DVA-C02 Deployment Practice Question

Answer analysis

Option-by-option breakdown

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

• ✓

  Configure automatic rollback based on CloudWatch alarms.

  Why this is correct

  When a deployment causes performance degradation or errors, CloudWatch alarms can detect these issues by monitoring key metrics such as error rates, latency, or CPU utilization. Configuring automatic rollback to trigger upon these alarm states ensures that the application quickly reverts to a stable previous version, minimizing the blast radius and user impact of a faulty deployment. This proactive measure significantly reduces the duration of service disruption and enhances reliability.

• ✓

  Use a canary deployment to shift traffic gradually.

  Why this is correct

  A canary deployment strategy involves releasing the new version to a small, controlled subset of users or instances first, while the majority of traffic still goes to the stable old version. This allows for real-time monitoring of key performance indicators and user behavior on the new version without impacting the entire user base. If issues are detected, traffic can be immediately rolled back from the canary, preventing a widespread outage and enabling a safe, controlled release.

• ✗

  Disable health checks to prevent false positives.

  Why it's wrong here

  Disabling health checks, such as those utilized by Elastic Load Balancers or Auto Scaling Groups, removes a critical mechanism for identifying and isolating unhealthy instances or application components. While it might prevent 'false positives' in some specific edge cases, it fundamentally prevents the detection of genuine failures, leading to users being routed to non-functional instances. This significantly increases the risk of service degradation and outages, as problematic instances remain in service.

• ✗

  Require a manual approval step before deployment.

  Why it's wrong here

  While a manual approval step can enforce policy compliance or ensure human oversight before a deployment, it does not inherently reduce the technical risk of the deployment failing due to underlying code bugs, configuration errors, or environmental incompatibilities. In fact, it can introduce delays, slow down the deployment process, and potentially lead to human error during the approval or execution phase, rather than preventing the core technical issues that cause failures.

• ✗

  Deploy to all instances at once to ensure consistency.

  Why it's wrong here

  Deploying to all instances simultaneously, often referred to as an 'all-at-once' or 'in-place' deployment, maximizes the blast radius if the new version contains a critical bug or configuration error. While it might seem to ensure immediate consistency across the fleet, any failure will instantly impact every single user, leading to a complete service outage rather than a partial degradation. This strategy offers no opportunity for gradual testing, monitoring, or graceful rollback.

Go deeper

Related to this question