Interest in automating network operations is growing rapidly. We are working with many customers to help them automate. Small steps in some cases, big ones in others, but ambitions for the long term are almost always extensive.

As these automation projects progress, we are getting a clear picture of the benefits, the pitfalls, and the future of automation.
 

History of network automation

The move to automate network operations began with webscalers two decades ago. While they proved that the IT networks behind their cloud operations could be automated, the telecom industry and enterprises at large weren’t that motivated. That changed as the scale of networks grew and the cloud arrived.

My team is focusing on automation of IP and optical networks, around service performance, assurance, and optimization. We have thousands of customers around the globe, ranging from the largest communications service providers (CSPs) to verticals like transportation, energy, and public sector customers.

Automation drivers

One of the universal automation drivers is the desire to be more responsive and agile. Cloud-based applications and services are dynamic by nature, making it much more difficult to predict what connectivity and SLAs will be needed. New businesses and new categories of services seem to be launched daily. So much more of our lives is lived online, especially over the last two years with the pandemic. We have grown used to and expect instant gratification. As we grow to depend on network-based services and products, they become more critical. The ante keeps getting upped for network operations.

This is all happening while revenues for telecom services remain flat. Increased expectations about agility, scalability, and higher service levels are all occurring against a background of total cost of ownership (TCO) concerns. That means automation is also about finding ways to do things for less.

Business benefits of automation

Fortunately, the news is good on that front. Having passed through the first wave of automations, we no longer have to rely on our intuition. The data is coming in, and frankly, it’s making the business case. A recent Analysys Mason study shows that automation has reduced labor time by 68 percent. This is operator data from networks where they have implemented automation projects in three areas: service deployment, network lifecycle management and assurance.

In terms of service delivery, the study showed there is an 88 percent savings in the time it took to define and introduce new services, which is a significant improvement for time to revenue. On the assurance side, error processing and mean time to repair improved by 70–80 percent.

First step: Prioritize

When it comes to kicking off an automation journey, CSPs have choices. Automation can save money, improve network and service quality, and increase responsiveness to customer demands. I recommend the CSP decide which they want to prioritize first.

The biggest cost savings area is life cycle management. Simple upgrades, backups and restores, rollout of new devices: these are very labor-intensive tasks. Du Telecom in the Middle East dropped the hours needed to upgrade a single router from eight hours to one. In a single maintenance window, they can now upgrade four routers. Anyone running OAM can’t help taking notice of these kinds of automation numbers.

If the priority is improving network and service quality, assurance should be the focus. A customer in Australia finished an automation project in 2020 that focused on alarm triage, troubleshooting, mean-time-to-repair, identifying new causes of outages and similar processes. Two years later, they estimate their savings to be 80–90 percent.

For those who prioritize the need to position themselves for the future by being innovative and more agile in responding to customer demands, this is arguably the most important thing that automation can do for them. Especially if they want to compete with hyperscalers and webscalers that tend to be very good in this area. Here we are seeing a real interest on DevOps and NetOps working together. Automation will be key to ensuring that the continuous integration and continuous development (CI/CD) pipeline automates not only application and service development, but infrastructure-as-code too.

This is where we are focused on developing a programmable and open platform. You cannot expect automation to happen in code that's released as a new software release. It must be code-pluggable with workflows, intents and adaptors that are plugged into the network run time so the platform can be quickly updated as needs evolve, new services are introduced, or when new elements of the network are deployed.

Complex network operations

We're talking about very sophisticated and complex environments that include cloud and edge clouds, decomposed and distributed network functions, using opensource tools and a very disaggregated environment. To put it all together it's quite a complex problem.

For our biggest customers, providing them with the programmable platform, tools, and highly capable software is enough; they have the software engineers to run with it. At the other end of the spectrum, we have many smaller customers who don’t have that kind of talent in-house and it’s hard to acquire and retain it. We help them to build automations working from a catalog of use cases using our own engineers.

Automation is a journey

We are working with our customers today to automate assurance, optimization, full service and network lifecycle. As a vendor and as a network automation team, our mission is to help our customers along the automation journey.