ARA-C01 Question 240
Single answerAvailabilityA global financial services company runs a customer-facing analytics application on Snowflake. The application has a strict requirement to continue serving read and write workloads even if an entire cloud region becomes unavailable. The current design uses one Snowflake account in a single AWS region with hourly data loads and continuous transactional updates to several core tables. The architects need a design that minimizes recovery time and data loss while keeping the secondary environment ready for rapid promotion. Which architecture best meets these availability requirements?
- A
Configure database replication and failover groups between primary and secondary Snowflake accounts in different AWS regions, and use client connectivity that can be redirected to the promoted secondary account during a regional outage.
- B
Enable Time Travel with the maximum retention period on all databases so the application can continue operating from historical data if the primary region fails.
- C
Create reader accounts in another AWS region and share the production databases to those reader accounts so they can accept both read and write traffic during a regional outage.
- D
Use a multi-cluster warehouse in the primary region and increase warehouse size to improve resilience against a regional cloud outage.
Show answer and explanation
Correct answer: A
Explanation
For Snowflake availability requirements that explicitly include surviving a complete regional outage, the recommended pattern is cross-region replication with failover between Snowflake accounts. Database replication keeps supported objects synchronized to a secondary account, and failover groups extend this to broader account-level disaster recovery use cases. In practice, architects must also plan application-level connection redirection, credential/secret handling, and operational runbooks for promotion and failback. Time Travel and Fail-safe are data protection features, not regional availability solutions. Similarly, warehouse scaling and multi-cluster configuration improve compute elasticity and concurrency but do not mitigate region failure. Snowflake documentation for replication and failover, business continuity/disaster recovery, and account object replication best practices supports this design.
- A. Correct.
Correct. Cross-region database replication combined with failover groups is the Snowflake-native approach for business continuity and disaster recovery across regions. Failover groups can replicate databases and account-level objects supported by the feature, allowing a secondary account to be kept synchronized and promoted during an outage. This design supports low RPO/RTO relative to manual restore approaches and is the appropriate architecture when the requirement includes surviving a full regional outage. Redirecting client connectivity to the promoted secondary is an essential part of the operational design because applications must reconnect to the active account after failover.
- B. Incorrect.
Incorrect. Time Travel protects against logical issues such as accidental updates, deletes, or object drops within the same account, but it does not provide availability if an entire cloud region hosting that account is unavailable. Historical data retention does not make the Snowflake service in that failed region accessible, nor does it create a secondary environment for continued operations.
- C. Incorrect.
Incorrect. Reader accounts are created for data consumers and are not the mechanism for cross-region disaster recovery of a production application requiring both reads and writes. Data sharing is primarily for sharing data, and reader accounts cannot serve as full writable production failover environments. This option reflects a common misconception that sharing data to another account is equivalent to creating a DR-ready active environment.
- D. Incorrect.
Incorrect. Multi-cluster warehouses improve concurrency and can help absorb workload spikes, but they do not address regional availability. Scaling compute within one region does not protect against loss of the region itself. This option confuses performance and workload resilience with disaster recovery and cross-region availability.