Telemetry is the data a system emits about its own operation, such as logs, metrics and traces, so that it can be monitored and analyzed. Logs record individual events, metrics measure values over time, such as request rates or error counts, and traces follow a request through multiple services. Some teams also collect profiles that show where code spends its time.
Telemetry is the raw material of observability. Without it, engineers have to guess why a system is slow or failing, and with it they can detect problems, investigate causes and track trends.
It brings costs and risks. Collecting, transporting, storing and querying telemetry takes money, and volume can grow faster than expected. Telemetry can also contain sensitive information, such as personal data, tokens or internal details that were written into logs or attached to traces, so it needs the same care as other data.
To manage it, decide what each signal is for, set retention periods, keep sensitive fields out through filtering or redaction, restrict who can read it and review costs by signal type. Remove data that nobody uses.