Cloud Digital Leader Scaling with Google Cloud operations — Timed 20 Questions

Question 1

A company's on-premises IT team spends 70% of their time on routine maintenance tasks: patching servers, replacing failed hardware, and upgrading storage. After migrating to Google Cloud managed services, which operational outcome should they expect?

20 minute time limit — choose an answer to begin.

Accepted Answer

The IT team can redirect time from maintenance to higher-value activities like innovation and feature development.. By migrating to fully managed Google Cloud services such as Cloud SQL, Google Kubernetes Engine, or Cloud Run, the cloud provider handles routine maintenance tasks like patching, hardware replacement, and storage upgrades. While IaaS like Compute Engine shifts hardware maintenance, OS patching remains the customer's responsibility. The key outcome is that teams can redirect time away from undifferentiated maintenance to higher-value activities. Option B correctly identifies this shift.

Answer

The IT team will need to hire more staff to manage additional cloud infrastructure.

Answer

The IT team will still perform the same tasks but remotely via the Cloud Console.

Answer

The IT team will be fully automated out of their roles by Google's AI.

Question 2

A company currently spends $200,000 annually on data center costs (hardware, power, cooling, staff). After migrating to Google Cloud, their cloud bill is $120,000 annually, but they also save $50,000 in data center costs they no longer pay. What is their net annual savings from the migration?

Accepted Answer

$80,000 annual savings ($200,000 previous cost minus $120,000 cloud cost). The net annual savings are simply the difference between the previous total cost of $200,000 and the new cloud cost of $120,000, resulting in $80,000 savings. The $50,000 data center cost savings are already reflected in the reduction from $200,000 to $120,000, as the company no longer pays those costs. No other costs remain after migration.

Answer

$30,000 (cloud cost increase of $120K minus the $50K DC savings)

Answer

$50,000 (only the data center cost savings count)

Answer

$120,000 (the entire cloud bill is savings)

Question 3

A company running critical applications on Google Cloud wants access to technical support with a response time under 1 hour for critical issues and a dedicated Technical Account Manager (TAM). Which Google Cloud support tier should they purchase?

Accepted Answer

Premium support. Premium support is the only Google Cloud tier that includes a dedicated Technical Account Manager (TAM) and meets the sub-hour response time for critical (P1) issues. Basic, Standard, and Enhanced do not include a TAM, and while Enhanced offers faster response than Standard, it still lacks the TAM. Therefore, Premium is the correct choice.

Answer

Basic support

Answer

Standard support

Answer

Enhanced support

Question 4

A company wants to optimize their Google Cloud spending. They have baseline compute workloads that run continuously 24/7 for at least one year. Which pricing option provides the greatest savings for these stable, long-running workloads?

Accepted Answer

Committed Use Discounts (CUDs) — 1-year or 3-year commitment for up to 55% savings.. Committed Use Discounts (CUDs) are ideal for stable, long-running workloads because they offer significant savings in exchange for a predictable resource usage commitment. For Compute Engine, 1-year commitments typically save about 37% and 3-year commitments save up to 55% (for resource-based CUDs). Since the workloads run continuously 24/7 for at least one year, a CUD directly matches this usage pattern and provides the greatest savings among the listed options for such non-preemptible, always-on VMs. Spot VMs offer higher raw discounts but are preemptible and unsuitable for uninterrupted 24/7 workloads.

Answer

Spot VMs — up to 91% savings over on-demand pricing.

Answer

On-demand pricing — pay per minute with no commitment.

Answer

Sustained Use Discounts — automatically applied to all running VMs.

Question 5

A company's cloud team is asked to reduce the cost of a batch data processing workload that runs for 4–6 hours each night and can tolerate interruptions. The workload currently uses standard on-demand Compute Engine VMs. Which pricing option should the team evaluate first?

Accepted Answer

Spot VMs, which offer up to 91% discount for workloads that can tolerate interruption and checkpoint/resume their work. Spot VMs are the correct first evaluation because the workload runs for a fixed 4–6 hour nightly window, can tolerate interruptions, and can checkpoint/resume its work. Spot VMs offer up to a 91% discount compared to standard on-demand VMs, making them the most cost-effective option for fault-tolerant batch processing that does not require continuous availability.

Answer

Committed Use Discounts (CUDs), by committing to 1 or 3 years of VM usage

Answer

Sustained Use Discounts (SUDs), which automatically apply when VMs run for more than 25% of a month

Answer

Capacity reservations, which guarantee capacity but do not provide a discounted pricing option for the nightly batch window

Question 6

A DevOps team wants to adopt GitOps practices for managing their Google Cloud infrastructure. Which combination of tools and practices defines a GitOps approach to cloud infrastructure management?

