8.3 Availability, SLOs and error budgets
You can state a reliability target in numbers and design to it instead of to 'always up'.
Before:03. Storage & DatabasesUnlocks:10. Architecture & Migration11. Projects, Certification & Portfolio
Always up is not a target, it is a mood — this topic replaces it with numbers: SLIs measure, SLOs commit, SLAs promise outward, and each extra nine multiplies the cost. Error budgets turn the SLO into a release-speed decision: burn the budget and slow down, keep it and ship. It sits late in the observability module because targets need measurement to mean anything. The design counterpart is graceful degradation — a system that gets worse politely beats one that fails completely, and building that choice takes intent.
Work through these
SLI, SLO, SLA and who each is for
An indicator is what you measure, an objective is the target you set yourself, and an agreement is the contractual promise to a customer. They are frequently confused, and each has a different audience.
What each extra nine costs
Every additional nine of availability costs disproportionately more in engineering and architecture. Knowing the shape of that curve is what makes a reliability target a business decision rather than an aspiration.
Error budgets as a release-speed decision
The gap between your target and perfection is a budget you can spend on releasing changes. Framing it that way turns an argument about caution into an arithmetic question.
Graceful degradation over hard failure
A service that sheds features under stress serves more users than one that fails completely. Designing which parts may fall away first is a decision to make before the day you need it.
Sign in to keep your progress.
Free resources
We haven't checked most of these for screen reader use yet.
Links last checked 29 Aug 2026.
Stuck here?
Ask a mentor. A real person answers, and they can see exactly which topic you're on. Usually within a couple of working days.
Checking your session…
Topics shown in module order.