Observability: From Logs to Traces

For years we debugged with logs and intuition. It worked until a request touched eight services and the slow one wasn't the obvious one.

Spans over lines

A trace ties one request's journey across services into a single timeline. Suddenly "the API is slow" becomes "the auth service adds 400ms on cache miss."

const span = tracer.startSpan("db.query");
try {
  return await pool.query(sql);
} finally {
  span.end();
}

Logs answer "what happened." Traces answer "where did the time go." You need both, but the second one ends incidents faster.