and automation of heterogeneous network components.

CSPs are in full swing in terms of building new cloud platforms and agile, on-demand connectivity services, especially for enterprise verticals. But, telcos are not an agile ecosystem by design. Consequently, 5G opportunities require that CSPs accelerate innovation at the pace of software. The right question for CSPs to ask when adopting automation solutions is not whether it can support multi-vendor deployments and/or brownfield environments. Instead, automation products that cast the back doors associated with the specialized, hardwired telco processes and systems stand to be key pillars for a more agile telco value chain. To that end, CSPs should aim to achieve cohesiveness in their operations as they seek to implement automation capabilities. It is critical that technology works well regardless of the vendor that it is coming from. For technology providers to fuel cohesiveness in a complex and heterogeneous ecosystem, they need to be a cloud vendor, a core vendor, a mobile vendor, and a fixed vendor. They should think about the evolution of cellular networks in a holistic manner and seek to implement automation in an end-to-end fashion. For example, hyperscalers are already partnering with CSPs to help them implement public cloud platforms and add agility to current operations. On the other hand, companies like Canonical, Mavenir, Red Hat, and VMware will also play a key role to help CSPs simplify automation and management of their (hybrid) cloud and cellular environments. The role of cloud platforms in telcos is a topic that will be explored further in an upcoming new research report from ABI Research.