Business needs for communication and commerce rely more heavily on the network than ever before, and this drives an increased rate of changes with minimal downtime. To keep up with the pace of the business, IT teams must begin to leverage automation and orchestration as much as possible. In today’s typical network, automation is underused due to the initial cost, complexity, and even cultural history that seeks to limit changes to the network. Increasing the use of automating network services usually requires transitioning skill sets from network protocols to programming languages.

Many organizations do not have the headcount or time to take on the required training to build and maintain a “home-grown” automation infrastructure. Lean IT teams without dedicated development require a platform for automation with turn-key features built in to get started automating time-consuming tasks that are done manually today. Organizations with development resources need an open platform allowing them to customize and build the required automation and migrate from limited home-grown solutions.

Most of the existing, installed base of networking equipment is configured via command line through the command line interface (CLI). This type of interface may be acceptable in a lab or small network, but when you need to configure dozens to hundreds to thousands, manual CLI does not scale. In addition to being time-consuming to configure, using CLI requires deep knowledge of that particular vendor’s command structure and semantics, which must be configured in a very specific order.

The first and most important benefit of adopting network automation is the reduction of time it takes network engineers or IT operations to implement changes in the network. This affects everything from new site installations to upgrades and ongoing changes. A recent Forrester research report shows that the majority of IT operations' time is spent performing maintenance on the existing network. This takes away from time to address incoming change requests to keep up with the business needs.

A second important benefit is the reduction of network downtime. Automated tasks are executed much faster and with a higher degree of accuracy over manual changes. One Gartner study shows that close to half of all network outages are due to manual misconfiguration. Leveraging network automation enables an operations team to accomplish more tasks with the existing team without needing to hire expensive contractors or reject change request, which slows the flow of business.

On average, enterprises have 13 to 20 different network device platforms deployed in their network. Some organizations selected “best of breed” devices, such as firewalls, load balancers, switches, routers, and wide area network (WAN) acceleration. Other organizations have multiple platforms from a single vendor, yet each has a different operating system and management system.

Although each device has its own merit, this creates challenges and additional cost for network operations. IT teams must hire or train team members on each vendor platform, which results in siloed skill sets. Each vendor platform will also have its own specific management requirements, which could include manual CLI or a web-based interface. Some of the newer solutions provide a software-defined networking (SDN) controller that enables management of many devices, but this also provides an application or an application program interface (API) that must be developed on. Each of these vendor management platforms usually comes with annual support costs as well.

When applying network automation to a multivendor, multiplatform network, it is important that it supports those vendors and is extensible to add features and additional vendor platforms. Ideally, the automation platform is able to abstract feature configuration away from the vendor-specific method so that when configuring the same feature, like a virtual local area network (VLAN) ID on a switch, the workflow is the same even if the underlying network is using multiple switch vendors.

Once an organization has made the decision to automate, several steps need to be taken:

• Identify what tasks are the most time-consuming for IT operations and use them as a starting point to implement network automation. It may be spin-up of new sites, ongoing change management like quality of service policies, access control lists, or upgrading devices.
• Decide on the network automation platform to determine if it has the feature set and vendor support required. This often means moving away from a home-grown platform and determining if the right fit is a platform for developers or one that has features ready for IT operations – ideally, one that is open and extensible as well.
• Before any changes, it is critical to get to a known state on the network and normalize all the configurations that will involve inventory, discovery, and remediation to get control of the existing network, since it has likely undergone manual changes for many years.
• Implement audit capabilities and regularly check that the network “policy” is properly configured for each feature and is enforced on the network.
• Start the definition of the automation process early, ideally in the test lab when the network engineer is providing the “golden config” as to how a network feature works. This is the time to define how it can be automated for implementation at scale. Instead of a single working configuration, develop a feature template that can be automated across all network devices.
• Update the change management control process to drive all changes through the automation platform and begin to minimize manual changes.
• Practice continuous integration. When implementing network automation, there is no stopping point; there is always more to do. Start small, get some of the mundane tasks automated and then move on to more and more tasks over time.

Networks today are more complex and demanding than ever. There is far too much going on for humans to handle on their own. Automation and orchestration have become essential to the proper functioning of the network. As organizations make the shift from manual to automatic, IT teams may feel the awkward jolts and frustrations of the learning curve. However, by putting the steps above into practice, the network can run like a well-oiled machine, freeing the team to pursue more critical and time-sensitive task while automating the mundane ones.