PCDOE Optimizing performance and cost Practice Test 5 — 15 Questions

Question 1

A company has a steady-state workload of 100 vCPUs running 24/7. They want to get the maximum discount possible without long-term commitment. What discount should they expect?

Accepted Answer

Sustained use discount of up to 30%. The correct answer is C: Sustained use discount of up to 30%. For steady-state workloads that run continuously, Google Compute Engine automatically applies sustained use discounts, and the maximum discount for instances running the full month is 30% (for N1 machine types). Because the company wants maximum discount without any long-term commitment, this automatic discount is exactly what they should expect. Option A is wrong because committed use discounts require a 1- or 3-year commitment, which the company explicitly wants to avoid. Option B is incorrect because the sustained use discount maximum is 30%, not 20%. Option D is wrong because sustained use discounts apply automatically to eligible steady-state usage.

Answer

Committed use discount of up to 57%

Answer

Sustained use discount of up to 20%

Answer

No discount available

Question 2

A company uses BigQuery for analytics. They notice high costs due to queries scanning large amounts of data. They want to reduce costs without sacrificing performance for urgent queries. Which approach is most cost-effective?

Accepted Answer

Partition and cluster tables, and use BI Engine for acceleration. Partitioning and clustering tables reduces the amount of data scanned by queries, and BI Engine provides an in-memory acceleration layer for fast, interactive queries. Together, they reduce costs by minimizing bytes processed while maintaining performance for urgent queries.

Answer

Use flat-rate pricing with reservations

Answer

Use on-demand pricing with query caching

Answer

Use materialized views and limit query jobs to interactive priority

Question 3

A startup runs a batch data processing job every night. The job processes large datasets and takes about 6 hours to complete. The job is designed to handle interruptions gracefully by saving checkpoint files to Cloud Storage every few minutes. The startup wants to minimize compute costs. The current setup uses a managed instance group with 10 n1-standard-4 VMs running for the entire 6-hour window. They are considering using preemptible VMs. However, they are concerned about cost stability and potential preemption. What should they do?

Accepted Answer

Use preemptible VMs for all instances, as the job can resume from checkpoints.. The job is fault-tolerant and can resume from checkpoints, making preemptible VMs suitable and significantly cheaper. Option A is not relevant as sole-tenant nodes do not reduce cost. Option B adds unnecessary complexity and still uses regular VMs. Option D requires a long-term commitment and is more expensive than preemptible VMs.

Answer

Use sole-tenant nodes to improve performance.

Answer

Use a mix of preemptible and regular VMs to ensure at least some progress if preempted.

Answer

Use regular VMs with committed use discounts for a 1-year term.

Question 4

Refer to the exhibit. A budget alert has fired for project dev-123 indicating that the cost has exceeded the budget of $1000. What should the team do next to investigate the cost overrun?

Accepted Answer

Open a BigQuery query on the billing export table, filtering by project 'dev-123' and service.. The correct option is B: opening a BigQuery query on the billing export table, filtering by project 'dev-123' and service. This is the right next step because the detailed billing export provides granular, line-item cost data (by project, service, SKU, and labels) that lets the team pinpoint exactly which services in dev-123 drove the overrun. Option A is wrong because disabling all APIs is a disruptive, non-investigative action that could break workloads without identifying the cause. Option C is wrong because an automated shutdown Cloud Function is a remediation control, not an investigation step, and could cause outages. Option D is wrong because the Cloud Console billing cost table gives only high-level aggregate costs, not the per-service breakdown needed to diagnose the overrun. Option E is wrong because creating a lower-threshold budget only changes alerting and does nothing to investigate the existing cost overrun.

Answer

Disable all APIs in project dev-123 immediately.

Answer

Set up a Cloud Function to automatically shut down resources when the budget is exceeded.

Answer

View the billing account's cost table in the Cloud Console.

Answer

Create a new budget with a lower threshold to get alerted earlier.

Question 5

A company runs a batch processing job on a Compute Engine managed instance group (MIG) that processes a nightly queue. The job typically takes 3 hours to complete, and the company wants to minimize cost while ensuring the job finishes reliably. The MIG is configured with a single instance type and no autoscaling. What should the company do to reduce cost for this workload?

Accepted Answer

Schedule the MIG to scale down to zero instances when the job is not running and scale up before the job starts using an instance schedule.. The workload runs predictably for only 3 hours each night. The most effective cost optimization is to eliminate charges during the idle 21 hours. Using an instance schedule to stop instances when not in use and start them before the job begins directly addresses this. Preemptible VMs introduce reliability risk, CUDs are for steady usage, and autoscaling does not eliminate idle costs.

Answer

Enable autoscaling on the MIG based on CPU utilization and set the minimum size to 1 and maximum size to 10.

Answer

Use a committed use discount (CUD) for the instance type used by the MIG for a 1-year term.

Answer

Configure the MIG to use preemptible VMs and set the instance template to automatically restart on preemption.

