Automation continues to be a key strategic pillar for communications service providers (CSPs) to improve operational efficiency and reduce time to market for new products. For example, Telenor, working closely with Enea and other vendors, recently developed an innovative End-to-End (E2E) automated deployment solution in the 5G core domain that promises to increase service deployment speeds by 70%. Deutsche Telekom highlights that 70% of tickets in its NOCs are automatically dispatched. As a result, vendors are investing significantly to compete to enlarge their slice of the automation market. Global technology intelligence firm ABI Research forecasts overall revenue from AI-enabled automation product sales to grow from just under US$ 15 Billion in 2023 to US$ 35 Billion in 2028 at a Compound Annual Growth Rate (CAGR) of 19%.
Today’s maintenance solutions are statically defined solutions often isolated from the applications layer and underlay infrastructure. “So, the traditional assurance and automation approach, involving several siloed teams, systems, and processes, may not be a robust way to handle maintenance. Network healing is predominantly manual, with the sequence of remedial steps considered tacit knowledge not able to be codified. Intelligent decision-making, prediction and analysis, and event execution are mostly manually conducted by network engineers. In this case, network maintenance is reactive. This siloed and manual approach to fault identification and remediation is ill-equipped to handle the complexity of large volumes of 5G network data, network events, and Key Performance Indicators (KPIs),” explains Don Alusha, Senior Analyst at ABI Research.
Complexity may well be the hallmark of the industry for the next decade or two. Cloud software and edge computing will create networks that cannot be assured with statically defined point solutions. Virtualization, for instance, expand the scope of automation. Each stack layer is an individual ecosystem with its own objects and properties, consumable resources, and lifecycle management states. In a diverse 5G ecosystem spanning physical, virtual, and cloud architectures, there will be thousands of KPIs generated every few seconds. Alusha says, “In fact, it is reported that there are more than 400 network procedures in 5G, each of which has its dedicated KPIs running against those procedures. This means that end-to-end instrumentation for automation must traverse a much more complex environment than before.”
CSPs vary in their automation maturity strategies. Some network domains can be fully closed-loop with no human intervention. But for other networks, the risk is considerably higher. For example, in the access network, changes are predictable, and there are hundreds of thousands of endpoints. There is little or no risk of one end-point fault propagating to the wider network. So, the RAN domain is considered a good first step for automation. Amdocs and Netcracker, for example, are among a growing group of suppliers that target RAN, transport, and edge DC domains where a considerable number of tickets can be automatically reduced to optimize overall operations. By contrast, in the core network, a single change can impact thousands of subscribers, so it must be cautiously approached.
Suppliers that are likely to dominate in the cloud automation market are those that help CSPs shed the back doors associated with existing siloed and disjointed networks. “Specifically, three dimensions are key for CSPs when selecting an automation partner: one, scale and correlation – it remains complex to assure services and correlate hundreds of thousands of KPIs across multiple NFs being generated every few seconds; two, big data expertise – for machine-guided automated troubleshooting and maintenance, big data expertise is indispensable, but so is domain knowledge around key data relationships within RAN, core and even transport; and three, full and correlated visibility – the ability to offer full and correlated visibility across network domains, across all stack layers up to the application/device layer is a key selection factor,” Alusha concludes.
These findings are from ABI Research’s Automated Service Assurance in Cloudified Networks application analysis report.
