What technique enables zero-downtime database schema changes?

What technique enables zero-downtime database schema changes?

• A. Backward-compatible online migrations with feature-flagging
• B. Immediate full schema rewrite during deployment
• C. Relying on downtime windows scheduled at off-peak hours
• D. Removing backups prior to migration

Answer: (A)

Zero-downtime schema changes come from doing migrations that don’t disrupt ongoing operations and can be deployed while the app stays live. The technique here is backward-compatible online migrations paired with feature flags. By making non-breaking changes—such as adding a new column that's nullable, introducing a new field with a safe default, or adding an index in a way that doesn’t block reads/writes—you keep existing queries and code paths working. The feature flag then controls when the new behavior is used, allowing you to roll out the change gradually, monitor impact, and flip the flag to enable the new path only after it’s proven safe. If issues appear, you can revert by toggling the flag while the migration itself remains non-disruptive. This combination prevents downtime because both the schema change and the code deployment avoid long locks and hard outages, and the old path remains available until you’re ready to switch over. In contrast, an immediate full schema rewrite during deployment is likely to cause downtime due to locking; relying on downtime windows isn’t zero-downtime; and removing backups before migrating is unsafe and unrelated to keeping the system available.

Zero-downtime schema changes come from doing migrations that don’t disrupt ongoing operations and can be deployed while the app stays live. The technique here is backward-compatible online migrations paired with feature flags. By making non-breaking changes—such as adding a new column that's nullable, introducing a new field with a safe default, or adding an index in a way that doesn’t block reads/writes—you keep existing queries and code paths working. The feature flag then controls when the new behavior is used, allowing you to roll out the change gradually, monitor impact, and flip the flag to enable the new path only after it’s proven safe. If issues appear, you can revert by toggling the flag while the migration itself remains non-disruptive. This combination prevents downtime because both the schema change and the code deployment avoid long locks and hard outages, and the old path remains available until you’re ready to switch over. In contrast, an immediate full schema rewrite during deployment is likely to cause downtime due to locking; relying on downtime windows isn’t zero-downtime; and removing backups before migrating is unsafe and unrelated to keeping the system available.