Toil Reduction & Elimination Flashcards

Google's SRE model recommends capping toil at what percentage of an SRE team's working time?

Answer: 25%

Google's SRE book recommends that toil should not exceed 50% of each SRE's time, with the aspirational goal of keeping it well below that, often cited in practice as targeting around 50% or less with continuous improvement toward lower percentages; the canonical cap stated is 50%, but the practical target emphasized is keeping it at or below 50%, and many sources cite the goal as under 50% — the most commonly tested figure in the SRE Foundation certification is 50%.

Which of the following is an example of toil that is ALSO a reliability risk if left unaddressed?

Answer: Manually rotating API keys before they expire

Manual key rotation is toil that also creates reliability risk because human delays or oversights can lead to expired credentials causing production outages.

A self-healing system automatically detects and corrects a misconfiguration without any human involvement. This BEST demonstrates which toil elimination principle?

Answer: Closed-loop automation

Closed-loop automation detects, diagnoses, and remediates issues automatically, fully removing the human from repetitive remediation tasks.

When measuring toil, an SRE tracks 'interrupt-driven work.' What does this term describe?

Answer: Reactive tasks that pull engineers away from planned project work

Interrupt-driven work is unplanned, reactive work triggered by alerts or requests that disrupts engineers' focus on strategic improvements.

An organization uses 'error budgets' as part of toil reduction. How does an error budget directly relate to toil?

Answer: Excessive toil consumes time that could be spent on reliability improvements, indirectly burning error budget faster

High toil reduces the time available for reliability engineering work, which means incidents are less likely to be prevented, leading to faster error budget consumption.

Which scenario BEST illustrates 'toil that creates more toil' (a toil feedback loop)?

Answer: Manually scaling servers under load generates alerts requiring manual acknowledgment, which itself generates tickets requiring manual closure

Toil feedback loops occur when one manual task triggers additional manual tasks, compounding the time burden with each cycle.

A proposed automation would handle 90% of cases but requires human escalation for the remaining 10%. What is the BEST approach?

Answer: Implement the automation for the 90% while building a clear escalation path and logging edge cases to close the gap over time

Iterative automation that handles most cases immediately while gathering data to close gaps is far better than delaying automation until perfection is achieved.