Network observability: What it means to vendors and to you

Network observability can be dismissed as a buzzword, but to be useful it should include monitoring, security, data, and actionable insights.

man with binoculars observability research spy
Getty Images

As an industry analyst who focuses on network management, I make it my business to cut through hype and buzzwords. When vendors started talking about “network observability” last year, I found myself at an impasse. It seemed everyone had their own take on what this term means.

Google network observability, and you will find an endless list of varying definitions. I recently read a blog that boiled it down to this: “Network monitoring is fault management. Network observability is performance management.”

No! No, it is not!

I decided to do something about this chaos. I talked to multiple network tool buyers and users, and I surveyed 402 IT stakeholders. I asked them to define network observability and explore what it means in the context of their network management tools and processes.

First, 90% of respondents believe it’s a useful term to describe the tools they use to monitor and troubleshoot their networks, but only 20% prefer network observability to more established phrases, specifically network monitoring and network performance management.

How IT stakeholders think of network observability

Here is what I heard from people I talked to one on one. First, the cynics:

“I think it’s just another catchphrase for network monitoring.” – Monitoring engineer at a $15 billion financial services company.

“For me, it’s just another buzzword. I’m still for the terms enterprise management or network management.” – Network engineer at a $14 billion aerospace and defense company.

Next, the people who think vendors might be on to something:

“I think of monitoring as something that is not ‘actioning.’ Network observability is about integration into event management, turning your observable metrics into something that alerts you or drives an action.” – Network operations manager at a $500 million SaaS provider

“It’s about getting insights from the network. It could mean more of a focus on the business impacts of the network.” – Network tools engineer at an $8 billion technology company

Action and insights came up a lot. In fact, I asked survey respondents to select words or phrases they most associate with network observability. The top four responses were: monitoring, security, data, and actionable insights. I think these four are a good foundation.

Network observability: Sifting meaning from the marketing

Network observability represents an evolution of network monitoring. Network observability solutions should dive deeper into networks, collecting a more diverse and voluminous set of data to give network teams total end-to-end visibility into operations. Those solutions should broaden their scope, looking not just at network performance, but end-user experience, business impacts, and security.

Finally, network observability should focus less on tinkering with how it presents data, which ultimately forces network engineers to glean insights themselves and doing too much of the heavy lifting in their heads. Instead, network observability should emphasize actionable insights derived in a variety of ways, including AI and machine learning and low-code scripted automation.

The former relies on algorithms to make tools more intelligent. Many vendors are driving toward actionable insights with AIOps, and our research shows that NetOps pros see tremendous potential with these algorithms. The latter approach involves scripted workflows, reports, and automations that can capture and execute on network engineering knowledge. This former approach might be available out of the box or it might be built in-house using APIs and other building blocks that vendors offer.

The network tools engineer mentioned above sees the potential of both approaches. “When something happens, we want a seamless fix with automation. It should trigger an alert that triggers an automation script that checks issues, fixes the problem, and closes the ticket, all without paging a person. I don’t see many tools that can do it, but you can build it in-house if your vendor has web hooks or APIs. Productized automated troubleshooting with AI and machine learning would be great.”

This is what you should look for from your network monitoring vendors today.

(Shamus McGillicuddy is Vice President of Research, Network Management, at Enterprise Management Associates, and is hosting EMA’s November 9 webinar about the firm’s network-observability research.)

Copyright © 2022 IDG Communications, Inc.

The 10 most powerful companies in enterprise networking 2022