Propagating context

Getting the information from one service to another across the network boundary requires some additional work, namely, propagating the context. Without this context propagation, each service will generate a new trace independently, which means that the backend will not be able to tie the services together at analysis time. As shown in Figure 4.5, a trace without propagation between services is missing the link between services, which means the traces will be more difficult to correlate:

Figure 4.5 – Traces with and without propagation

Specifically, the data needed to propagate the context between services is span_context. This includes four key pieces of information:

• span_id: The identifier of the current span
• trace_id: The identifier of the current trace
• trace_flags: The additional configuration flags available to control tracing levels and sampling, as per the W3C Trace Context specification (https://www.w3.org/TR...