In the previous chapter, we reviewed several techniques that rely on distributed context propagation, which often comes built in with the tracing libraries. The notion of a request context that travels along the requests in a distributed transaction is becoming increasingly important for monitoring distributed systems, not just by means of distributed tracing, but even with the more traditional monitoring tools such as metrics (or stats) and logs. By enriching metrics and logs with the metadata from request context, we can observe behavioral patterns in the applications that otherwise would be hard to notice just by looking at the aggregates. This is quickly becoming a new norm in the observability industry and is one of the reasons that projects such as OpenCensus come out with combined APIs for metrics, logs, and tracing, all of which are tightly linked with distributed context propagation.
In this chapter, we will look at many integration points...