Practical Guide to Modern Networking Telemetry

Chapter 5. Using Multiple Types of Network Telemetry Together

Building on Chapter 4, we’ll now explore the ways that multiple, disparate data points combine to provide a better picture of not only your network, but your entire environment. In this chapter, we are still narrowly focused on network data. After making such a compelling case in Chapter 4 for the power and value a single data element can have, you might wonder whether multielement issues are significantly better. To that point, we’ll quote one of our favorite “mind palace” moments from the BBC series, “Sherlock”:

Mycroft: “Oh, Sherlock. What do we say about coincidence?”

Sherlock: “The universe is rarely so lazy.”

“The Sign of Three,” Sherlock 2014

As with coincidences, the universe is rarely so lazy that two (or more) metrics, happening within a similar time period, are mere coincidences. Metrics from a single domain don’t always tell us everything we need to know about an outage—how it happened, where to look, or how to go about fixing it. In this chapter, we will share some examples of (realistic, if not common) issues where two or more network telemetry elements provide the answers and insights you need to return your infrastructure to peak performance.

Use Case 1: When Synthetic Is Better than the Real Thing

There are times you’re pretty sure there’s a problem, but you lack the data to prove it. You know the data you could use to prove it, but it relies on specific data generated by specific actions that users ...