Prepare and test your skills
Prepare and test your skills
Free sample question, answered and explained below. The Professional Cloud Architect practice bank has 2,335 more like it. Try 10 free: instant score, no account.
An enterprise runs a batch and transaction processing workload on a Compute Engine regional Managed Instance Group (MIG). The workload faces diverse bottlenecks depending on customer behavior:
You need to configure an autoscaling policy that combines these signals to ensure the application maintains high performance without dropping requests.
How does the Compute Engine autoscaler evaluate these multiple signals to determine the target size of the MIG?
Additive metric aggregation assumes that each autoscaling metric operates as an incremental delta, adding its required VM count to the counts requested by all other signals (for example, 10 VMs for CPU + 15 VMs for Queue Depth + 10 VMs for Schedule = 35 VMs).
Compute Engine autoscaling does not sum recommended counts together:
max(signal_1, signal_2, ...)).Different autoscaling signals were mapped to dedicated, non-overlapping pools of instances within separate specialized instance groups rather than a single unified MIG.
Compute Engine autoscaling allows a Managed Instance Group (MIG) to automatically adjust its number of virtual machine instances according to changing workload demands. When an autoscaling policy is configured with multiple signals—such as CPU utilization, Cloud Monitoring custom metrics, load balancing serving capacity, and scaling schedules—the autoscaler independently evaluates the required capacity for every single signal during each collection cycle.
In this scenario, the workload has multiple independent bottlenecks:
By taking the maximum (largest number of VMs) among all individual recommendations, the autoscaler guarantees that whichever bottleneck or schedule demands the most capacity is fully satisfied, preventing performance degradation or request drops.

minNumReplicas and maxNumReplicas bounds across all combined evaluations.Picking the largest recommended instance count ensures high availability and eliminates under-provisioning. If one metric demands 20 VMs due to high CPU while a queue metric only requires 8 VMs, scaling to 20 VMs ensures the CPU-bound requests do not fail.
Hierarchical signal evaluation implies a strict rule-based priority order where higher-tier triggers completely mask or disable lower-tier metric signals when active.
Compute Engine autoscaling does not use a strict hierarchy where schedules or custom metrics suppress CPU utilization:
minRequiredReplicas) during their active duration. If real-time demand (e.g., CPU utilization) spikes beyond what the schedule provided, the autoscaler dynamically scales out further to meet the higher load.Google Cloud autoscaling operated under an exclusive override model where scheduled events locked the instance group to an immutable static size and ignored live telemetry.
Calculating an arithmetic average across multiple scaling signals is a conceptual method sometimes considered in custom mathematical models to smooth out short-term variance. In this hypothetical approach, if CPU utilization recommends 30 VMs and a custom queue metric recommends 10 VMs, the system would provision the average of 20 VMs.
Google Cloud Compute Engine autoscalers do not average instance recommendations across different signal types:
scaleInControl) and initialization periods (coolDownPeriodSec), not metric averaging.A custom external scaling script explicitly implemented a custom smoothing algorithm across metrics before calling the MIG API resize command, but this is neither built-in nor recommended for production SLA-backed services.
Answer them yourself and get your score the moment you finish. No account needed.
Full access includes
Keep the momentum going with these hand-picked practice scenarios
Last updated