Site Reliability Engineering Flashcards

What does 'toil budget' mean in SRE?

Answer: A maximum percentage of engineering time allowed for repetitive manual work, beyond which automation becomes mandatory

SRE teams typically cap toil at 50% of time; if toil exceeds the budget, it signals the need to invest in automation and pay down operational debt.

What is request retrying with exponential backoff?

Answer: Retrying failed requests with increasing delays (1s, 2s, 4s) plus jitter to avoid thundering herd

Exponential backoff with jitter spaces out retries to reduce load on a recovering service, preventing all clients from retrying simultaneously and amplifying the problem.

What is a 'thundering herd' problem?

Answer: When many clients simultaneously retry or connect after a service recovers, overwhelming it again

A thundering herd occurs when a large number of clients all attempt to reconnect or retry simultaneously, often triggered by a service restart, overwhelming the recovering system.

What is a feature flag?

Answer: A configuration switch that enables or disables a feature at runtime without deploying new code

Feature flags (or feature toggles) decouple deployment from release, allowing teams to deploy code to production while keeping it dormant until enabled for specific users or segments.

What is 'on-call rotation' in SRE?

Answer: A schedule where team members take turns being the primary responder for production incidents

On-call rotation distributes the burden of production incident response across the team, with clearly defined primary and secondary responders and escalation paths.

What does 'mean time between failures' (MTBF) measure?

Answer: The average time a system operates without a failure — a measure of reliability

MTBF measures the average duration between consecutive failures of a system; a higher MTBF indicates a more reliable system.