Question 6

A company runs a stateless batch processing workload on a managed instance group (MIG) with autoscaling based on average CPU utilization. Jobs finish in about 10 minutes and are not sensitive to interruption. During month-end peaks, the MIG scales out to 120 instances and cost spikes. You want to reduce cost without lengthening the average job completion time. What should you do?

Accepted Answer

Configure the MIG to use Spot VMs with a fallback to standard VMs, and keep CPU-based autoscaling.. Spot VMs are designed for fault-tolerant, stateless, or batch workloads that can tolerate preemption. Because the jobs are short and stateless, preemption has minimal impact on average completion time, and the discount directly reduces the cost of peak scale-out. Committed use discounts, metric changes, and instance resizing do not address the spiky, interruptible nature of this workload.

Answer

Change the autoscaling metric to average memory utilization and set a lower target.

Answer

Purchase a 1-year committed use discount for the peak vCPU and memory, and keep the same autoscaling policy.

Answer

Reduce the maximum number of instances in the MIG and increase the per-instance machine size proportionally.

Question 7

Your team runs a stateless batch processing workload on a managed instance group (MIG) of Compute Engine VMs that runs for approximately six hours each night. The workload is fault-tolerant and can be interrupted. You want to minimize compute cost for this workload while keeping the same machine type. What should you do?

Accepted Answer

Configure the MIG to use Spot VMs and set the instance redistribution type to opportunistic.. Spot VMs are the appropriate cost optimization for fault-tolerant, interruptible workloads because they use spare capacity at a significant discount. Because the nightly batch job can be interrupted and does not need to run continuously, using Spot VMs in the managed instance group with opportunistic redistribution minimizes cost while preserving the ability to complete the workload within its window.

Answer

Switch the MIG to a smaller machine type and enable autoscaling based on CPU utilization.

Answer

Purchase a 1-year committed use discount for the vCPUs and memory used by the MIG.

Answer

Move the workload to a single large Compute Engine instance and purchase a sustained use discount.

Question 8

A company runs a microservices application on Google Kubernetes Engine (GKE). They observe that many pods request more CPU and memory than they actually use, leading to low node utilization and high cost. They want to optimize resource usage without impacting application performance. Which approach should they take?

Accepted Answer

Enable the Vertical Pod Autoscaler (VPA) in recommendation mode and use its recommendations to manually adjust resource requests.. Over-provisioned resource requests lead to reserved but unused capacity. VPA in recommendation mode provides data-driven suggestions to right-size requests without automatic changes, allowing the company to adjust safely. HPA and node auto-provisioning address scaling, not request sizing, and setting limits equal to requests does not reduce waste.

Answer

Configure a Horizontal Pod Autoscaler (HPA) for each deployment based on CPU utilization.

Answer

Use node auto-provisioning to automatically add nodes when pods are pending due to insufficient resources.

Answer

Set resource limits equal to resource requests for all pods to ensure guaranteed QoS.

Question 9

You operate a Google Kubernetes Engine (GKE) cluster with several node pools. Some workloads have highly variable CPU demand throughout the day, while others are steady. You want to reduce cost by matching node pool capacity to actual demand without disrupting running pods. Which configuration should you use?

Accepted Answer

Enable cluster autoscaler on each node pool and set appropriate minimum and maximum node counts.. Cluster autoscaler is the GKE feature that resizes node pools in response to workload demand, scaling up when pods cannot be scheduled and scaling down when nodes are underutilized. Configuring minimum and maximum node counts per pool preserves capacity for steady workloads while allowing variable workloads to drive scaling, and the graceful drain behavior avoids disrupting running pods.

Answer

Use node auto-provisioning with a single default node pool and remove all custom node pools.

Answer

Configure a horizontal pod autoscaler for each deployment and leave node pool sizes fixed.

Answer

Enable vertical pod autoscaler in recommendation mode for all workloads and manually resize node pools weekly.

Question 10

Your organization runs a production service on Compute Engine and wants to reduce cost without reducing performance. Monitoring shows that average CPU utilization across the fleet is about 15 percent and memory utilization is about 40 percent, while the service is latency-sensitive. You want to right-size the fleet. What should you do first?

Accepted Answer

Use the Google Cloud console's VM right-sizing recommendations to identify over-provisioned instances and resize them.. Right-sizing recommendations use historical utilization data to propose machine types that match actual workload needs, which is the correct starting point when instances are over-provisioned. Because the service is latency-sensitive, using data-driven recommendations preserves necessary headroom while reducing waste, and it should precede commitment purchases so that discounts apply to correctly sized resources.

Answer

Enable autoscaling based on CPU utilization and leave the machine type unchanged.

Answer

Purchase committed use discounts for the current machine types to lower the hourly rate.

Answer

Immediately switch all instances to the smallest available machine type to maximize savings.

