1Z0-1067-25 Question 78
Select 2Your team maintains a microservices application on Oracle Cloud Infrastructure (OCI) that experiences unpredictable spikes in traffic. You have configured an instance pool with an auto-scaling policy, but you notice that scaling operations sometimes lag behind rapid traffic surges, and other times the environment scales out too aggressively, raising costs. Which two steps should you take to improve your auto-scaling strategy for better responsiveness and cost control?
- A
- Create separate scale-in and scale-out policies with distinct thresholds and cooldown periods to fine-tune responsiveness and prevent rapid fluctuations.
- B
- Disable custom metrics and rely solely on CPU utilization as the primary trigger for all scale-out actions.
- C
- Configure dynamic scaling rules to set the minimum number of instances in the pool to zero, preventing any running instances during off-peak times.
- D
- Integrate custom application metrics with the auto-scaling configuration to achieve more granular scaling decisions beyond purely CPU-based thresholds.
Show answer and explanation
Correct answers: A, D
Explanation
When designing auto-scaling for fluctuating workloads on OCI, best practices include configuring multiple, targeted scaling policies and taking advantage of custom metrics. Separate scale-in and scale-out policies let you specify different thresholds and cooldown periods to avoid excessive scaling. Enhancing auto-scaling triggers with application-level or custom metrics can significantly improve accuracy. For more information, refer to the OCI documentation on instance pool auto-scaling (https://docs.oracle.com/en-us/iaas/Content/Compute/Tasks/autoscalinginstancepools.htm).
- A. Correct.
- Correct. Having separate and well-structured scale-in and scale-out policies, including appropriate utilization thresholds and cooldown periods, allows you to avoid over-scaling or under-scaling in response to sudden traffic changes. Cooldown periods prevent thrashing, where instances are repeatedly added and removed in quick succession.
- B. Incorrect.
- Incorrect. Relying solely on CPU utilization may overlook other critical performance indicators (e.g., custom application metrics or memory usage). Disabling custom metrics generally decreases the auto-scaling mechanism� accuracy.
- C. Incorrect.
- Incorrect. Setting the minimum number of instances to zero might reduce costs drastically, but it also risks zero availability during unexpected spikes. A baseline of at least one or more instances typically ensures the service remains continuously available.
- D. Correct.
- Correct. Integrating custom application metrics (e.g., request queues, memory usage, or request rate) with auto-scaling provides deeper insights into real-time load. This helps the system scale more accurately and cost-effectively than CPU-only triggers.