If you work in an IT organization, you've likely heard the term "observability" lately. If you're a DevOps pro, you probably know exactly what vendors are talking about when they use the term. If you're a NetOps pro, you might be scratching your head.
DevOps Knows Observability
The DevOps community is very familiar with the observability concept. It refers to the ability to understand the internal state of a system by measuring its external outputs. In DevOps, the system is the application, and the outputs are metrics, logs, and traces. DevOps pros know how to navigate messaging from application performance management and cloud monitoring vendors to find solutions that can deliver the observability they need.
More recently, network monitoring vendors have started talking about network observability. Here is where things get fuzzy. In my opinion, DevOps observability and network observability are not interchangeable. Why would they be?
DevOps teams want to understand the state of applications and the infrastructure on which they reside. NetOps teams need to understand a much larger universe of networks, from the cloud to the user edge.
Both DevOps observability and network observability refer to the need to understand the internal state of a system, but that need to understand is only a problem statement. The solution to that problem is where the differences occur.
Does Anyone Have Network Observability?
First, most NetOps teams care about application performance. They want to collect data from the application environment if they can, such as hypervisors and containers. But they don't stop there. They need to monitor data center networks, wide-area networks (WANs), and campus and branch networks. More recently, they've had to worry about home office networks.
Each network they monitor has become more complex. The data center network has been virtualized and partially extended into the public cloud. The WAN has hybridized, with a mix of managed WAN connectivity, public internet, and 4G/5G. Office networks are a mix of ethernet and Wi-Fi, connected via home internet.
A network observability system must monitor and analyze an extremely diverse and ever-growing data set to understand end-to-end network state. A NetOps team might use five, ten, or even fifty tools to monitor a network by collecting packets, flows, device logs, device metrics, test data, DNS logs, routing table changes, configuration changes, synthetic traffic, and more.
It's a lot to keep track of, and it's hard to find a single tool that can handle it all. In fact, my new research on the concept of network observability found that 83% of IT organizations are interested in streaming data from their network observability tool(s) to a central data lake. Why? Nearly half of them believe a data lake will help them correlate network data across their tools.
Earlier, I wrote that network observability is a bit "fuzzy." I'd argue that it's fuzzy because the problem of network observability is much bigger and more complex than DevOps observability. It may prove impossible for any single tool to solve this problem. That's perfectly okay. But IT organizations must keep this in mind and take a comprehensive approach to network operations tools as they steer toward the promise of network observability.
To learn more about network observability, check out EMA's November 9 webinar, which will highlight market research findings on the topic.
