Using libraries to instrument efficiently
Instrumenting your application code to emit the telemetry of logs, metrics, and traces can be complex, time-consuming, and difficult to maintain. There are two main approaches to solving this problem – automatic instrumentation and manual instrumentation – with a wide selection of SDKs and libraries available to support them. Here is a brief overview of them:
- Automatic instrumentation: Automatic instrumentation is the simplest to implement but can lack the level of control that’s often required when building an observability platform. In a very short space of time, it will provide visibility into your application and help you start answering your observability questions. Without careful configuration and design, this will lead to other problems such as performance and cost issues, and, in the worst case, render the observability platform useless.
The approach varies depending on the programming language; for example...
