How Telecom Italia Built End-to-End Network Automation to Increase Efficiencies and Reduce Time to Market
When you’re a telecommunications service provider, your number one priority is supplying reliable high-speed connectivity to your customers. Whatever you have to do behind the scenes to achieve this is your concern. As long as they get a quality product at an attractive price, your end users will not care how you manage your network infrastructure.
Providing a seamless customer experience can be tricky, even for an established provider. Network standards are continually evolving, but new technology doesn’t always supplant the old. Many providers find themselves struggling to manage new and aging infrastructure at the same time. Moving users to new networks can often be challenging.
Traditional Network Management in Two Countries
Traditional network management is a labor- and time-intensive process that involves manual configuration, testing, updating, upgrading, and deployment of physical and virtual infrastructure. It often requires scripting and working on devices one by one, either across the network or site.
Fortunately, this manual approach is fast becoming obsolete. Network automation is making it easier and faster to maintain, deploy, and consolidate single- and multi-vendor wired and wireless networks. As a result, service providers can accelerate the delivery of revenue-generating services while reducing operational overhead and better allocating human resources.
TIM is the unique brand of the Telecom Italia Group that operates in the market, through the shared value strategy for the company and the community, offering fixed and mobile telephony services, internet, digital content, and cloud services to over 100 million customers in Italy and Brazil.
Our scalable and flexible offerings are available on handheld, mobile, and desktop devices. We provide families, businesses, public services, and government entities with next-generation solutions, including smart homes and other Internet of Things technologies.
Our advanced intelligent infrastructure powers smart cities, and thus improves the quality of life of the populations we serve. From advanced traffic management to environmental monitoring, we are helping to build tomorrow’s green communities.
Too Many Tools
The problem with network management is its complexity. It can be time-consuming and expensive. It also requires the advanced skills and expertise of people working in different departments.
In a manually managed network, the responsibilities of one team often overlap with the activities of another. As a result, any given parameters can be set from multiple touch points, with very different tools. This can lead to configuration errors, but it can also lead to confusion. In our case, we often end up with the impossibility to go back to when and why specific configuration changes have been implemented.
For example, Team A would write a switch configuration section. Team B would come along and perform some troubleshooting activities by analysing that specific section, but they couldn’t assess the reason for those command lines.
Another potential source of confusion is the many network topologies and technologies that we have in place. Our five-year-old network is not designed like our two-year-old network, and it won’t operate the same way. Our ten- or twenty-year-old networks cause even more headaches.
We were working with a handful of different scripts and solutions, many of which were proprietary and known to very few people. Every time we needed to make a change, we had to figure out which of these solutions to use, and to ascertain who knew how to operate it.
On top of this, we couldn’t model a scenario to see if our solution worked. We used to run preliminary tests, deploy the new configuration, and then test it again. If it didn’t work, we had to repeat the process.
All of this took time and effort, and while it may not have been a front-facing problem, it did have an impact on our customers and on our bottom line. We were unable to make dynamic adjustments to service levels, and we could not provide our users with the level of control they needed.
If a government client needed a new VPN, we had to create it on our end. When a corporate client required additional bandwidth for a new tool or for a special event, they had to put in a request. The industry was moving to an on-demand model, and Telecom Italia didn’t want to be left behind.
A Single SDN
It was time to embrace network automation. Doing so would improve efficiency, simplify network management, and shorten our time to market. To do this, we have been leveraging new technologies coming from different network domains. For IP and newer optical networks, we had to build a software-defined network (SDN) platform on top of our physical infrastructure.
The SDN became a shim layer that freed us from having to deal with proprietary tools and multiple Command Line Interface flavours. It allowed us to build a single touchpoint and to use high-level declarative APIs for our network management activities. We used this virtual infrastructure to hide the complexity of multiple topologies and technologies, thus eliminating many of the cumbersome processes their operation and maintenance required.
We no longer deal with a patchwork of systems and technologies from various vendors, when designing automated procedures. Instead, we now have an abstracted view of our entire network. It’s a lot like driving a car. When you’re on the highway looking for the next exit with a gas station, you’re not thinking about your fuel injector or your timing belt. You’re operating at a much higher level.
The SDN empowers us to look at the bigger picture. As a result, we can deal with network issues faster. Our various departments can all contribute their specialized skills within a single environment. Software, hardware, and network teams are all working on a common platform.
The Heart of Our Automated Network
At the heart of our SDN platform and our automated network is Cisco Network Services Orchestrator (NSO). We chose their solution because it allows us to configure network equipment in an intuitive way that makes it easy to roll back new settings if they aren’t working properly.
Cisco NSO uses Netconf/YANG to orchestrate in a flexible simple language implementation, as well as traditional interfaces like SSH and CLI to configure legacy architecture. The flexibility of using a model-driven (YANG) platform to deliver service intent allows support of future technologies and network architecture implementations.
The ability to abstract network layers further streamlines the way we deliver network services. Our network administration teams can now focus on managing our network infrastructure while our network service teams can concentrate on providing custom solutions to our clients, all through a single orchestration point.
Cisco Network Services Orchestrator also takes live snapshots of our existing network. We can look at the way specific elements within our infrastructure are configured, and then model these components to find ways of migrating them to new physical or virtual technologies. We can also use Cisco NSO’s extensive network discovery features to help us plan and deploy network services.
Another benefit of Cisco NSO is its easy integration. We’ve adopted Itential as our workflow management solution. It is a low-code platform that extends and enhances Cisco’s robust tools with purpose-built maintenance automation, service order management, and configuration management tools.
Itential and Cisco NSO work together to provide a single touch point for network automation that crosses departmental silos. Combined, these two platforms speak a language that our operations, service, IT, and network management teams can all understand.
End-to-End Network Configuration
Our eventual goal is to use Itential and Cisco NSO to automate end-to-end network configuration within the IP and SDN enabled transport domain. We would like to eliminate all manual processes, but this will take time. Our long-term goals include software and hardware updates on live networks, along with self-service ordering and activations.
For the time being, we are focusing on the quick wins that we have amassed since we started using Cisco NSO. For example, we were able to automate the migration of customers to a new carrier network from aging infrastructure dating back to early 2000s. We have also established a workflow that automates the switches configuration and the activation test for a Carrier Ethernet service that is crucial for the wholesale market.
Both of these victories helped build trust internally. When our various teams see what their colleagues have automated, they start to look at ways to adopt Cisco NSO as part of their own workflows. One group’s ingenuity sparks another’s creativity. Soon, automation will become second nature company-wide, and we can start to build our bigger initiatives.
As I look back on our implementation, I have some advice for others. The smart approach is to work across your organization. Instead of changing everything from a single point, introduce automation to your various teams, and empower them to create the workflows they lack.
Don’t just tell your employees what you want; allow them to show you what they need. It is far more effective to have them automate existing processes than to create new ones for them. People are far more resistant when you try to effect change from the outside in.
Finally, consider the impact of a top-down change on your customers. If you force automation on your teams, it may disrupt service levels as everyone is in the learning phase at the same time. A gradual transition will help you roll out changes only when they will have no perceptible impact on the end-user experience. And that should be your end goal. Your customers don’t care how you get there—their only concern is that reliable high-speed connectivity. What’s happening behind the scenes won’t matter to them, but it will make all the difference in the world to your business.