Question 670
An architect proposes this change: Design multi-region entry routing, compatible identity dependencies and replicated application configuration/secrets for the failover region. The workload context is: A two-region application needs resilient login and DNS dependencies as well as redundant web app servers. Which evidence would most directly validate the change?
- Verify connections follow the failover-group listener after planned failover, with read-only workloads on intended replicas.
- Disable one regional origin and verify global clients are routed to a healthy origin over HTTPS.
- Test URI-path routing and zone outage behavior while WAF inspects requests.
- Deactivate primary DNS/routing and verify secondary web login and Key Vault configuration still function.
Explanation
Correct Answer
D:The decisive evidence is: Deactivate primary DNS/routing and verify secondary web login and Key Vault configuration still function. This measures the mechanism of the planned change, not an unrelated Azure service: Design multi-region entry routing, compatible identity dependencies and replicated application configuration/secrets for the failover region.
Incorrect Answers
A:The alternative test targets: Use Azure SQL failover group endpoints and supported read-only listener routing with region-appropriate replication. The scenario's actual proposed remedy is: Design multi-region entry routing, compatible identity dependencies and replicated application configuration/secrets for the failover region. Verify that one by: Deactivate primary DNS/routing and verify secondary web login and Key Vault configuration still function.
B:The decisive evidence is: Deactivate primary DNS/routing and verify secondary web login and Key Vault configuration still function.
C:That observation pertains to a separate failure: A single self-managed reverse-proxy VM fails and all regional web routes stop. Its cause is: L7 regional routing was deployed without managed gateway zone resilience. It does not confirm the result of this design. The direct proof would be: Deactivate primary DNS/routing and verify secondary web login and Key Vault configuration still function.
Question 671
During architecture troubleshooting, a team reports: The secondary web app serves static pages but authentication and configuration retrieval fail. Context: A two-region application needs resilient login and DNS dependencies as well as redundant web app servers. Which underlying design mistake explains the observed behavior most precisely?
- Client connection configuration bypasses the service's failover-group listener.
- There is no global HTTP edge entry point with origin health-based routing.
- L7 regional routing was deployed without managed gateway zone resilience.
- Regional failover planning omitted shared identity/configuration dependencies and allowed network paths.
Explanation
Correct Answer
D:The recorded behavior follows from: Regional failover planning omitted shared identity/configuration dependencies and allowed network paths. The appropriate mitigation would be: Design multi-region entry routing, compatible identity dependencies and replicated application configuration/secrets for the failover region. Verify with: Deactivate primary DNS/routing and verify secondary web login and Key Vault configuration still function.
Incorrect Answers
A:The alternative diagnosis “Client connection configuration bypasses the service's failover-group listener” does not match the observed symptom: The secondary web app serves static pages but authentication and configuration retrieval fail. Context: A two-region application needs resilient login and DNS dependencies as well as redundant web app servers. The recorded behavior follows from: Regional failover planning omitted shared identity/configuration dependencies and allowed network paths. The appropriate mitigation would be: Design multi-region entry routing, compatible identity dependencies and replicated application configuration/secrets for the failover region. The evidence instead supports “Regional failover planning omitted shared identity/configuration dependencies and allowed network paths”.
B:The alternative diagnosis “There is no global HTTP edge entry point with origin health-based routing” does not match the observed symptom: The secondary web app serves static pages but authentication and configuration retrieval fail. Context: A two-region application needs resilient login and DNS dependencies as well as redundant web app servers. The recorded behavior follows from: Regional failover planning omitted shared identity/configuration dependencies and allowed network paths. The appropriate mitigation would be: Design multi-region entry routing, compatible identity dependencies and replicated application configuration/secrets for the failover region. The evidence instead supports “Regional failover planning omitted shared identity/configuration dependencies and allowed network paths”.
C:The alternative diagnosis “L7 regional routing was deployed without managed gateway zone resilience” does not match the observed symptom: The secondary web app serves static pages but authentication and configuration retrieval fail. Context: A two-region application needs resilient login and DNS dependencies as well as redundant web app servers. The recorded behavior follows from: Regional failover planning omitted shared identity/configuration dependencies and allowed network paths. The appropriate mitigation would be: Design multi-region entry routing, compatible identity dependencies and replicated application configuration/secrets for the failover region. The evidence instead supports “Regional failover planning omitted shared identity/configuration dependencies and allowed network paths”.
