DOF case studies
DOF · Put your knowledge to work in context.
This independent study page provides practice for DOF. Its editorial learning areas include DevOps adoption, culture and delivery flow; Automation and safe delivery technology; Measurement, reliability and learning signals; Feedback and continuous improvement. The questions develop subject knowledge and application through worked reasoning, distinctions between competing answers, and practical scenarios. Study the relevant guide before attempting a mixed set, then explain the evidence or calculation that supports each answer. Topics on this exam: DevOps adoption, culture and delivery flow, Automation and safe delivery technology, Measurement, reliability and learning signals, and Feedback and continuous improvement. The DOF case studies are 4 cases and about 876 words.
1A stable queue and a WIP experiment
A fictional delivery service has a stable long-run average of twelve items within its defined work boundary and completes four items per week. That boundary begins when the team accepts an item and ends when it is delivered to the customer. The team suspects that excessive simultaneous work contributes to waiting, but it has not yet tested the effect of a new work-in-progress limit. A manager proposes announcing that halving the posted limit will certainly halve delivery time even if staffing, item size, dependencies and throughput all change.
225 words, about 1 min2A pipeline that protects the artifact and the secret
A fictional team builds a release candidate, tests it and then rebuilds from the same source branch during production deployment. Its dependency ranges are not locked, so the second build can obtain different packages. The deployment script also contains a shared API secret in plain text inside the repository. A test has recently failed only in production, and the team wants to fix the uncertainty without bypassing security review.
223 words, about 1 min3A deployment scorecard without an individual league table
A fictional service records two hundred production deployments during a quarter. Twelve require immediate remediation after causing a production failure. The team also records eight unplanned deployments that were made because of incidents; the supplied definitions and denominators are fixed for this exercise, and these two counts need not describe the same events. A sponsor wants to compare the service directly with an unrelated safety-critical system and publish an individual engineer ranking based on deployment volume.
209 words, about 1 min4An incident review with an action that can be checked
A fictional delivery team restores service after a release encounters an undocumented database dependency. The initial meeting ends with the statement that the engineer should be more careful, but no owner, test or review date is recorded. The same engineer followed the current runbook, and several team members knew parts of the dependency but had no shared place to record them. A manager now wants a review that preserves a factual timeline and produces a defensible improvement.
219 words, about 1 min