Site Reliability

Purpose

Make reliability a measured, budgeted property rather than an aspiration. Perfect availability is neither achievable nor desirable — the goal is to be exactly as reliable as the product requires, and to spend the remaining budget on shipping.

When to Use

• Defining SLOs and an error-budget policy.
• Designing a system's behavior under overload or dependency failure.
• Capacity planning ahead of a known load event.
• Setting retry and timeout policy across services.

Capabilities