Microsoft AZ-104 Exam Dumps, Practice Test Questions

More workers spread requests across instances but each instance still has the original memory limit. A request that is too large for one worker can continue failing.

Autoscale adds parallel instances of the same SKU. A memory-heavy request still runs within one worker and can exhaust its unchanged RAM allocation.

The process architecture can alter address-space limits, but does not guarantee the underlying plan has more physical RAM. The case identifies the per-worker memory SKU as insufficient.

Apps in a common plan share that plan’s underlying allocated compute. They do not automatically receive isolated per-app worker pools just because one was deployed later.

App Service scaling changes worker resources, not tenant subscription ownership. It cannot create a separate subscription per app.

Worker scale changes do not inherently remove custom-domain DNS configuration. It is an unrelated application connectivity setting.

Azure Monitor autoscale uses explicitly configured metrics or schedules and applies scaling decisions at App Service plan scope, affecting the plan’s hosted apps. It is a valid choice for measured CPU thresholds but does not meet the stated requirement for hands-off HTTP traffic-based per-app scaling without custom metric rules. The team needs the separate automatic scaling feature.

Manual scale out changes the worker capacity to an administrator-selected count and does not react automatically to individual apps experiencing different HTTP bursts. Scheduling someone to adjust instances each day adds operational work and can leave one app either underprovisioned or overprovisioned. The scenario requires automatically varying capacity based on actual incoming requests.

Deployment slots isolate application versions and support testing or swap operations; they are not a separate worker autoscaling mechanism. Adding staging slots does not establish HTTP traffic-triggered per-app instance limits or prewarmed capacity. The correct control is the App Service scale-out feature on the supported plan, with the applicable instance settings.

More workers can increase capacity and cost, but does not guarantee the site’s application process remains loaded after inactivity. Always On addresses the identified unload behavior.

Health checks monitor endpoint responsiveness and influence unhealthy instance routing; they are not the same as the Always On periodic keep-warm setting.

A deployment slot's routing configuration does not set the production site's idle-unload behavior. Always On on an eligible App Service tier sends periodic keep-alive requests to keep the process warm.

A plan’s OS worker type is set by its platform configuration. An app setting cannot make Windows worker instances run Linux containers.

Scaling out increases capacity of the same plan characteristics, including OS. It cannot transform the platform type of the worker pool.

Selecting an application runtime does not change the operating system underlying the plan’s worker instances. A compatible Linux plan remains required.

Automatic scaling is primarily driven by HTTP demand and configured app limits. Custom CPU-based conditions use Azure autoscale rules instead.

A fixed instance count does not define CPU-driven scale-out and scale-in rules. Azure Monitor autoscale policies need appropriate metric thresholds and capacity bounds.

Always On maintains an application worker’s activity but does not define CPU thresholds or plan-level scale-out rules.

RBAC authorization cannot unlock features excluded by the hosting plan SKU. Plan tier, not role privilege, is the blocker.

Deployment slots are a hosting-tier capability, not a setting that grants a Free plan an unsupported feature. Upgrade to a tier with slots before configuring the staging-to-production swap.

A folder does not create an independently addressable managed slot with swap semantics. It cannot satisfy the staging and promotion requirement.

More workers increase the chance that a later request reaches an instance without the user's local cart. The state remains fragmented.

ARR affinity can increase the chance a client returns to the same available worker, but it does not make process memory durable or synchronize carts after the worker restarts, fails, or is replaced. Reliable scale-out needs shared session storage or a genuinely stateless design.

More memory can accommodate larger local carts but cannot share them between workers. A later request sent to another instance or a replacement worker still finds no prior cart state. Change state architecture rather than only worker capacity.

Adding shared plan workers may reduce capacity pressure, but all apps still use the same pool. It does not guarantee that a noisy application cannot compete with other apps for worker CPU resources at peak time.

A resource group changes management and access scope, not the worker pool hosting the web app. Without changing the hosting plan, the CPU-intensive and quiet apps still share the same compute allocation.

Always On helps keep an app loaded by preventing idle-unload behavior on applicable tiers; it does not allocate a dedicated CPU pool. The apps continue competing for the original shared plan workers.

Auto-heal can recycle workers under supported conditions, but it does not replace Health Check probes that determine which running instances receive traffic after repeated failing application readiness responses.

Always On helps prevent an idle app from unloading; it is not a readiness probe that excludes a responsive worker whose health endpoint returns failures. App Service Health Check is designed for that behavior.

Adding instances does not cause the load-balancing system to remove a specific unhealthy-but-running instance based on an application endpoint. Configure Health Check on the app with a meaningful path.

Always On keeps the application process warm on existing workers; it does not create or preserve a minimum of three workers under autoscale.

Default capacity is not an enforced floor after autoscale actions. If minimum remains one, a scale-in rule may still reduce the app below three workers overnight; set the minimum to three.

The maximum caps scale-out but does not prevent automatic scale-in to one instance. It cannot establish the requested overnight floor.