4 of the 7 questions we have for this objective, shown with the answer and the reasoning.
Determine a strategy to improve overall operational excellence
Brennick Logistics reviewed six months of operations tickets for its EC2-based warehouse system. Disk-full incidents on the Linux application servers averaged 120 tickets per month, each fixed by the same runbook that enlarges the Amazon EBS volume and extends the file system. A Regional failover drill runs four times per year and takes a full day. Public TLS certificates issued by AWS Certificate Manager (ACM) with DNS validation are renewed by an engineer from a calendar reminder. The CTO wants to automate the work that removes the most toil with the lowest risk first.
Which automation should the solutions architect prioritize?
- ABuild an AWS Step Functions workflow that performs the full Regional failover drill end to end.The drill happens four times a year and is high risk to automate, so it removes little toil relative to the effort and risk.
- BWrite a Lambda function that renews the ACM certificates every night.ACM renews DNS-validated public certificates automatically while the validation record exists, so this is unnecessary work; the manual calendar task can simply be retired.
- CPublish disk_used_percent with the CloudWatch agent, alarm on it, and use an Amazon EventBridge rule on the alarm state change to start a Systems Manager Automation runbook that modifies the EBS volume and extends the file system.This is the highest-frequency task, it already follows a well-defined runbook, and Automation can codify the ModifyVolume API call and the in-guest extension command with low risk.
- DCreate an AWS Config custom rule that marks instances noncompliant when their disks are nearly full.AWS Config records resource configuration, not in-guest disk utilization, and a compliance finding would not fix the problem.
Why: Automation candidates are best ranked by frequency, repeatability and risk: a task that happens often and already has a stable runbook pays back fastest. The disk-full runbook is triggered by a measurable signal, so a CloudWatch agent metric, an alarm, an EventBridge rule and a Systems Manager Automation runbook close the loop without people. Rare, complex procedures such as Regional failover are better automated later, after the frequent toil is gone. The certificate option is a reminder to check whether a managed service already handles the task before building automation for it.
Determine a strategy to improve overall operational excellence
Ostrander Payments deploys its card-authorization Lambda functions with AWS SAM. Each deployment updates the function code and immediately sends all traffic to the new version. Two releases last quarter caused a spike in errors that lasted until an engineer redeployed the previous code by hand. The company wants new versions to receive 10% of traffic first and to roll back automatically if an error alarm fires, with minimal custom code.
What should the solutions architect recommend?
- AAdd a DeploymentPreference of Canary10Percent5Minutes to each function in the SAM template, publish versions behind an alias, and list the functions' error alarms under the deployment preference's Alarms property.SAM uses AWS CodeDeploy to shift alias traffic in the chosen canary pattern and rolls back automatically when any listed CloudWatch alarm enters ALARM state.
- BConfigure CloudFormation rollback triggers on the stack that reference the error alarms.Rollback triggers monitor alarms during a stack operation, but the traffic still moves to the new code all at once rather than in a 10% canary.
- CCreate a weighted alias manually after each release and have an engineer adjust the weights after checking dashboards.A manual weighted alias provides a canary but no automatic rollback and depends on a person watching dashboards.
- DEnable provisioned concurrency on the new version before it receives traffic.Provisioned concurrency reduces cold starts but does nothing to limit the impact of faulty code.
Why: An all-at-once deployment means every bad release affects all customers until someone notices. AWS SAM's DeploymentPreference integrates with AWS CodeDeploy to shift a Lambda alias gradually, using canary or linear configurations, and to roll back when specified CloudWatch alarms fire. Pre-traffic and post-traffic hooks can add validation if needed. The trap is CloudFormation rollback triggers, which watch alarms but do not stage traffic.
Determine a strategy to improve overall operational excellence
Fenwright Health runs a patient portal on an EC2 Auto Scaling group across three Availability Zones in us-east-2, with an Amazon Aurora PostgreSQL cluster. The architecture review claims the portal survives the loss of an Availability Zone, but this has never been tested. The operations team wants a repeatable monthly exercise that simulates an Availability Zone failure in production and stops on its own if the customer-facing error rate exceeds 2%.
Which solution meets these requirements?
- AHave an engineer terminate the instances in one Availability Zone each month and watch dashboards to decide when to stop.Manual termination is not repeatable, does not simulate network or database effects, and relies on a person to stop the test.
- BRun an AWS Resilience Hub assessment each month and treat a passing score as proof of recovery.Resilience Hub evaluates the architecture against RTO and RPO targets, but an assessment is not a live test of recovery actions.
- CCreate an AWS Fault Injection Service experiment template from the AZ Availability: Power Interruption scenario, target the portal resources by tag, and add a stop condition that uses a CloudWatch alarm on the 2% error rate.The scenario injects the effects of losing a zone, such as stopped instances, disrupted subnets and Aurora failover, and the alarm-based stop condition ends the experiment automatically.
- DStart an Amazon Application Recovery Controller zonal shift away from one Availability Zone each month.A zonal shift is a recovery action that moves traffic away from a zone; it does not simulate the failure of resources in that zone.
Why: Resilience claims must be exercised, and AWS Fault Injection Service provides controlled, repeatable experiments with guardrails. The AZ Availability: Power Interruption scenario from the FIS scenario library combines several actions to reproduce what an Availability Zone outage looks like to an application: it stops EC2 instances in the zone, blocks new capacity there, disrupts subnet connectivity, and runs aws:rds:failover-db-cluster when the Aurora writer is in the affected zone (targeted by tag). Stop conditions tied to CloudWatch alarms make the experiment safe to run in production because FIS halts when the customer-impact threshold is crossed. Assessments and zonal shifts are useful complements but do not inject failure.
Determine a strategy to improve overall operational excellence
Dunmore Insurance runs 300 Windows Server EC2 instances managed by AWS Systems Manager. Every instance must have the Amazon CloudWatch agent installed with the company's standard configuration and a specific registry hardening setting applied. Administrators sometimes change these settings while troubleshooting, and the drift goes unnoticed for weeks. The company wants the desired configuration reapplied automatically on a schedule.
Which solution should the solutions architect recommend?
- ARun the configuration once with Systems Manager Run Command targeted at all instances.Run Command executes a one-time action and does not keep reapplying the configuration when it drifts later.
- BCreate Systems Manager State Manager associations that apply the configuration documents to instances selected by tag on a 30-minute schedule.State Manager associations define a desired state and reapply it on the schedule, so drift is corrected automatically and compliance is reported per instance.
- CAdd the configuration commands to the EC2 user data of the launch template.User data runs at first boot, so it does not correct drift on running instances.
- DCreate an AWS Config rule that reports instances whose CloudWatch agent configuration differs from the standard.A Config rule only detects and reports noncompliance; without additional remediation it does not reapply the configuration, so the drift remains until someone fixes it.
Why: Configuration management means declaring a desired state and continuously enforcing it, not running a script once. Systems Manager State Manager associations bind an SSM document, such as AWS-ConfigureAWSPackage or a custom document, to targets and reapply it on a schedule or when new instances match the target. Association compliance then shows which instances drifted. Run Command and user data are point-in-time actions, and AWS Config evaluates recorded configuration and reports compliance rather than enforcing in-guest settings.