Accepted Answer

Storing all infrastructure as code (Terraform or Config Connector) in a Git repository, using pull requests for all changes, and automated CI/CD pipelines that apply changes and detect drift from the declared state. GitOps is defined by using a Git repository as the single source of truth for declarative infrastructure, with pull requests driving changes and automated CI/CD pipelines reconciling the actual state with the declared state. This approach enforces version control, auditability, and drift detection, which are core to managing Google Cloud infrastructure at scale with tools like Terraform or Config Connector.

Answer

Manually applying Terraform changes from engineers' local machines and documenting changes in a shared wiki

Answer

Using the Google Cloud Console to make infrastructure changes and exporting the configuration to Git after each change

Answer

GitOps only applies to application code deployment, not to cloud infrastructure management

Question 7

A cloud team wants to automatically enforce that all new Compute Engine VMs are created with a specific label (environment: production) and that no VMs are created with external IP addresses in the production project. Which Google Cloud capability enforces these organizational policies at resource creation time?

Accepted Answer

Organization Policy Service constraints that enforce no external IPs and required labels at resource creation time, blocking non-compliant VMs before they are created. Organization Policy Service constraints, specifically `compute.vmExternalIpAccess` for blocking external IPs and custom constraints for required labels, are evaluated at resource creation time. They block non-compliant VM creation before the API call succeeds, enforcing policies like 'no external IPs' and 'required labels' without relying on post-creation detection or IAM permissions.

Answer

Cloud Monitoring alerting policies that detect and notify when non-compliant VMs are created

Answer

Cloud IAM roles that prevent developers from creating VMs without the proper labels

Answer

Cloud Billing budget alerts that detect when VM spending exceeds expected amounts for labeled resources

Question 8

A team wants to set up alerts for when the error budget of their service is nearly exhausted. The service has an SLO of 99.9% availability over a 30-day rolling window. Which condition should they use in Cloud Monitoring alerting?

Accepted Answer

error budget remaining < 10%. The error budget is the amount of allowable downtime over the 30-day window (0.1% of total time). Setting an alert when the remaining budget drops below 10% gives the team early warning before the budget is fully exhausted, allowing proactive remediation. Cloud Monitoring alerting policies can use the `error_budget_remaining` metric to trigger on a threshold like < 10%. Note: In SRE practice, 'exhausted' is often used loosely to mean 'nearly exhausted' for alerting purposes; an alert at exactly 0% would be reactive and less valuable.

Answer

burn rate > 1 over 1 hour

Answer

latency > 100ms

Answer

SLI < 99.9%

Question 9

A site reliability engineer is implementing SRE practices in Google Cloud. Which TWO of the following are key principles of SRE? (Choose TWO.)

Accepted Answer

Using error budgets to balance reliability and feature velocity. Error budgets are a core SRE principle that define the acceptable level of failure (e.g., 99.9% uptime allows 0.1% errors). This budget is used to balance the tension between releasing new features (velocity) and maintaining system reliability, allowing teams to halt deployments when the budget is exhausted. Automation is another key SRE principle that focuses on eliminating toil and reducing manual, repetitive work by automating operational tasks, which improves reliability and frees teams to focus on high-value activities. Together, error budgets and automation help organizations implement SRE practices effectively.

Answer

Measuring everything with SLIs, SLOs, and SLAs

Answer

Centralizing all operations in a single team

Answer

Deploying changes only during maintenance windows

Question 10

Refer to the exhibit. A DevOps engineer notices that the alert fires even when there is only a single 5-second spike of errors that lasts for one minute. What is the most likely cause?

Accepted Answer

The trigger count is set to 1, so a single minute of high rate fires the alert. The trigger count is 1, meaning the alert fires after just one alignment period (1 minute) with a rate above threshold. A single 5-second spike can cause a high rate during that minute, triggering the alert. Option B is incorrect because the alignment period of 60s is actually appropriate; the issue is the trigger count being too low. Option C is incorrect because while a low threshold could cause false positives, the question specifically points to a single spike causing the alert, which is due to the trigger count. Option D is incorrect because the trigger count is 1, not 2.

Answer

The alignment period is too short (60s)

Answer

The threshold value (5) is too low

Answer

The trigger count is set to 2, so two consecutive periods are needed

Question 11

An application deployed on Cloud Run is experiencing increased latency. The team suspects it's not scaling quickly enough. They have set a maxScale of 10 and minScale of 0. What should they adjust to reduce cold start latency?