Question 672
The design change is: Design multi-region entry routing, compatible identity dependencies and replicated application configuration/secrets for the failover region. The original failure was: The secondary web app serves static pages but authentication and configuration retrieval fail. Which ongoing control best prevents recurrence of the same root cause?
- Review Front Door health probes and origin priority before application releases.
- Run complete synthetic login and settings tests against each region during continuity reviews.
- Monitor replication lag and exercise listener failover after major releases.
- Monitor backend health, WAF logs and certificate expiry for the gateway.
Explanation
Correct Answer
B:The preventive control is: Run complete synthetic login and settings tests against each region during continuity reviews. It targets the earlier cause: Regional failover planning omitted shared identity/configuration dependencies and allowed network paths. Confirm ongoing effectiveness by: Deactivate primary DNS/routing and verify secondary web login and Key Vault configuration still function.
Incorrect Answers
A:The alternative diagnosis “Review Front Door health probes and origin priority before application releases” does not match the observed symptom: The design change is: Design multi-region entry routing, compatible identity dependencies and replicated application configuration/secrets for the failover region. The preventive control is: Run complete synthetic login and settings tests against each region during continuity reviews. It targets the earlier cause: Regional failover planning omitted shared identity/configuration dependencies and allowed network paths. The evidence instead supports “Run complete synthetic login and settings tests against each region during continuity reviews”.
C:The alternative diagnosis “Monitor replication lag and exercise listener failover after major releases” does not match the observed symptom: The design change is: Design multi-region entry routing, compatible identity dependencies and replicated application configuration/secrets for the failover region. The preventive control is: Run complete synthetic login and settings tests against each region during continuity reviews. It targets the earlier cause: Regional failover planning omitted shared identity/configuration dependencies and allowed network paths. The evidence instead supports “Run complete synthetic login and settings tests against each region during continuity reviews”.
D:The alternative diagnosis “Monitor backend health, WAF logs and certificate expiry for the gateway” does not match the observed symptom: The design change is: Design multi-region entry routing, compatible identity dependencies and replicated application configuration/secrets for the failover region. The preventive control is: Run complete synthetic login and settings tests against each region during continuity reviews. It targets the earlier cause: Regional failover planning omitted shared identity/configuration dependencies and allowed network paths. The evidence instead supports “Run complete synthetic login and settings tests against each region during continuity reviews”.
Question 673
A design workshop has this requirement: An engineering company runs a legacy Windows service that requires installing a kernel driver and full access to the guest operating system. Operators currently observe: The app cannot install its required host driver on a managed web platform. Select the ONE design remedy that best addresses the root cause.
- Select Azure Virtual Machines with appropriate VM size, networking and disk configuration rather than a restricted PaaS host.
- Evaluate Azure Dedicated Host with eligible VM families and documented licensing eligibility.
- Use an appropriate VM temporary/local scratch disk only for disposable files and store durable results on managed disks or services.
- Use Azure VMware Solution if supported and economically justified for the required VMware operational model.
Explanation
Correct Answer
A:This option addresses the actual design defect: The chosen hosting abstraction does not expose the operating system features the app requires. The correction is: Select Azure Virtual Machines with appropriate VM size, networking and disk configuration rather than a restricted PaaS host. Confirm the outcome with: Install the driver on a test VM and validate service startup and OS-level operations.
Incorrect Answers
B:The alternative diagnosis “Evaluate Azure Dedicated Host with eligible VM families and documented licensing eligibility” does not match the observed symptom: The app cannot install its required host driver on a managed web platform. This option addresses the actual design defect: The chosen hosting abstraction does not expose the operating system features the app requires. The correction is: Select Azure Virtual Machines with appropriate VM size, networking and disk configuration rather than a restricted PaaS host. The evidence instead supports “Select Azure Virtual Machines with appropriate VM size, networking and disk configuration rather than a restricted PaaS host”.
