Observability Tools

Observability tooling answers three separate questions, and it helps to keep them apart: what is the system doing right now (metrics), what happened in this specific case (logs and traces), and what broke for this user (error tracking). Many tools cover more than one; few cover all three equally well.

The main decision is commercial rather than technical. Open-source stacks give you the same core capability with no per-host or per-GB bill, and cost you the time to run them. Commercial platforms are genuinely good and genuinely expensive, and the expense scales with exactly the thing you want more of — data. Teams frequently start open-source, hit an operational wall, and buy; others buy first and later move high-volume telemetry back in-house to control spend.

Whatever you pick, the instrumentation is the durable asset. Standardised, vendor-neutral instrumentation is what makes switching possible later; proprietary agents are what makes it painful.