Stage 2 · Lesson 6 of 7
Sampling and correlating signals
At scale you cannot keep every , so you sample. Head decides when a trace starts, such as keep 10 per cent; it is cheap but may throw away the rare failure. Tail sampling decides after the trace finishes, so it can keep every error and slow request and sample the boring ones; it needs a that sees whole traces.
Then connect the signals: put trace and span IDs in every log line, and attach (example trace IDs) to latency metrics. Now a spike on a graph links to a slow trace, and the trace links to its logs.
Common trapsampling 1 per cent at the head and discovering during an that the failing request you need was never recorded.
A question to check yourself, an AI practice session and 2 cheat sheet entries come with the full roadmap. Unlock
Resources for this stage
Start with· follow one of these
- GuideFreePrometheus Documentation
Prometheus · Intermediate
Free: the first steps guide, then the querying basics, functions and histograms pages
- CoursePaidPrometheus Certified Associate (PCA) Prep Course
KodeKloud · Intermediate · Subscription
Paid, hands-on labs for PromQL, recording rules, Alertmanager and Kubernetes monitoring
Practise
Keep handy
Also worth your time
The chapter that defines the four golden signals and what makes monitoring worth paging on