C:The alternative diagnosis “Use an appropriate VM temporary/local scratch disk only for disposable files and store durable results on managed disks or services” does not match the observed symptom: The app cannot install its required host driver on a managed web platform. This option addresses the actual design defect: The chosen hosting abstraction does not expose the operating system features the app requires. The correction is: Select Azure Virtual Machines with appropriate VM size, networking and disk configuration rather than a restricted PaaS host. The evidence instead supports “Select Azure Virtual Machines with appropriate VM size, networking and disk configuration rather than a restricted PaaS host”.
D:The alternative diagnosis “Use Azure VMware Solution if supported and economically justified for the required VMware operational model” does not match the observed symptom: The app cannot install its required host driver on a managed web platform. This option addresses the actual design defect: The chosen hosting abstraction does not expose the operating system features the app requires. The correction is: Select Azure Virtual Machines with appropriate VM size, networking and disk configuration rather than a restricted PaaS host. The evidence instead supports “Select Azure Virtual Machines with appropriate VM size, networking and disk configuration rather than a restricted PaaS host”.
Question 674
The planned architecture is: Select Azure Virtual Machines with appropriate VM size, networking and disk configuration rather than a restricted PaaS host. Workload context: An engineering company runs a legacy Windows service that requires installing a kernel driver and full access to the guest operating system. Which acceptance test demonstrates that the required behavior works?
- Inspect assigned VM physical host placement and verify the licensed workload runs on the dedicated host.
- Install the driver on a test VM and validate service startup and OS-level operations.
- Redeploy the worker and confirm business results remain while disposable scratch may disappear.
- Migrate a pilot VM and validate vSphere management, network extension and application connectivity.
Explanation
Correct Answer
B:The acceptance test is specific: Install the driver on a test VM and validate service startup and OS-level operations. Its purpose is to establish that the proposed architecture works: Select Azure Virtual Machines with appropriate VM size, networking and disk configuration rather than a restricted PaaS host.
Incorrect Answers
A:Inspecting assigned VM physical host placement and verifying the licensed workload runs on the dedicated host would not install the driver on a test VM and validate service startup and OS-level operations.
C:The alternative test targets: Use an appropriate VM temporary/local scratch disk only for disposable files and store durable results on managed disks or services. The scenario's actual proposed remedy is: Select Azure Virtual Machines with appropriate VM size, networking and disk configuration rather than a restricted PaaS host. Verify that one by: Install the driver on a test VM and validate service startup and OS-level operations.
D:Migrating a pilot VM and validating vSphere management, network extension and application connectivity would not install the driver on a test VM and validate service startup and OS-level operations.
Question 675
The customer states: An engineering company runs a legacy Windows service that requires installing a kernel driver and full access to the guest operating system. An incident reveals: The app cannot install its required host driver on a managed web platform. What cause best accounts for this evidence?
- Critical state was stored on nonpersistent temporary local storage.
- A replacement IaaS platform was chosen despite strict VMware compatibility requirements.
- The architecture does not provide the required customer-controlled physical-host tenancy.
- The chosen hosting abstraction does not expose the operating system features the app requires.
Explanation
Correct Answer
D:The incident is explained by: The chosen hosting abstraction does not expose the operating system features the app requires. The reported symptom is: The app cannot install its required host driver on a managed web platform. The corresponding design fix is: Select Azure Virtual Machines with appropriate VM size, networking and disk configuration rather than a restricted PaaS host.
Incorrect Answers
A:Critical state was stored on nonpersistent temporary local storage. The chosen hosting abstraction does not expose the operating system features the app requires.
B:A replacement IaaS platform was chosen despite strict VMware compatibility requirements. The chosen hosting abstraction does not expose the operating system features the app requires.
C:The architecture does not provide the required customer-controlled physical-host tenancy. The chosen hosting abstraction does not expose the operating system features the app requires.
Question 676
Workload context: An engineering company runs a legacy Windows service that requires installing a kernel driver and full access to the guest operating system. The implemented remedy is: Select Azure Virtual Machines with appropriate VM size, networking and disk configuration rather than a restricted PaaS host. Which ongoing practice most specifically prevents recurrence of this failure: The app cannot install its required host driver on a managed web platform.?
- Audit scratch versus persistent paths in application configuration before image changes.
- Review AVS node capacity, connectivity and license implications before each migration wave.
- Review VM image patching and guest configuration when changing OS versions.