Question 11

Your company runs a latency-sensitive web application on Compute Engine behind a global external Application Load Balancer. Costs are rising because the backend fleet is consistently underutilized during off-peak hours but must handle sharp traffic spikes. You need to reduce cost while maintaining responsiveness during spikes. Which two actions should you take? (Choose two.)

Accepted Answer

Configure the managed instance group to use autoscaling based on load balancing capacity and set a minimum size that covers baseline traffic.. Load-balancing-capacity autoscaling with a baseline minimum lets the fleet follow traffic, shrinking during off-peak periods and expanding for spikes, while Cloud CDN reduces backend requests by serving cacheable content at the edge. Together they lower cost without sacrificing responsiveness, because capacity is matched to demand and less work reaches the origin.

Answer

Switch the backend instances to Spot VMs to reduce the hourly compute rate.

Answer

Replace the global external Application Load Balancer with a regional external passthrough Network Load Balancer.

Answer

Purchase a one-year committed use discount covering the maximum fleet size observed during spikes.

Question 12

You manage a Google Cloud project with several services. Finance wants to be notified when the project's spend is projected to exceed a monthly threshold, and they want to see which services contribute most to the cost. Which combination of Cloud Billing features should you configure?

Accepted Answer

Create a budget with a specified amount, enable email alert rules at percentages such as 50, 90, and 100 percent, and use the Cloud Billing cost table grouped by service.. Cloud Billing budgets support alert thresholds based on actual and forecast spend, which notifies finance before the monthly threshold is exceeded. The Cloud Billing cost table can be grouped by service to reveal which services drive cost, providing the visibility finance requested. Together these built-in features meet the requirement without custom data pipelines.

Answer

Enable the Cloud Billing Budget API and create a Pub/Sub topic, then rely on Cloud Monitoring dashboards grouped by service.

Answer

Set a project-level quota cap on each service and configure Cloud Monitoring to notify when quotas are approached.

Answer

Export billing data to BigQuery and build a custom report, then configure a Cloud Monitoring alert on the exported table.

Question 13

A startup runs a stateless web API on a managed instance group of Compute Engine VMs behind an external HTTP(S) load balancer. Traffic is predictable during business hours but drops to near zero at night. The team wants to reduce VM cost without changing the API code or breaking the load balancer health checks. What should they do?

Accepted Answer

Enable the managed instance group autoscaler with a CPU utilization target and set the minimum number of instances to a small non-zero value.. Autoscaling with a CPU target shrinks the managed instance group when night traffic falls and expands it when business hours resume. Keeping a minimum of one or two instances ensures the load balancer always has a healthy backend, so the API remains reachable. This matches capacity to demand without code changes and avoids the availability risk of preemptible VMs or a zero-instance schedule.

Answer

Change the load balancer to a network load balancer and enable connection draining on the backend service.

Answer

Create a schedule that scales the managed instance group to zero VMs at night and back to the normal size in the morning.

Answer

Replace the managed instance group with preemptible VMs and keep the same fixed instance count.

Question 14

A company runs a GKE cluster with a single node pool of n1-standard-4 nodes. The operations team notices that many nodes are underutilized, with average CPU and memory usage below 20%. They want to reduce cost without changing the application architecture. What should they do?

Accepted Answer

Enable cluster autoscaling with a lower minimum node count, and set pod resource requests so the scheduler can pack pods efficiently.. Cluster autoscaling is the primary tool for right-sizing a GKE node pool. When pods have accurate resource requests, the scheduler can pack them onto fewer nodes, and the autoscaler removes idle nodes. Lowering the minimum node count reduces baseline cost. The other options either do not change node count, rely on manual guesswork, or address pod count rather than node utilization.

Answer

Change the node pool machine type to n1-highmem-8 and reduce the number of nodes by half.

Answer

Enable node auto-repair and node auto-upgrade on the node pool.

Answer

Configure a horizontal pod autoscaler for all deployments and leave the node pool unchanged.

Question 15

Your organization runs a multi-tenant Google Kubernetes Engine Standard cluster. The platform team wants to reduce compute cost without degrading the latency of the tenant services. They observe that many pods request far more CPU and memory than they actually use, and that some namespaces have no resource requests at all. Which two actions should the platform team take? (Choose two.)

Accepted Answer

Configure a LimitRange in each namespace to set default CPU and memory requests for containers that do not specify them.. The two actions that address the evidence are VPA recommendations and LimitRange defaults. VPA in recommendation mode surfaces the gap between requested and actual usage so the team can right-size manifests, and LimitRange supplies default requests for pods that omit them. Together they let the scheduler place pods based on realistic numbers, which improves density and lowers node cost without touching tenant application code or latency.

Answer

Set a ResourceQuota on each namespace to cap the total CPU and memory that the namespace can request.

Answer

Enable the Horizontal Pod Autoscaler on every deployment with a CPU target of 80%.

Answer

Enable node auto-provisioning and set the cluster autoscaler to scale down aggressively during off-peak hours.