Database Migration Reviews With AI: Catching the Lock You Didn't See

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-ai-coding-database-migration-review-creators

1. A database migration that adds a new column to a table takes 500ms in development but takes 45 minutes in production. What is the most likely cause of this dramatic difference?

   1. The production database is running a different database engine than development
   2. The production table contains millions of rows, causing lock acquisition and data rewriting to take much longer
   3. The production database has more CPU cores allocated to it
   4. The development environment is using a newer version of the database software

2. What is the primary reason AI tools can help prevent production outages during database migrations?

   1. AI can automatically execute migrations without human oversight
   2. AI can scan migration scripts against known-bad patterns like exclusive locks on large tables
   3. AI can replace the need for any testing in non-production environments
   4. AI can generate the actual data that will exist in production

3. Which of the following statements about exclusive locks during migrations is correct?

   1. Exclusive locks only block write operations, not read operations
   2. An exclusive lock on a large production table can block all writes until the migration completes
   3. Exclusive locks are released automatically after 30 seconds regardless of migration progress
   4. Exclusive locks are safe to acquire on any table size in modern databases

4. A team wants to add a NOT NULL constraint to a column in a large production table. Why might this be considered a non-reversible change that requires a backfill?

   1. NOT NULL constraints cannot be removed once applied
   2. The column may contain null values that must be identified and fixed before the constraint can be safely applied
   3. NOT NULL constraints automatically delete rows with null values
   4. The database engine does not support rolling back NOT NULL constraints

5. What is an online schema change tool and why would AI recommend using one?

   1. A tool that automatically approves all schema changes without review
   2. A tool that performs schema changes without acquiring exclusive locks by using table-copying techniques
   3. A tool that runs migrations in a separate test database before production
   4. A tool that converts SQL migrations to NoSQL schema changes

6. What is one capability that AI tools currently cannot provide when reviewing database migrations?

   1. AI cannot identify lock-acquiring statements in migration scripts
   2. AI cannot catch every database-engine-specific edge case
   3. AI cannot read SQL syntax
   4. AI cannot distinguish between different types of database tables

7. What does 'lock contention' refer to in the context of database migrations?

   1. When multiple applications try to read the same data simultaneously
   2. When a lock on a database resource prevents legitimate operations from proceeding
   3. When database performance improves due to reduced query load
   4. When two databases attempt to synchronize their schemas automatically

8. Before running a risky migration on a production database, what should a team always prepare?

   1. A marketing announcement about the upcoming downtime
   2. A rollback plan that estimates how long reversal will take
   3. A celebration for the engineering team
   4. A request for additional budget for new hardware

9. Why do read replicas can be affected by certain database migrations?

   1. Read replicas ignore all migration commands and remain unchanged
   2. Some migrations require schema changes that must be replayed on replicas, causing replication lag
   3. Read replicas automatically approve all migrations without review
   4. Migrations have no impact on read replicas in any scenario

10. What is the difference between testing against 'production schema' versus 'production-shaped data'?

    1. There is no difference between these two concepts
    2. Production schema means testing with the same table structures; production-shaped data means testing with similar data volumes
    3. Production-shaped data refers to testing in a different database system
    4. Production schema is only relevant for development environments

11. When AI reviews a migration, which of the following is a valid output it might provide?

    1. A guarantee that the migration will complete without errors
    2. A lock-acquisition analysis with severity rating and blocking duration estimate
    3. An automatic approval to deploy to production immediately
    4. A complete rewrite of the migration in a different programming language

12. Why should a senior database engineer still review high-stakes migrations even when using AI tools?

    1. Senior engineers are required by law to approve all changes
    2. AI tools have already caught all possible issues, so the review is just a formality
    3. AI cannot replace the domain expertise and context that senior engineers bring to edge cases and specific infrastructure considerations
    4. Senior engineers need something to do to stay busy

13. What is a 'multi-step migration' and when would AI recommend one?

    1. A migration that runs on multiple databases simultaneously
    2. A migration split into stages to reduce lock time, such as adding a column first then populating it
    3. A migration that automatically rolls back if any step fails
    4. A migration that executes the same change multiple times for redundancy

14. Which of the following migration behaviors represents the highest risk to production availability?

    1. Adding a new table to the database
    2. Acquiring an exclusive lock on a table with 50 million rows
    3. Changing the name of a small lookup table
    4. Creating a new index during off-peak hours

15. What does it mean for a database migration to be 'reversible'?

    1. The migration can be automatically undone by the database system
    2. The migration changes can be safely rolled back without data loss or corruption
    3. The migration will always complete successfully
    4. The migration does not require any locks