An application has elevated request latency across several services. The team needs to locate the slow dependency and inspect the detailed events for the same requests. Which two telemetry practices are most useful? Select 2 answers.

1. Inspect only total project CPU usage and assume it identifies the slow downstream call.
2. Include correlation identifiers in searchable structured application logs.
3. Use the monthly billing invoice as the primary request-level latency record.
4. Increase log retention without adding request identifiers or traces.
5. Propagate trace context and inspect distributed request spans.