ARA-C01 Question 252
Select 2Replication and failoverA global retailer runs a Snowflake account in AWS us-east-1 and has configured a secondary account in AWS us-west-2 for disaster recovery. The primary account contains a database with critical sales data and several secure views consumed by downstream applications. During a planned DR exercise, architects must ensure the west region can be promoted quickly with minimal data loss and that application queries continue to work against replicated objects after failover. Which TWO actions should the architect include in the design?
- A
Create a failover group that includes the database and the account-level objects required by the workload, then schedule replication to the secondary account.
- B
Rely on database replication alone because secure views automatically recreate any required account objects in the target account during failover.
- C
Use client connection configuration that can be updated to point applications to the secondary account after the failover group is promoted.
- D
Promote the secondary database to primary without configuring cross-region replication first, because Snowflake can backfill missing replication metadata during failover.
- E
Replicate warehouses together with the database so compute capacity and warehouse definitions are preserved automatically in the target account.
Show answer and explanation
Correct answers: A, C
Explanation
The best design is to use a failover group for the DR scope and to ensure applications can reconnect to the promoted secondary account. In Snowflake, replication and failover are configured ahead of time; failover is not a substitute for setting up replication. Database replication covers database objects, but DR designs often require supported account-level objects as well, which is why failover groups are used for broader business continuity scenarios. Architects must also plan for application connectivity because account endpoints differ between primary and secondary accounts. Snowflake documentation and best practices for replication and failover emphasize: defining replication groups or failover groups based on the required object scope, scheduling or triggering replication to meet RPO targets, and preparing client redirection procedures to meet RTO targets. Warehouses are not replicated with databases, so compute must be provisioned separately in the DR account.
- A. Correct.
Correct. For DR scenarios that require more than just database objects, Snowflake failover groups are the appropriate construct. A failover group can replicate supported databases and certain account-level objects needed for business continuity. This is important when workloads depend on account objects in addition to database contents. Replication must be configured and refreshed to keep the secondary current before a failover can occur.
- B. Incorrect.
Incorrect. Database replication does not automatically recreate all required account-level dependencies in the target account. A common misconception is that secure views make the environment self-contained. In practice, if the workload depends on supported account objects that must also exist in the DR account, architects should use failover groups rather than assume database replication alone is sufficient.
- C. Correct.
Correct. After failover, applications need to connect to the promoted secondary account. Snowflake does not transparently redirect all clients to a different account URL. A practical DR design includes an application connection strategy such as configurable connection strings, DNS indirection where appropriate, or deployment automation so applications can be redirected quickly to the promoted account.
- D. Incorrect.
Incorrect. Replication must be configured in advance. Snowflake failover is performed on already replicated objects or groups. You cannot promote a secondary that has not been established through replication, and Snowflake does not retroactively create the required replication state during a failover event.
- E. Incorrect.
Incorrect. Warehouses are compute objects and are not replicated as part of database replication. They also are not included simply because a database is replicated. A DR account should have its own compute strategy and warehouse provisioning plan. Choosing this option reflects the common but incorrect assumption that replication covers both storage objects and compute resources.