Examkin

Question

Difficulty: HardManaging Compute Engine Resources

An organization runs an asynchronous media rendering service on a Compute Engine Managed Instance Group (MIG) that scales dynamically using CPU utilization metrics. During peak operational hours, brief, intermittent drops in CPU usage cause the autoscaler to prematurely terminate VM instances that are still processing long-running rendering tasks. You need to prevent premature instance termination during transient CPU fluctuations while maintaining automated horizontal scaling capabilities. Which configuration change should you make following Google-recommended best practices?

  1. Configure scale-down control settings on the MIG autoscaler by establishing a stabilization period window to delay instance deletion during temporary metric drops.Answer
  2. B
    Set the autoscaling mode of the Managed Instance Group to ONLY_SCALE_OUT so that downscaling is completely disabled.
  3. C
    Migrate the MIG worker instances to Spot VMs so that task interruptions trigger automatic instance recreation.
  4. D
    Assign the primitive Owner role (roles/owner) to the Compute Engine default service account to allow the autoscaler permission to inspect active background worker threads.

Answer

Configure scale-down control settings on the MIG autoscaler by establishing a stabilization period window to delay instance deletion during temporary metric drops.
The correct answer provides the standard Compute Engine autoscaling solution for premature scale-down events. By configuring scale-down control settings—specifically setting a stabilization period—the autoscaler calculates capacity based on peak load observed within that window. This prevents transient drops in CPU utilization from triggering premature instance deletions while preserving full autoscaling functionality.

Step-by-Step Solution

1
Analyze the operational problem and autoscaling behavior.
Identified that rapid metric fluctuations cause premature scale-down actions while background processing tasks remain incomplete.
Default autoscaler evaluation windows may react too quickly to temporary drops in CPU utilization.
2
Evaluate Compute Engine MIG autoscaler stabilization mechanisms.
Scale-down control stabilization periods allow administrators to specify a time window (e.g., 600 seconds) during which the autoscaler considers the peak required capacity rather than instantaneous minimums.
This ensures VMs are kept active long enough for temporary workload dips to subside or tasks to complete.
3
Apply Google-recommended best practices for autoscaler configuration.
Configuring `--scale-down-control time-window` on the autoscaler achieves metric smoothing without disabling automated scale-down capabilities entirely.
It maintains cost efficiency while guarding against premature VM termination.

Key Concept

Compute Engine MIG Autoscaler Scale-Down Controls and Stabilization Windows
Rate this question

Topics

Managing Compute Engine Resources

More questions from this topic