Flow is a key property of event driven systems and yet understanding flow through a distributed system is notoriously hard. This manifests in how we think about flow: “what path does this operation take through the system” or “what message entering this component led to another message leaving this component” or even “why did this flow take the fault path”. This goes further into how we think about resolving errors like “in which component did this flow error and why”

Traditional approaches like logging inevitably create more noise than signal when we try to observe flow across components, and other methods like intercepting messages using a wiretap have limited effectiveness in telling us “why” something happened. Both of these techniques often create as many questions as they answer.

In this talk we look at how utilising modern observability techniques can help you to understand the flow of your event driven system.

We’ll cover different techniques for modelling your telemetry and look at the Semantic Conventions for Messaging. We’ll discuss how using standardised naming helps us provide trusted signal data, as opposed to the noise. We will look at the W3C Trace Context standards, and understand how this gives us a view of flow within our system along with live demos on how to make it all work together

With examples in C# and JavaScript throughout we will show you different approaches to observing the flow of your system, and their trade-offs. We will show a range of middleware transports including Azure Service Bus and Kafka