Loading video player...
OpenTelemetry is empowering the cloud native industry to leave behind the old concept of isolated “observability pillars”, enabling different signals —traces, baggage, metrics, logs, and more recently profiles— to be produced as a set of correlated, contextualised streams of telemetry data. This allows us to understand complex distributed systems more effectively. However, do you always know when to use one signal over another when instrumenting workloads? In this talk, we'll explore how the very structure and aggregation of these signals inherently carry meaning. From the choice of metric instruments, to the relationships between spans in a trace. They all have a purpose. Their correct usage, paired with other OpenTelemetry concepts like context, resources, or semantic conventions, allows us to describe our systems more efficiently. We'll dedicate a special focus to logs, and the recent convergence with events as a type of structure log type. Learn how this shift is unlocking powerful new instrumentation possibilities thanks to complex attribute values, like nested objects and arrays, crucial for emerging use cases like Gen-AI observability and client-side instrumentation. Join us to understand the profound depth of information your OTel signals can unlock for your workloads, and understand why logs are not just a “legacy signal”.