Question 130 An application consists of unrelated services; one reporting function fails and should not bring down core checkout. What architecture principle is relevant? Make every checkout request block on the reporting service Give all components one required shared process Isolate failures and critical dependency paths Disable independent health checks Explanation Correct Answer C:Separating nonessential components can prevent a reporting outage from propagating. Incorrect Answers A:That couples critical operations to an optional dependency. B:A process failure could take down every function. D:Without component health information, isolation is harder to operate.
Question 131 A product recommendation engine is unavailable but shopping checkout must continue. What behavior is preferable? Restart every database automatically for each recommendation timeout Block all purchases until recommendations return Keep checkout working with recommendations temporarily omitted Erase the shopping cart after any recommendation error Explanation Correct Answer C:Graceful degradation preserves core functions while optional components fail. Incorrect Answers A:Unnecessary restarts can broaden the incident. B:An optional failure would become a full outage. D:That adds avoidable data loss.
Question 132 A team has configured automatic backups but never performs a restoration test. What reliability gap remains? Backups eliminate the need for authorization controls A green storage meter proves application consistency Actual recoverability and recovery time are unproven Backup existence guarantees every recovery objective is met Explanation Correct Answer C:A backup job can succeed while a real restoration is unusable or too slow. Incorrect Answers A:Copies do not prevent access abuse. B:Storage status alone does not validate usable restored data. D:RTO and RPO require tested processes.
Question 133 A system has redundant instances but no way to detect an unhealthy service. Which operational capability is missing? A large compute bill as the only signal Rely only on Azure Service Health notices for platform-wide incidents More resource names without telemetry Monitoring and actionable alerts for service health Explanation Correct Answer D:Redundancy needs health detection and response to support reliable operations. Incorrect Answers A:Spending does not reliably identify application health. B:Service Health does not replace application-specific health probes and operational alerts for failing instances. C:Inventory labels are not operational monitoring.
Question 134 A web app meets its targets with 100 users but fails under 10,000. What should the team do to assess performance predictability? Assume test performance is identical at every load Test under representative concurrency and resource limits Remove all logging before measuring success Only change the web app's name Explanation Correct Answer B:Representative load exposes queuing and bottlenecks missed by small tests. Incorrect Answers A:Response times can degrade under saturation. C:Without evidence, the cause and scale of degradation remain unclear. D:Naming does not improve throughput.
Question 135 An application has good average response time but frequent very slow requests. Which measurement gives more useful performance insight? Only the count of Azure subscriptions High-percentile latency under realistic load Average CPU usage without request timings Look only at mean response time over the whole day Explanation Correct Answer B:Tail latency captures user-visible slow requests hidden by averages. Incorrect Answers A:Subscription count does not represent request experience. C:CPU may not expose slow request distribution. D:The mean can hide rare but severe slow requests; high-percentile latency exposes the tail.
Question 136 Two unrelated jobs share limited compute and slow each other unpredictably. Which improvement is most defensible? Delete usage metrics so contention is invisible Right-size or isolate their workloads according to usage patterns Give every job the same global administrator password Disable all job scheduling and run everything manually Explanation Correct Answer B:Workload separation reduces unpredictable competition for resources. Incorrect Answers A:Loss of telemetry does not improve performance. C:Credentials do not isolate compute consumption. D:Manual runs do not systematically solve resource contention.
Question 137 A project estimates compute and storage but ignores retention and backup growth. What budget risk remains? All copies are deleted automatically at the end of each day Backup capacity and retention can increase ongoing charges Backups always have zero storage cost Only CPU usage determines a storage bill Explanation Correct Answer B:Predictable costs require estimating secondary data growth and policies. Incorrect Answers A:Actual retention behavior depends on configuration. C:Retention can consume billable capacity. D:Storage quantities and redundancy can matter independently.
Question 138 A global application is deployed only far from most customers, and users report slow responses. What performance factor should be considered? Changing the employee directory group name A belief that all Azure regions are equidistant from every user Only database table colors Network round-trip latency to the chosen Azure region Explanation Correct Answer D:Physical distance and network paths affect response time. Incorrect Answers A:Identity labels do not move application compute closer to users. B:Geography matters. C:Presentation choices do not change propagation delay.
Question 139 An autoscaling application rapidly alternates between adding and removing instances as a metric fluctuates. What design detail can help? Appropriate evaluation windows and cooldown behavior Delete the autoscaling metrics permanently Require every scaling event to be a new datacenter purchase Assume one threshold works for all time without tuning Explanation Correct Answer A:Stabilizing scale decisions prevents needless thrashing around a threshold. Incorrect Answers B:No metrics prevents demand-aware decisions. C:Hardware procurement is not elastic scaling. D:Workload patterns and startup times require review.
Question 140 A cloud customer is concerned about unauthorized entry to Azure facilities. Which layer of security is principally owned by Azure? Customer-managed administrative passwords Source code authorization checks Datacenter physical access controls Application-specific payroll permissions Explanation Correct Answer C:The provider protects its buildings and physical infrastructure. Incorrect Answers A:Credential control remains a tenant responsibility. B:The application owner implements its own authorization rules. D:The organization must grant users access based on business roles.
