When Humana needed to transform its fragmented network data from scattered spreadsheets and tools into a comprehensive, automated system, the healthcare insurance provider turned to Network to Code and its Nautobot open-source platform to overhaul its approach to network infrastructure management.
Humana needed a strategy to automate workflows and reduce manual efforts around specific tasks, says Matthew Schwen, associate director of network automation engineering at Humana. He took the role to help reduce long development times, organize scattered and uncoordinated scripts, and foster an understanding of network automation for end users.
“The company didn’t yet have a network automation strategy. There were scripts and tools essentially spread out all over the place. They hired me to build the strategy for network automation,” Schwen says. “There were a lot of scripts that were run from local workstations, which were taking engineers a lot of time to build—weeks to months—and then the time it took to deploy and utilize. No one had a single view of what all the scripts were doing and how they were organized.”
Establishing a single source of network truth
To kick off the automation effort, Schwen interviewed 15 to 20 teams to analyze their existing infrastructure—which led to him recognizing that Humana needed a centralized, current source for all of its disparate data. With multiple disconnected databases, existing data was scattered across Excel files, third-party tools, and other drives.
“Each team had their own Excel file or their own access database or their own something containing data that we had to pull from. I don’t even remember how many sources,” Schwen says.
Schwen decided to partner with Network to Code due to his prior experience using NetBox, an open-source platform used to document network devices, IP address management, data center assets, and other components. At Humana, Schwen recognized the need for an extensible, automated approach. Network to Code had previously been a major contributor to NetBox, but later Nautobot further to meet the evolving needs of network automation teams, the company says. Network to Code supports Nautobot as both an open-source project and a commercial product, providing services around customization, integration, and automation workflows.
The Nautobot platform consolidates information from various sources —including other management platforms, configuration management databases (CMDB), and IP address management tools—and aggregates it into a single repository, which then acts as the authoritative source for network automation and management. This single source of truth is critical for creating a “true self-service network automation strategy,” Schwen says, and once it’s established, automation can begin.
“Automation in itself, if you have tunnel vision on automation, the tech is actually pretty easy. It’s been around for a long time, and there is a lot of technology out there for it,” Schwen explains. “What’s hard is the people and the processes. That’s more than half the work.”
Having a centralized platform allows for consistent onboarding, change requests, and deployment processes across the entire network infrastructure. With Network to Code and Nautobot, Humana was able to:
- Create a uniform platform for network requests.
- Standardize cloud deployments.
- Eliminate the need to ask different teams about network information.
- Provide a consistent location for anyone interacting with the network.
“We wanted to get to the point where we had kind of self-service, where anybody who needed resources from the network side, or interactions from the network side, there was one place to go and they knew where that was,” Schwen says.
Schwen explains with Nautobot, there is a workflow that is running on a schedule, and it will pick up any network changes and compare them to the configuration standards for that device. It will auto-remediate any discrepancies found in the configuration of that device. The network operator just has to “add a device to the database and walk away, because everything you need for that device is done,” Schwen says. Network to Code’s recent release of NautobotGPT, an AI-powered assistant to the platform, also helps by accelerating network automation workflows.
“One thing that is incredibly useful is automation tasks as prompts. We took existing workflow automations that we had and turned the objectives into prompts. We were able to recreate those in GPT and automatically deploy specific workflow systems,” Schwen explains.
Reaping the rewards of real-world automation
At Humana, network automation delivered substantial efficiency gains across multiple operational areas.
Schwen quantifies the impact in hours saved. He notes that approximately 1,000 hours have been saved monthly through strategic automation. For instance, in tier-one support alone, processing 400 monthly tickets with an estimated four-hour reduction per ticket translates into significant time savings.
Administrative tasks also saw improvements, with senior engineers reclaiming between 80 and 100 hours per month through automated reporting and data collection processes. By automating complex workflows such as circuit maintenance notifications and creating self-service network request systems, Humana was able to transform time-consuming manual processes into streamlined, efficient network operations.
The automation strategy also extended beyond pure time savings, enabling tier-one support to execute network commands independently and freeing senior engineers from attending low-level support meetings. These improvements allowed network professionals to shift focus from repetitive tasks to strategic network architecture and innovation projects, reshaping how Human’s network team operates.
“In a lot of cases, you don’t even want to work on the automation because you don’t know how to automate. You need to convert processes into automation. You have to understand the processes first, and you have organizations that don’t know the process so you spend a majority of your time documenting the process. Then you convert that into automation and that part’s usually easier than understanding the process,” Schwen explains.
Source:: Network World