As telecom networks become more complex, automation will become a necessity, not a choice. Without the benefits of automation and artificial intelligence (AI), it will become impossible for operators to introduce new technologies and compete for customers, let alone maintain and run their 5G networks.
But telcos and their network operations teams are historically conservative when it comes to making big changes to the network so it’s no surprise that moving to a fully automated system makes many in the telecom industry uncomfortable. Instead, it will likely take time for telcos to develop enough confidence in automation and AI’s role in running their most valuable asset -– the network.
However, most telcos are beginning to realize that network automation will inevitably be part of their future and are prioritizing their investment in this area.
According to Appledore Research’s latest network automation report (as reported by TelecomTV), investments in network automation software by telecom operators grew by more than 42% between 2020 and 2022 to $6.21 billion. In addition, Appledore analysts forecast that telco investments in network automation software will continue to grow and be worth more than $17 billion by 2027.
The concept of automation has been a part of the telecom industry for quite some time and the European Telecommunications Standards Institute (ETSI) even developed a tool set called ETSI MANO, which is a management and orchestration (MANO) stack, for operators to incorporate automation into their networks.
But as 5G network functions moved away from virtual machine-based systems and toward more containerized and cloud native network functions, the ETSI MANO stack has struggled to gain traction and today’s automation platforms such as Microsoft Azure Operator Service Manager, are not dependent upon ETSI MANO.
Microsoft last February introduced Azure Operator Service Manager and Azure Operator Insights, two AI for operations (AIOps) services intended to help operators streamline the complexity of 5G networks.
Azure Operator Service Manager is designed to assist telecom operators in managing their network services. It is intended for telecom operators that are in the process of migrating their workloads to Azure and Arc-connected cloud environments.
Many telcos today are using some form of automation, but those efforts are typically focused on single use cases, such as using automation in customer service or billing and revenue management. But few, if any, have started to incorporate automation more fully across their network functions and lifecycle management.
One reason it’s difficult to track is that automation is often called a number of different things. One common term is intent-based networking, which is when a user describes an outcome or the state of the system and the network configures itself to achieve that desired state. Ideally, intent-based networking tools employ machine learning and are closely tied to analytics platforms like Azure Operator Insights, and become more robust over time, making it more reliable and useful. According to Paul Brittain, principal product manager at Microsoft, in the telecom industry intent-based networking typically refers to the management of the underlying service. While it can also refer to a composite of multiple vendors and multiple systems—that type of intent-based networking is more difficult to accomplish but is also more valuable because of its ability to handle more complex situations.
Intent-based networking offers many benefits. For example, it is able to detect if one component of the network is running an old version of software and correct it before the problem cascades. Humans often miss these types of accidental errors. In addition, intent-based networking can reduce manual tasks that can slow down updates. Plus, it offers better security because the information and updates are being correlated in real-time making it easier to identify and mitigate threats and suspicious activity. Finally, intent-based networking also offers better performance. Network performance is improved because network administrators can specify the desired state of the network and the network will work to achieve that state.
Another common term that is often used alongside automation is GitOps, an operational framework that takes the best elements of DevOps and applies them to infrastructure automation. For example, GitOps will use files stored as code and configure those files to generate the same infrastructure environment every time.
Telcos may be slow to embrace automation but once they do start automating their networks many will likely approach it the same way hyperscalers have approached automation —by using a canary deployment strategy.
A canary deployment is a progressive rollout of an application or network function that gives an operator the benefit of continuous integration/continuous deployment (CI/CD). In other words, the rollout typically starts in a lab and then progressively expands until it achieves 100% deployment.
By rolling out automated services in this manner, it allows the telco to see how a particular network function will perform in a controlled manner before rolling it out more widely throughout the network. Some telcos, which may select multiple vendors for their network services, may choose to automate only one vendor, or they may decide to just automate a select market.
By conducting a staged rollout, telcos can quickly correct any issues that may develop without putting the entire network in jeopardy. This is a much more agile approach and can be used for rolling out new functions, new security features and even software updates.
AI’s role in automation
Although many assume that one of the goals of automation is cost savings, in reality, automating the network will probably not result in telcos being able to reduce their head counts. Instead, it will likely just mean that their existing operations staff will be able to handle much more complicated processes without requiring them to expand headcount.
Although AI is being used in many aspects of the network, it does work hand-in-hand with automation. “Automation provides levers and AI provides the insight into what levers to pull,” said Microsoft’s Brittain.
But telcos, which typically are reluctant to hand over full control of the network to AI and their automated systems, can still maintain some control. There are two types of automation: open loop automation provides recommendations for action, whereas a closed loop automation actually makes the decisions. AI-based analysis is typically used to enhance both open and closed loop systems.
How likely are telcos to embrace closed loop automation? Initially, most believe telcos will only use closed loop automation on a few specific network functions but it’s unlikely to become prevalent throughout the network. Instead, it’s expected that telcos will use an open loop system where they have more control.
A top priority for telcos is keeping network control but as they move further down the 5G network path, automating their network functions will likely become a necessity if they want to simplify the complexity of their 5G network.