Accepted Answer

Set minScale to 1 to keep at least one instance warm.. Setting a minScale greater than 0 keeps at least one instance warm, reducing cold start latency. Option A (decrease container concurrency) may affect scaling behavior but does not directly address cold starts. Option C (increase CPU limit) provides more resources per instance, not warm instances. Option D (increase maxScale) only raises the upper scaling limit and does not affect cold starts.

Answer

Decrease container concurrency.

Answer

Increase CPU limit.

Answer

Increase maxScale to 20.

Question 12

A company runs batch processing jobs using preemptible VMs to reduce costs. They need to ensure these jobs can scale out significantly during peak hours. Which Compute Engine pricing model should they combine with autoscaling to optimize cost for these workloads?

Accepted Answer

Preemptible VMs with no further discounts.. Preemptible VMs already provide significant cost savings (up to 60-91% off standard VMs) and are ideal for fault-tolerant batch jobs. Combining them with autoscaling allows scaling during peak hours without any additional discounts. Sustained use discounts do not apply to preemptible VMs, and committed use discounts require upfront commitments unsuitable for variable workloads. Sole-tenant nodes are designed for dedicated hardware, not cost optimization. Therefore, using preemptible VMs with no further discounts is the optimal cost-effective solution.

Answer

Sole-tenant nodes.

Answer

Sustained use discounts.

Answer

Committed use discounts.

Question 13

Which TWO are recommended practices when configuring autoscaling for Compute Engine managed instance groups?

Accepted Answer

Use a load balancer in front of the instance group.. Options B and D are correct. Setting min and max instances prevents over- and under-scaling. Using a load balancer distributes traffic to the group. Option A is wrong because disabling the cool-down period can lead to thrashing. Option C is wrong because using a single metric is not recommended; multiple metrics provide better scaling decisions. Option E is wrong because custom metrics can be combined with standard metrics.

Answer

Disable the cool-down period to react faster to spikes.

Answer

Use a single metric (e.g., CPU utilization) for simplicity.

Answer

Scale based on custom metrics only.

Question 14

Which THREE components should a company include in their architecture to design a global web application with low latency for users worldwide?

Accepted Answer

Cloud CDN.. Cloud CDN caches content at edge locations worldwide, reducing latency by serving users from a nearby PoP. Anycast IP addressing ensures that users connect to the nearest Google edge PoP, optimizing network paths globally. The External HTTP(S) Load Balancer is a global load balancer that uses anycast IP to route user traffic to the closest available backend, providing global distribution and low latency. Backend buckets for static content are just storage backends and do not inherently reduce latency; they must be paired with CDN or a global load balancer. Regional internal load balancers only serve traffic within a single region and do not contribute to a global low-latency architecture.

Answer

Backend buckets for static content.

Answer

Regional internal load balancer.

Question 15

Refer to the exhibit. An operations team configured this Cloud Monitoring alert. They notice that the alert fires, but the associated managed instance group autoscaler does not scale up. What is the most likely reason?

Accepted Answer

The duration of 300 seconds (5 minutes) is too long, and the autoscaler uses a shorter evaluation period.. Cloud Monitoring alerts do not trigger or influence managed instance group autoscaler scaling. The autoscaler independently evaluates its configured metric and scaling policy. Therefore, none of the listed alert configuration settings (duration, threshold, alignment period, or filter) can explain why the autoscaler did not scale up; the question has no correct option.

Answer

The threshold value of 0.8 is too low.

Answer

The alignment period of 60 seconds is too short, causing unstable metrics.

Answer

The filter is missing the project ID, so it applies to all projects.

Question 16

A company runs an e-commerce platform on Google Kubernetes Engine (GKE) using autoscaling. They have a baseline workload and occasional traffic spikes during promotions. They configured a Horizontal Pod Autoscaler (HPA) for their web application pods and a Cluster Autoscaler for the node pool. The HPA targets 70% CPU utilization. During a recent sales event, traffic exceeded expectations. The operations team observed that the HPA increased the desired number of replicas to 50, but only 20 pods were running. The remaining 30 pods were in 'Pending' status. The Cluster Autoscaler logs show repeated messages: 'no capacity to scale up node pool'. The node pool is configured with a maximum of 10 nodes, each with 4 vCPUs, and currently all 10 nodes are running. The team checked the node pool's current utilization and found that nodes are near capacity. What should the team do to ensure the application scales correctly during future events?

