Automation has captured the human imagination since the earliest days of recorded history. A fanciful example of this comes from Homer’s classic “The Iliad” in the form of a bronze robot called Talos, which was designed to automatically monitor and defend the island of Crete from marauding intruders. The appeal of automation for IT organizations, including network operations (NetOps) teams, may not be so life-and-death but it’s undeniable, nonetheless. More and more NetOps teams are turning to automation to achieve higher network uptime (by reducing manual errors), increase efficiency and reduce costs and help deliver a better and more consistent operational experience for all users.
Due to the allure of these promised benefits, the temptation for NetOps teams to integrate automation as soon as possible may be strong. However, it is important for these teams to remind themselves that automation is a tool, not the destination itself. Indeed, simply overcoming the technical challenges of deploying automation may not result in the successful outcome teams want, without identifying and defining the intended outcome of what their automation journeys should be.
This article will explore some practical steps and best practices that NetOps teams should use throughout their own automation journeys.
Getting the Journey Started on the Right Step
A well-known maxim in business is that you should visualize success and start with the end in mind when embarking on a major project or initiative. This is a good reminder for NetOps teams as they begin their automation journey. That is, these teams should be able to answer the question, “What is our desired end state when it comes to automation?” before they even evaluate what tools or technologies they may want to consider.
Teams who have this mindset will have a clearer understanding of their overall mission as well as an enhanced view of the various projects, processes and products they can improve through automation. It is critical for NetOps teams to take the time to put in the work of defining the desired IT experience at the end of the automation journey, even if the task itself may be difficult. For example, this experience may help define how NetOps engineers will interact with an automation tool through its code, its dedicated user interface, or even its generic service portal. As the saying goes, “no pain, no gain.” The teams who invest in the time and effort to plan ahead will be far more likely to maximize the benefits of automation.
What to Automate?
In addition to visualizing what success looks like, it is important for NetOps teams to identify what tasks or processes they should be automating. This is important because there is a vast universe of NetOps tasks that can be automated. Taking the time to identify tasks to be automated will generate huge benefits for the NetOps team as it will help them select the correct automation strategy and toolset that will deliver them the results in an efficient way. And while specific projects may be unique to each NetOps team, in general, there are two broad categories of tasks that are ideal for automation:
1. Highly repetitive tasks. These are manual or semi-automated tasks that are not necessarily hard but can be cumbersome for NetOps teams. Examples of repetitive tasks that can be automated include assigning an interface into a specific VLAN or perhaps updating network security policies or rules.
2. Highly complex tasks. These can be more strategic and difficult tasks that can be prone to human error due to their scope, scale, and complexity. Automating these tasks may require additional planning and steps to ensure success. Examples of complex tasks that are ideal for automation include service redirection from one device to another or BGP peering with third parties.
Think Big, Start Small With NetOps
Like all people, NetOps professionals enjoy the results of a job well done. So, while the vision of their automation journey may be big, it’s important to start with a small, short-term project that can be completed quickly. There are a couple of benefits to this approach:
- Quick automation wins will give NetOps teams confidence for future projects.
- Projects like this can generate data and feedback that NetOps teams can convert into learnings and insights for the next project.
This approach can also be applied to bigger, more complex automation projects. Instead of taking on the entire scale of the project at once, NetOps teams can break it down into smaller components. For example, infrastructure-as-code (IaC) is a broad category of automation entailing the managing and provisioning of systems through machine-readable definition files. There are many considerations to large-scale IaC projects such as identifying the use case, e.g., Day 0 or Day 2 operations. The advantages of this approach are the same as with the quick-win scenario: There is a better likeliness of success and more immediate feedback and data to guide the NetOps teams through this entire process.
Finally, as talented as most NetOps teams are, they are not likely to have all of the automation expertise in-house at any given time. So, it is important that they address skills gaps (automation, programming) through strategic vendor partnerships, internal/external recruitment and additional training for their teams. Again, instead of recruiting or investing en masse, NetOps teams are well-advised to take the time to plan, execute, learn and modify as they go along.
The human fascination with automation is stronger than ever. The industry buzz around one particular form of automation—especially artificial intelligence and machine learning—is indicative of this. NetOps teams have much to gain from investing in the tools and skills required to do automation right.
Automation of data center network pods, self-contained zones that represent a section of the overall data center, is a good example of a strategic project requiring forethought and planning to execute well. A logical first step in this process might be to anticipate how to manage future changes when automation is in place. Options for this may include code manipulation, Git (open source code management) or CI/CD pipelines, all elements of infrastructure-as-code automation. Or it may involve improving user experience through a specific UI interface or through a self-service portal. The next step then might be to decompose this large configuration into smaller bits starting with the registration of the new switches into the controller. Once this is done, we can now go into port configurations and then eventually move toward larger and more sophisticated elements such as the automation of VLAN assignments and L3 peering.
For long-term and large-scale automation projects like this, it is especially important to follow the recommended key steps outlined in this article, which are:
1. Start with the end in mind. It’s critical that the teams take the time and effort to define what success looks like, especially in the form of the IT experience they want to achieve.
2. Know what you want to automate. There is an endless supply of tasks and processes that can or should be automated. Prioritizing automation projects that NetOps wants to take on is a major milestone in any automation journey.
3. Think big, start small, learn as you go: Quick automation project wins are key to building confidence and “muscle memory” for future projects. For bigger projects, the best approach is to break them down into more manageable projects that can provide vital data and insights during the journey.