Explain SLIs, SLOs and error budgets, and how you would use them day to day.

An SLI is a measurement of user experience, such as the proportion of requests served successfully under three hundred milliseconds. An SLO is the target for that measurement over a window, say 99.9 percent over twenty eight days. The error budget is the remainder, so 99.9 percent leaves about forty three minutes of failure a month. When the budget is healthy you ship faster; when it is spent, reliability work takes priority.

The indicator is the thing you measure from the user's side, so for an API it is usually the share of requests that are both successful and fast enough. The objective is the target over a rolling window, and the budget is what is left over. Three nines over twenty eight days is roughly forty three minutes of badness you are allowed to spend, and framing it as a budget is the whole point, because it turns reliability from an argument into arithmetic. If we have spent five percent of the budget this month, we ship features and take sensible risks. If we have burned eighty percent by the tenth, the next sprint is reliability work and risky launches wait. For alerting I use multi window burn rates rather than paging the moment the ratio dips, so a fast burn that would exhaust the budget in an hour pages immediately, while a slow burn creates a ticket. That killed most of our night time noise. The hardest part is honestly picking the indicator, because a target on something users cannot feel is theater.