Site reliability engineer algorithms interview questions

As asked

Given a 99.9 percent SLO over 28 days, write a function that takes the last hour's success and total request counts and returns the burn rate as a multiple of the budget.

Sample answer outline

Error budget per hour at 99.9 percent over 28 days is 0.1 percent of an hour's request count. Burn rate = (observed_error_rate / 0.001). A burn rate of 1 means you are burning exactly at budget. A burn rate of 14.4 means you will exhaust the budget for the month in the next hour. Common alert thresholds: page on 14.4x over 1 hour (fast burn), 6x over 6 hours (slow burn). The function is small but the framing is what the interviewer is testing.

Reference implementation (typescript)

export function burnRate(
  successCount: number,
  totalCount: number,
  sloTarget = 0.999,
): number {
  if (totalCount === 0) return 0;
  const errorBudget = 1 - sloTarget;
  const observedErrorRate = 1 - successCount / totalCount;
  return observedErrorRate / errorBudget;
}

Expect these follow-ups

• How would you alert on both fast and slow burn at the same time?
• What is the failure mode if traffic volume drops by 90 percent?
• Why is a single threshold alert worse than a multi-window one?