Automation goes mainstream
The pandemic nearly wrong-footed service providers. While their networks, thankfully, kept up with the dramatic shifts in traffic patterns, it was often a close thing. Much of the developed world sheltered at home and telecommuting went from being an occasional thing to normal behavior. Digitalization plans were fast-tracked by many enterprises, and video conferencing realized its decades-long-delayed promise. Meanwhile, 5G launched and mobile usage accelerated.
These events have injected a sense of urgency in the industry. IP network operators now understand that many of their current complex operations need to be streamlined. From where I am sitting, automation is going to be the near- and medium-term focus, and there are a few things to watch for.
Automation goes mainstream
Hyperscalers lead the industry in terms of automation in their data centers, but within IP networks generally, the adoption rate both internally and customer-facing is also set to increase. Operation teams are struggling to deal with the volume and fluctuation of activity. As digitization picks up steam, everything is more consumable. Shifts just happen faster, especially in a world that has grown accustomed to instant gratification. Network operations, as a result, need to provide a higher level of responsiveness.
Several studies indicate that there are strong business benefits to adopting automation across the WAN and data center networks. A study by Analysys Mason shows that operators avoided up to 65% of their costs after implementing automation across all operations categories in the WAN. Our Bell Labs Consulting group have also published a Business Case Analysis for data center fabrics showing up to 40% cumulative effort savings for all operational tasks in the operations life cycle.
5G and Industry 4.0
Hyperscalers have conditioned customers to expect frictionless consumability — you get what you want, when you want it. These expectations are driving many enterprises to offer the same consumable, instant gratification services to their customers. And, with the focus on enterprise agility and the embrace of Industry 4.0 use cases, this means that network service providers must provide consumable network services to support these B2C and B2B enterprise applications. Automation is a key enabler for providing flexible and on-demand consumability such as Network as a Service (NaaS).
Enterprise customers pursuing Industry 4.0 are also very interested in 5G slicing to meet the needs of their more demanding use cases. Slicing can provide precise SLAs to meet application needs for low latency, high bandwidth or support for low-powered IoT devices. To meet these SLAs, however, a service slice delivered by 5G will also have to be supported by the IP and optical layers end to end. There will be no practical way to do this without automation.
Automation as a product
The biggest obstacle to automation is often the people and processes behind the scenes. The failure to automate is usually rooted in organizational culture. Approaching automation in a manual, human-based, and more risk-averse manner fails. Automation is a product, and as such, organizations must learn to trust it. Network operations staff are used to configuring devices using Craft interfaces and CLI commands, and this is their comfort zone. They must shift from manual work, however, to trusting the pipeline.
At the end of the day, when network operations teams build automation workflows, they are building and operating a product. Hence operations have a lifecycle to it. Things evolve and require maintenance. ‘Infrastructure as code’ uses the same methodology as today’s continuous integration/continuous design (CI/CD) development processes.
Infrastructure as code
Using an infrastructure-as-code type platform allows teams to describe in a declarative way their demands within the network, such as setting up an EVPN or making a connection from A to B. A digital twin can provide the sandbox to test topology, configuration, and the general state of the network, without impacting production.
Another important aspect is the openness of the platform. The network is broken into multiple domains, each with its own specific requirements and constraints. Operators will need tools that are programmable so that they can more easily adapt to these varying domains. Additionally, they should look for vendors to build their platforms or products in a way that operators can use according to their needs and their requirements.
Synergies between network and IT tools
The automation journey is one where the network and IT merge. Although the tools used in the two domains are different and each side has its own domain expertise, the move to 5G and digitalization is bringing them closer together. We can see this on the CSP side in the transition to a more NaaS type of offer and, on the IT side, to a shift in how they view the data center fabric and the recognition that they need to think about NetOps as well as DevOps. Network and IT tools and teams must come together to deliver the experience. And CSPs should look for vendors that have adopted this consolidated approach.
The automation journey
Automation is still a journey, and it depends on the maturity of each organization. What I am seeing is an increase in the number of discussions we are having in this space, and the huge benefits of automation when done right with the appropriate technology. Service providers are improving their maturity levels as they journey towards a fully automated life cycle. And IT are beginning to understand that compute and storage are not everything and that network resources are critical to delivering their consumable experiences. As their worlds merge, automation will be the bridge.
Nokia network automation solutions help communication service providers, webscale companies and enterprises build more programmable, responsive, efficient and reliable IP and optical networks.