Accepted Answer

Increase the maximum number of nodes in the node pool to allow more capacity.. The HPA requested 50 replicas, but only 20 could be scheduled because the existing 10 nodes are near capacity and the node pool is already at its maximum size. The Cluster Autoscaler cannot add more nodes because the node pool's maximum node limit prevents it from provisioning the needed additional capacity. Increasing the maximum number of nodes in the node pool (Option C) allows the Cluster Autoscaler to add nodes and accommodate the pending pods, enabling the HPA to scale as needed.

Answer

Increase the HPA target CPU utilization to 90% to reduce the number of replicas needed.

Answer

Reduce the pod resource requests for CPU so that more pods can fit on existing nodes.

Answer

Enable extra capacity by creating a second node pool with preemptible VMs.

Question 17

An e-commerce platform uses Compute Engine instances in a managed instance group behind a Cloud Load Balancer. During a flash sale, the load balancer reports increased error rates. The operations team suspects the instances are overwhelmed. Which two steps should they take to troubleshoot the issue? (Choose TWO.)

Accepted Answer

Check the CPU utilization of the instance group in Cloud Monitoring.. Checking CPU utilization in Cloud Monitoring directly reveals whether the instance group is resource-constrained. High CPU utilization indicates that the instances are overwhelmed, which aligns with the increased error rates reported by the load balancer. This metric is a primary indicator of compute capacity issues and helps validate the team's suspicion before taking corrective action. Additionally, reviewing load balancer logs in Cloud Logging can reveal specific error codes and patterns, helping to confirm whether the errors are due to backend capacity issues or other causes such as configuration problems.

Answer

Switch to a Network Load Balancer for higher throughput.

Answer

Increase the size of the instance group without investigation.

Answer

Enable HTTP health checks on the load balancer.

Question 18

A gaming company runs a real-time multiplayer game server on Google Kubernetes Engine. They want to optimize costs while ensuring low latency for players across different regions. Which three strategies should they implement? (Choose THREE.)

Accepted Answer

Use committed use discounts (CUDs) for sustained resource usage.. Committed use discounts (CUDs) are ideal for sustained resource usage because they offer significant cost savings (up to 70%) in exchange for a 1- or 3-year commitment to a minimum level of compute resources. For a gaming server that runs continuously, CUDs reduce the per-hour cost of the underlying GKE nodes, directly optimizing long-term operational expenses without impacting latency. Node auto-provisioning automatically scales the cluster based on pod resource requests, ensuring you only pay for the compute you actually need and reducing over-provisioning. Deploying GKE clusters in multiple regions and using a multi-cluster ingress routes player traffic to the nearest available cluster, reducing latency and providing resilience.

Answer

Use spot VMs with a node taint and toleration for game server pods.

Answer

Use preemptible VMs for game server pods.

Question 19

A startup has deployed a Node.js application on Cloud Run. They are seeing a higher-than-expected bill for Cloud Run usage. The application is accessed by users worldwide, and traffic patterns show occasional spikes. They want to reduce costs while maintaining performance. They currently have no concurrency management and use the default Cloud Run settings. What should they do first?

Accepted Answer

Set a maximum number of concurrent requests per container instance to reduce over-provisioning.. Cloud Run bills for container instance time during request processing. The default maximum concurrency is 80, not unlimited. To reduce costs, increase the max concurrency setting to a value your application can handle, allowing each instance to serve more simultaneous requests and reducing the number of instances needed. This directly lowers costs while maintaining performance.

Answer

Implement a caching layer with Cloud CDN.

Answer

Move the application to Compute Engine with a smaller machine type.

Answer

Reduce the container memory limit to the minimum required.

Question 20

Which TWO statements about resource monitoring and scaling on Google Cloud are correct?

Accepted Answer

You can use Stackdriver Monitoring to set up alerting policies that trigger scaling actions in managed instance groups.. Options B and C are correct. B is correct because Google Cloud's operations suite (formerly Stackdriver Monitoring) provides alerting policies that can trigger scaling actions in managed instance groups. C is correct because autoscaling policies can be configured to use metric thresholds and cooldown periods to prevent thrashing. Option A is incorrect because managed instance groups can autoscale based on multiple metrics, not only CPU utilization. Option D is incorrect because standard metrics from Google Cloud Monitoring can be used; custom metrics are not mandatory for autoscaling. Option E is incorrect because load balancers do not directly trigger scaling actions; the autoscaler handles scaling decisions.

Answer

Managed instance groups automatically scale based on CPU utilization only.

Answer

To scale based on custom metrics, you must use the autoscaler with a custom metric from Stackdriver.

Answer

Load balancers can directly trigger scaling actions without autoscaling.