COF-C03 Question 195
Single answerData replication and failoverA company runs a business-critical Snowflake deployment in AWS us-east-1 and has a secondary account in AWS us-west-2 for disaster recovery. The primary account contains several databases and one virtual warehouse used by ETL jobs. The company wants to minimize downtime during a regional outage and ensure reporting users in the secondary account can continue querying the latest replicated data after failover. Which action should the Snowflake administrator take to meet this requirement?
- A
Configure replication for the required databases and promote the secondary database during an outage; separately create warehouses in the secondary account because compute objects are not replicated.
- B
Enable replication on the primary warehouse so that failover automatically recreates the warehouse and resumes ETL jobs in the secondary account.
- C
Use database replication and failover to replicate all account objects, including users, roles, and warehouses, so the secondary account is fully identical after promotion.
- D
Create a database share from the primary account to the secondary account and rely on the share during an outage because shared databases automatically become writable after failover.
Show answer and explanation
Correct answer: A
Explanation
The key Snowflake concept is that replication and failover are primarily about replicated metadata and data for supported objects, not compute. For disaster recovery across regions, administrators typically replicate the required databases from the primary account to a secondary account and, if needed, use account-level replication/failover features for supported account objects. During an outage, the secondary database can be promoted to serve workloads. However, virtual warehouses are not replicated, so the DR account must already have appropriately sized warehouses available for reporting or ETL. This aligns with Snowflake best practices for business continuity: replicate data objects, pre-provision compute in the target account, and test failover procedures regularly. See Snowflake documentation on database replication, account replication/failover groups, and the distinction between storage objects and compute resources.
- A. Correct.
Correct. Snowflake replication/failover supports replicating supported objects such as databases (and, with the appropriate features, certain account-level objects via failover groups), but virtual warehouses are compute objects and must be created separately in the target account. In a regional outage, the administrator would promote the replicated secondary database so it becomes writable/readable in the DR account, and reporting can continue using warehouses already defined there.
- B. Incorrect.
Incorrect. Virtual warehouses are not replicated through database replication or failover. Snowflake separates storage/metadata replication from compute, so administrators must provision compute resources independently in the secondary account. A common misconception is to assume failover reproduces the entire runtime environment, including warehouses.
- C. Incorrect.
Incorrect. Database replication does not replicate all account objects, and warehouses are not included. Some account-level objects can be protected using account replication/failover groups in supported editions and configurations, but even then warehouses are still not replicated. This option overstates the scope of failover replication.
- D. Incorrect.
Incorrect. Secure data sharing is not a disaster recovery mechanism for regional outages of the primary account in this scenario. A shared database in a consumer account is read-only and does not become writable after failover. This option confuses sharing with replication/failover.