passdrill
AWS SAA-C03 · RDS, DynamoDB & Databases · Card 023/024 easy

A team needs the ability to restore a DynamoDB table to its exact state as of any specific second within roughly the last five weeks, in case a bug silently corrupts data and isn't noticed for several days. The feature must be turned on explicitly per table before it is needed. Which feature should they enable, and how does a restore work?

  1. DynamoDB Streams, which retains a rolling window of item-level changes that can be replayed in place to reconstruct the table's state as of any past second
  2. Point-in-time recovery (PITR), which continuously backs up the table for a fixed 35-day window and restores the chosen second's state into a new table, since restores can't be applied in place
  3. On-demand backups taken manually every hour, restored in place over the existing table to roll it back to the desired hour
  4. Global tables, since replica tables in other regions retain older versions of items that can be queried directly for any past point in time
Next card → Shuffle