Enterprise networks are complex ecosystems that encompass multi-cloud, multi-domain computing. While networks become more complex, the expectations remain the same:
How are IT personnel to maintain networks without disruptions in service, especially when there is no such thing as office hours? Does DevOps have a place in modernizing network management? Will automation play a part in upgrading the network?
Possible answers to these questions were presented at ONUG Fall 2020 in a panel discussion moderated by Michael Clark, Chair of the Cloud Native DevOps Steering Committee. The panel was made up of:
The panel was asked if the global pandemic had changed the trajectory of network digital transformation. Everyone agreed that the direction did not change, but it was accelerated.
Network management is knowing what needs to be done and how to do it. Modernizing means separating the process. According to Chris Wade, the process should be divided into what needs to be done and how to do it. DevOps should play a role in establishing a pipeline so rules can be developed for testing and implementation.
To Brent Holden modernizing means looking at the time to value. Should an organization self-code solutions or find solutions that can be customized and supported? More and more companies are looking at highly customizable packaged solutions to shorten the time to delivery. Jeff Gray felt modernizing was finding ways to streamline the process and to reduce the blast radius — minimize the scope of errors.
All panelists agreed that legacy systems and legacy mindsets were the primary obstacles to modernization.
Legacy systems represent a substantial investment in hardware and expertise. While maintaining these systems may take more time and resources, IT is comfortable with how they operate and understands their behavior. What legacy systems cannot do is scale quickly. With the lack of scaling, IT staff are faced with delivering performance and availability without compromising security.
Network management as a group is risk-averse. That’s not surprising given past experiences. No one wants to repeat the days when a change brought down the network, or an update had to be backed to restore availability. Simultaneously, network managers want to reduce the number of errors and improve their time to market.
The panelist felt that DevOps could help networks, but they disagreed on how. They agreed that breaking down silos improves collaboration. By working together, network, development, and security teams develop an understanding of what each group does. The collaboration makes for a stronger technical team.
Whether a company developed a DevOps culture was not as important as removing silos. In fact, Michael suggested that DevOps might not apply to network modernization since it was designed for application development. He wondered if there were any use cases that confirmed that DevOps worked? Since few organizations actually measure key performance indicators such as mean time to restore, there is little data to confirm improvement.
It’s essential for network personnel to understand that the DevOps concept is designed to facilitate operations and not stand as gatekeepers. The intent is to enable networks to improve testing and speed up time to market. What organizations call the concept is not as important as implementing it.
Different technologies are available for modernizing a network. Some enterprises may look at automation or an API programming approach. Others may look at prepackaged solutions that can be deployed through commands.
Automation may be as simple as stringing a series of commands together to replace a manual process. More sophisticated automation uses machine learning and big data to determine how to configure and manage a network. Intent-based automation can help engineers scale a network based on desired outcomes, reducing the chance of a misconfiguration error. Chris suggested that Software-Defined Networking (SDN) combined with virtualization, can help balance network load by automatically redirecting traffic to a less used resource.
An API-first approach allows for more in-depth access to information. Device-level data can be collected and shared, making it easier to manage the overall process, including automation. If data is tied to network management, critical areas can be identified and the processes automated.
APIs can help organizations increase their network management tool kit. Understanding how to use tools to create a more in-depth view of the network means a faster path to identifying and correcting errors.
IT personnel can use prepackaged programs to automate processes. These programs can be scheduled and executed from an application’s front end. Once deployed, these solutions can automate network configuration, security, and provisioning. Automation could be as simple as running scripts that automate manual processes.
If networks are going to meet the enterprise demands, they must find a way to scale quickly and reliably. They must develop the agility to pivot to meet ever-changing expectations. It is imperative that organizations do something. The panel suggested that enterprises consider the following:
Don’t try to take on digital transformation all at once. Make changes a step at a time. As more projects become successful, people will become more comfortable with the solutions and processes, making it easier to change not only legacy systems but also legacy systems.