ADA-C01 Question 550
Single answerDesign and implement disaster recovery and business continuity plansA global retailer runs a business-critical Snowflake workload in AWS US_EAST_1. The company must meet a disaster recovery requirement of restoring service in another region within 1 hour if the primary region becomes unavailable, while minimizing ongoing compute cost in the standby environment. The workload includes databases, external stages, and several scheduled tasks that load and transform data. Which solution best meets these requirements?
- A
Create database replication and failover groups to a secondary account in another AWS region, replicate the required objects, and keep warehouses in the secondary account suspended until failover is needed.
- B
Use Time Travel and Fail-safe in the primary account to recover all databases, external stages, and tasks after the outage, because these features provide cross-region recovery without additional setup.
- C
Clone the production databases weekly into another region and manually recreate integrations, tasks, and users only if a disaster occurs, because zero-copy cloning provides the fastest cross-region recovery.
- D
Configure Snowpipe in the secondary region and rely on auto-ingest to rebuild the databases and scheduled tasks from the source files when the primary region fails.
Show answer and explanation
Correct answer: A
Explanation
The best answer is to use Snowflake replication and failover capabilities, specifically failover groups for business continuity and disaster recovery. Failover groups are designed to replicate and fail over a defined set of supported objects between accounts, which is the recommended approach for regional disaster recovery. This aligns with Snowflake best practices for meeting strict recovery time objectives while controlling cost by maintaining a warm standby account with suspended warehouses.
Key reasoning:
- Time Travel and Fail-safe are data protection and recovery features, but they are not substitutes for cross-region DR.
- Zero-copy cloning is useful for dev/test and operational scenarios inside Snowflake, but it is not the primary mechanism for cross-region business continuity.
- Rebuilding from staged files using Snowpipe is not sufficient for a complete DR plan because it does not preserve the broader environment and introduces too much recovery complexity.
Relevant Snowflake documentation topics include replication and failover groups, business continuity and disaster recovery planning, and the limitations of Time Travel, Fail-safe, and cloning in DR scenarios.
- A. Correct.
Correct. For cross-region disaster recovery in Snowflake, organizations should use replication and failover capabilities, typically through failover groups, to replicate supported objects between accounts in different regions/clouds where supported. This is the appropriate design for meeting a low RTO objective such as 1 hour. Keeping warehouses in the secondary account suspended minimizes ongoing compute cost because storage replication incurs cost, but warehouses only consume compute when running. This approach also supports planned or unplanned failover with a standby account ready to take over.
- B. Incorrect.
Incorrect. Time Travel and Fail-safe protect against data loss and object recovery within Snowflake, but they do not provide business continuity for a regional outage. They are not a cross-region disaster recovery mechanism and cannot restore service in another region if the entire primary region is unavailable. Also, they do not by themselves replicate account-level operational components into another region.
- C. Incorrect.
Incorrect. Zero-copy cloning does not work as a cross-region disaster recovery strategy in the way described. Clones are created within the same account and region context for existing objects and are not a substitute for replication to a standby region. Weekly cloning also creates a poor recovery point objective and manual recreation of operational objects would likely fail the 1-hour recovery requirement.
- D. Incorrect.
Incorrect. Snowpipe can help ingest files, but it is not a disaster recovery solution for restoring a complete Snowflake environment. Rebuilding databases and tasks from source files would be operationally slow, error-prone, and unlikely to meet the stated RTO. In addition, this option ignores the need to replicate metadata, security-related objects, and supported account objects required for business continuity.