DP-900 Practice Question: Describe considerations for working with non-relational data on Azure

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Data encryption at rest

  Why it's wrong here

  Data encryption at rest is a security control that uses AES-256 to protect stored data, ensuring confidentiality and compliance with regulatory standards. However, it does not affect the availability of the database during a regional outage or failover event. Encryption at rest is about preventing unauthorized access to the physical storage, not about keeping the service running or mitigating downtime, so it is not a high-availability feature.

• ✗

  Change feed

  Why it's wrong here

  The change feed is an event-driven mechanism that tracks item-level inserts, updates, and deletes, allowing you to react to changes or integrate with downstream systems. While it can enable patterns like event sourcing or replication to other stores, it does not provide any redundancy, automatic failover, or continued operations if a region becomes unavailable. The change feed is about processing data changes, not about sustaining availability, so it is irrelevant to high availability concerns.

• ✓

  Automatic failover

  Why this is correct

  Automatic failover is a core high-availability feature in Azure Cosmos DB that continuously monitors the health of regions and, if the primary region becomes unavailable, automatically redirects traffic to a secondary region without manual intervention. This ensures that the database remains accessible and can continue to serve read and write requests (depending on the consistency configuration) during an outage. It directly addresses the need for uptime and resilience, which makes it a correct answer for a high-availability question.

• ✗

  Point-in-time restore

  Why it's wrong here

  Point-in-time restore is a backup and recovery feature designed to protect against accidental data corruption, human error, or malicious deletions by allowing you to restore the database to an earlier state, typically with a granularity of seconds. However, performing a restore usually involves replacing the current database with the restored version, which can cause downtime or require redirecting applications to a new database endpoint. It is fundamentally a disaster-recovery feature for data integrity, not a proactive mechanism for maintaining continuous availability, so it does not qualify as a high-availability feature.

• ✓

  Multi-region writes

  Why this is correct

  Multi-region writes is a high-availability feature because it allows every region in a Cosmos DB account to accept both reads and writes, with data replicated synchronously across all regions. If one region experiences an outage, the remaining regions continue to handle read and write traffic without interruption, effectively removing a single point of failure. This capability directly enhances availability and is a correct answer for a question about high availability, often enabling a 99.999% uptime SLA.

Go deeper

Related to this question