Exercise 1: Metrics vs. Logs vs. Traces, With Examples — Possible Solution
====================================================================

METRICS: numeric, time-series data, like CPU percentage, request
count, or latency over time.
Best-suited question: "Is something wrong right now?" -- for example,
"has our error rate spiked in the last 10 minutes?" is a metrics
question, because it's asking about a trend or a threshold being
crossed, not the detail of any single event.

LOGS: discrete, timestamped records of individual events, like a
specific error message or a single request line.
Best-suited question: "What exactly happened?" -- for example, "what
was the exact error message and stack trace when request ID 4471
failed at 14:32?" is a logs question, because it needs the specific
detail of one particular event, not an aggregate trend.

TRACES: the path a single request takes as it moves across multiple
services.
Best-suited question: "Where in a multi-service chain did it break?"
-- for example, "this request touched the API gateway, then the auth
service, then the database -- which one of those three actually added
the extra 2 seconds of latency?" is a tracing question, because it's
about attributing time/failure to a SPECIFIC HOP in a chain of
services, which neither a single metric nor a single log entry from
one service alone would show on its own.

WHY THIS WORKS AS AN ANSWER
------------------------------
Each example question is deliberately chosen to require the SPECIFIC
property that distinguishes its signal type from the other two --
an aggregate trend (metrics), a single event's full detail (logs), or
attribution across multiple services in one request's journey
(traces) -- directly matching the "best suited to answer" column of
the chapter's own comparison table.