- Reassess host capacity, failure recovery and license inventory during renewal.
Explanation
Correct Answer
C:The key follow-up control is: Review VM image patching and guest configuration when changing OS versions. It avoids recurrence of the scenario in which: The app cannot install its required host driver on a managed web platform. It addresses: The chosen hosting abstraction does not expose the operating system features the app requires.
Incorrect Answers
A:Auditing scratch versus persistent paths in application configuration would not review VM image patching and guest configuration when changing OS versions.
B:Reviewing AVS node capacity, connectivity and license implications would not review VM image patching and guest configuration when changing OS versions.
D:Reassess host capacity, failure recovery and license inventory during renewal would not review VM image patching and guest configuration when changing OS versions.
Question 677
A solution architect is reviewing a workload: A stateless web API runs on identical Windows VMs and must automatically scale out when CPU remains above a target. The observed incident is: The API exhausts capacity because two manually provisioned VMs never scale. Which architecture choice most directly removes the identified weakness?
- Select eligible GPU VM sizes or supported GPU-backed managed compute with compatible drivers and allocation quotas.
- Select Azure Virtual Machines with appropriate VM size, networking and disk configuration rather than a restricted PaaS host.
- Deploy Azure Virtual Machine Scale Sets with a suitable autoscale rule, image and load-balancing integration.
- Choose Azure App Service with suitable plan and autoscale features for the supported web runtime.
Explanation
Correct Answer
C:This option addresses the actual design defect: Capacity management lacks automated VM instance scaling and health integration. The correction is: Deploy Azure Virtual Machine Scale Sets with a suitable autoscale rule, image and load-balancing integration. Confirm the outcome with: Apply controlled CPU load and verify new scale-set instances join the healthy serving pool.
Incorrect Answers
A:Selecting eligible GPU VM sizes or supported GPU-backed managed compute with compatible drivers and allocation quotas would not deploy Azure Virtual Machine Scale Sets with a suitable autoscale rule, image and load-balancing integration.
B:Selecting Azure Virtual Machines with appropriate VM size, networking and disk configuration rather than a restricted PaaS host would not deploy Azure Virtual Machine Scale Sets with a suitable autoscale rule, image and load-balancing integration.
D:Choosing Azure App Service with suitable plan and autoscale features for the supported web runtime would not deploy Azure Virtual Machine Scale Sets with a suitable autoscale rule, image and load-balancing integration.
Question 678
An architect proposes this change: Deploy Azure Virtual Machine Scale Sets with a suitable autoscale rule, image and load-balancing integration. The workload context is: A stateless web API runs on identical Windows VMs and must automatically scale out when CPU remains above a target. Which evidence would most directly validate the change?
- Benchmark inference throughput and verify GPU device use and memory under load.
- Apply controlled CPU load and verify new scale-set instances join the healthy serving pool.
- Install the driver on a test VM and validate service startup and OS-level operations.
- Deploy the portal and verify managed HTTPS, health checks and horizontal scaling.
Explanation
Correct Answer
B:The decisive evidence is: Apply controlled CPU load and verify new scale-set instances join the healthy serving pool. This measures the mechanism of the planned change, not an unrelated Azure service: Deploy Azure Virtual Machine Scale Sets with a suitable autoscale rule, image and load-balancing integration.
Incorrect Answers
A:That observation pertains to a separate failure: Inference falls below performance target on CPU-only VMs. Its cause is: Compute selection does not meet model accelerator requirements. It does not confirm the result of this design. The direct proof would be: Apply controlled CPU load and verify new scale-set instances join the healthy serving pool.
C:This measures the mechanism of the planned change, not an unrelated Azure service: Deploy Azure Virtual Machine Scale Sets with a suitable autoscale rule, image and load-balancing integration.
D:That observation pertains to a separate failure: The team operates unnecessary webserver VMs and manually patches them for routine hosting. Its cause is: A full IaaS VM hosting model was retained though managed App Service meets the constraints. It does not confirm the result of this design. The direct proof would be: Apply controlled CPU load and verify new scale-set instances join the healthy serving pool.
Question 679
During architecture troubleshooting, a team reports: The API exhausts capacity because two manually provisioned VMs never scale. Context: A stateless web API runs on identical Windows VMs and must automatically scale out when CPU remains above a target. Which underlying design mistake explains the observed behavior most precisely?
