5 exercises — practise answering Zero-Downtime Migration Engineer interview questions in professional technical English.
0 / 5 completed
1 / 5
The interviewer asks: "How would you migrate a production database from one engine to another — for example MySQL to PostgreSQL — with zero downtime for a high-traffic application?" Which answer best demonstrates Zero-Downtime Migration Engineer expertise?
Option B is strongest because it uses CDC-based continuous replication, ongoing reconciliation, and a near-instant flag-based cutover with rollback, avoiding downtime entirely. Option A explicitly requires a downtime window, which contradicts the zero-downtime goal. Option C's dump-and-restore approach requires the source to be frozen during export, causing an outage. Option D causes repeated downtime across multiple migration windows rather than eliminating it.
2 / 5
The interviewer asks: "Midway through a live traffic cutover to a new service, you notice error rates climbing on the new path. What do you do?" Which answer best demonstrates Zero-Downtime Migration Engineer expertise?
Option B is strongest because it treats rollback as a pre-rehearsed, instant action, preserves diagnostic data, and re-attempts via canary rather than a full cutover. Option A risks prolonged user impact while waiting passively. Option C attempts to fix a live incident in production without reverting to a known-good state first, extending risk exposure. Option D creates an inconsistent split-brain state that is harder to reason about and diagnose than a clean rollback.
3 / 5
The interviewer asks: "How would you validate that a migrated system behaves identically to the legacy system before fully decommissioning the old one?" Which answer best demonstrates Zero-Downtime Migration Engineer expertise?
Option B is strongest because it uses shadow traffic comparison over a representative time window including edge-case cycles, with explicit divergence review before cutover. Option A relies solely on pre-deployment tests, which cannot capture real production edge cases and data. Option C is an informal, unscaled manual check prone to missing subtle regressions. Option D removes the safety net immediately, eliminating any ability to compare or roll back.
4 / 5
The interviewer asks: "How do you handle schema changes that are not backward compatible during a zero-downtime migration, such as renaming or removing a column that the old application code still depends on?" Which answer best demonstrates Zero-Downtime Migration Engineer expertise?
Option B is strongest because it applies the expand-contract pattern explicitly accounting for rolling deployments where old and new code coexist, with a verified backfill before the contract phase. Option A ignores that a schema change and code deploy are never perfectly atomic across a fleet of instances. Option C reintroduces the downtime window this approach exists to avoid. Option D assumes ungraceful failure will not occur, which is a risky and untested assumption.
5 / 5
The interviewer asks: "How would you plan and communicate a large, multi-week zero-downtime migration across several dependent teams to minimise organisational risk?" Which answer best demonstrates Zero-Downtime Migration Engineer expertise?
Option B is strongest because it phases the migration with reversible checkpoints, a thorough upfront dependency discovery, and ongoing communication coordinated with dependent teams' schedules. Option A creates organisational risk by hiding a multi-team-impacting change until it is done. Option C compresses risk into a single high-stakes window, working against the goal of minimising risk. Option D assumes a one-time announcement remains sufficient across weeks of changing status, which typically leads to surprises.