By Song Pang, Chief Technology Officer at NetBrain Technologies

What if you could make your existing network engineers 10x more efficient? Just imagine what your team could do with all that extra time and work.

This may not just be a daydream. Technologies like AI and automation can make NetOps more efficient and proactive without adding more people. More specifically, automation can capture expert knowledge and apply it at scale, and AI can make automation accessible to everyone on the NetOps team. The best gains come from using both of them together. 

The core of many NetOps problems today is scale. Networks have gotten much more complex (hybrid environments, growth of SaaS and IoT devices, remote work, and more) but network operations processes haven't scaled up to match. Automation helps processes scale up but requires engineers with a specific mix of networking knowledge and coding skills. AI can allow engineers without that rare skillset to access automation. This means junior engineers can resolve more tickets themselves without escalating them, senior engineers can work more efficiently, and all of NetOps can become more proactive. 

How does this work in practice? Glad you asked. Here are five scenarios where combining AI and network automation makes NetOps more efficient and proactive.

Automated Diagnosis

Current AI models are good at interpreting natural language and analyzing data. Both of these qualities can be used along with automation to speed up diagnosis and troubleshooting. AI chatbots can be used as a translator or interface between engineers and the network. Engineers can ask questions in normal language, the chatbot interprets the questions and then uses automations that query the network (or a digital twin of the network) to answer them. This grants junior engineers access to network data like IP addresses, DNS, neighbors, and device logs without having to use CLI or a complex network management tool. For example, engineers could ask "Check the uptime of all devices" or "Check all devices of logs including the word "error" and summarize the results."

AI can also look at network traffic patterns to find performance bottlenecks, or analyze historical data and suggest potential causes (congestion, misconfigurations, or hardware failures) for a current issue. When a ticket comes in, AI can execute automated workflows to gather relevant data, such as latency measurements, packet captures, and device configurations. If an engineer doesn't have to do this manually, it saves time on every ticket. At enterprises with hundreds of tickets per day, these savings become significant very quickly.

AI-Assisted Observability

AI and automation together can run network assessments continuously to find issues before they affect customers. Assessments were historically done manually (if they were done at all). AI and automation allow them to be run on a regular schedule and check every rule network-wide for security, critical apps, config, connectivity, and performance. This allows NetOps to work more proactively, rather than waiting for issues to affect users and be reported to them.

AI-Assisted Change Automation

Uptime Intelligence found that configuration and change management is the root cause of 45% of all network outages. Automated checks can make sure that network changes don't accidentally break things or cause configuration drift. In advance, NetOps should define the "golden configurations" or ideal state for the network. Then automations can validate the actual state of the network against this desired state.

Before a specific network change, AI can define which rules are relevant for it. During the change, automated checks can verify that the actual configurations still match the golden configs. Afterward, AI can help integrate the new design into the golden config rules to prevent future configuration drift. In the event of a major network outage, AI could quickly analyze configuration changes made within the last 24 hours to identify potential causes of the outage, such as misconfigurations, accidental changes, or unauthorized modifications.

Auto-Remediation

When automated assessments find misconfigurations or violations within  network or security rules, AI can suggest actions to remediate it. Remediations should be verified by human engineers, but the AI suggestions speed up the troubleshooting process. Then, depending on the actions needed, AI can orchestrate automations to implement it. For example, if the configurations on a backup firewall didn't match the primary one, AI could suggest changes to keep configurations in compliance.

Find Devices Vulnerable to CVEs

AI and automation allow for proactive security checks. For example, when a new CVE comes out, automation can scan the network for devices matching the criteria of the affected Cisco products. An AI chatbot can use the results to generate a report listing all vulnerable devices, including their location, model, and software version. Reports can be easily shared with security teams for immediate action. This speeds up the patching process, reduces the time the network is exposed to threats and improves overall compliance with industry standards and regulatory requirements.

We are still in the early days of AI for networking. Enterprise networks are extremely complex, no two are alike, and the commands used to interact with routers, firewalls, and other network devices aren't standardized across vendors. This limits how much AI models can act without human involvement. Tasks like designing enterprise networks completely, diagnosing issues independently, or making network management decisions are beyond their capabilities - for now. In the future, these may become possible as AI's reasoning abilities improve and the cost of AI training comes down. However, organizations can still get dramatic improvements in their NetOps efficiency and proactivity today by using AI and automation together.

##

ABOUT THE AUTHOR

Song Pang is the Chief Technology Officer at hybrid network automation and visibility company NetBrain, responsible for Pre-Sales, Professional Services, Technical Support and Customer Success. He has been at NetBrain for almost ten years in a variety of customer support and engineering roles and formerly was an analyst at Stroud International. Pang has a B.S. in Electrical and Computer Engineering from Cornell University. Founded in 2004, NetBrain is the market leader for NetOps automation, providing network engineers with dynamic visibility across their hybrid networks and low-code/no-code automation for key tasks across IT workflows.