- Capacity management lacks automated VM instance scaling and health integration.
- A full IaaS VM hosting model was retained though managed App Service meets the constraints.
- The chosen hosting abstraction does not expose the operating system features the app requires.
- Compute selection does not meet model accelerator requirements.
Explanation
Correct Answer
A:This is the failure mechanism rather than an incidental symptom: Capacity management lacks automated VM instance scaling and health integration. It accounts for: The API exhausts capacity because two manually provisioned VMs never scale. The architectural repair is: Deploy Azure Virtual Machine Scale Sets with a suitable autoscale rule, image and load-balancing integration.
Incorrect Answers
B:The alternative diagnosis “A full IaaS VM hosting model was retained though managed App Service meets the constraints” does not match the observed symptom: The API exhausts capacity because two manually provisioned VMs never scale. Context: A stateless web API runs on identical Windows VMs and must automatically scale out when CPU remains above a target. This is the failure mechanism rather than an incidental symptom: Capacity management lacks automated VM instance scaling and health integration. It accounts for: The API exhausts capacity because two manually provisioned VMs never scale. The evidence instead supports “Capacity management lacks automated VM instance scaling and health integration”.
C:The alternative diagnosis “The chosen hosting abstraction does not expose the operating system features the app requires” does not match the observed symptom: The API exhausts capacity because two manually provisioned VMs never scale. Context: A stateless web API runs on identical Windows VMs and must automatically scale out when CPU remains above a target. This is the failure mechanism rather than an incidental symptom: Capacity management lacks automated VM instance scaling and health integration. It accounts for: The API exhausts capacity because two manually provisioned VMs never scale. The evidence instead supports “Capacity management lacks automated VM instance scaling and health integration”.
D:The alternative diagnosis “Compute selection does not meet model accelerator requirements” does not match the observed symptom: The API exhausts capacity because two manually provisioned VMs never scale. Context: A stateless web API runs on identical Windows VMs and must automatically scale out when CPU remains above a target. This is the failure mechanism rather than an incidental symptom: Capacity management lacks automated VM instance scaling and health integration. It accounts for: The API exhausts capacity because two manually provisioned VMs never scale. The evidence instead supports “Capacity management lacks automated VM instance scaling and health integration”.
Question 680
A design review identifies: A stateless web API runs on identical Windows VMs and must automatically scale out when CPU remains above a target. The failure is: The API exhausts capacity because two manually provisioned VMs never scale. Select TWO distinct decisions that together provide the immediate technical solution AND an ongoing prevention control. A verification-only step does not replace either responsibility.
- Review autoscale threshold, scale-in rules and image version before peak seasons.
- Review VM image patching and guest configuration when changing OS versions.
- Choose Azure App Service with suitable plan and autoscale features for the supported web runtime.
- Select eligible GPU VM sizes or supported GPU-backed managed compute with compatible drivers and allocation quotas.
- Deploy Azure Virtual Machine Scale Sets with a suitable autoscale rule, image and load-balancing integration.
Explanation
Correct Answers
A:This is an ongoing protection specific to the fault: Review autoscale threshold, scale-in rules and image version before peak seasons. The recurrence risk is: Capacity management lacks automated VM instance scaling and health integration. It should be checked using: Apply controlled CPU load and verify new scale-set instances join the healthy serving pool.
E:The targeted implementation is: Deploy Azure Virtual Machine Scale Sets with a suitable autoscale rule, image and load-balancing integration. The problem is not generic resource availability: Capacity management lacks automated VM instance scaling and health integration. Verify the result by: Apply controlled CPU load and verify new scale-set instances join the healthy serving pool.
Incorrect Answers
B:Guest patch and VM image reviews may reduce operational risks, but cannot make a fixed pair of VMs scale automatically when CPU remains high.
C:App Service could host a compatible web API, but the stated design keeps its Windows VM runtime and asks for autoscale of VM instances rather than a platform migration.
D:GPU VM selection addresses accelerator capacity, not autoscale of CPU-bound stateless Windows web server instances.
Designing Microsoft Azure Infrastructure Solutions (AZ-305) Practice Tests