Sample question 1 · Cost and Performance Optimization (12%)
Meridian Content Studios, a small independent film production company based in Southeast Asia, recently migrated its post-production rendering workload to AWS. The workload runs on a single on-demand r5.4xlarge instance and uses Amazon ElastiCache for Redis to cache frequently accessed scene metadata. The Head of Technology has observed that render job completion times have not improved despite the caching layer, and cache hit ratios reported in CloudWatch are consistently below 15%. A security awareness initiative at the studio has also surfaced concerns from the finance team about unpredictable monthly AWS bills. The Head of Technology must present a root-cause analysis and corrective action plan to both the technical team and the finance committee within one week. Which operational action best addresses the low cache hit ratio and allows the CloudOps engineer to present a credible, governance-aligned corrective action plan to both technical and finance stakeholders?
- A. Increase the ElastiCache node size from cache.r6g.large to cache.r6g.4xlarge to provide more memory for caching scene metadata, and report to the finance committee that the larger node will reduce render times by storing more data in memory.
- B. Audit the application's cache key design and TTL configuration using ElastiCache slow logs and CloudWatch cache hit/miss metrics to identify why metadata is not being retrieved from cache; remediate by correcting key naming inconsistencies or extending TTLs, and present the root-cause findings with before/after hit ratio targets as the governance deliverable for both stakeholder groups.
- C. Replace the ElastiCache for Redis cluster with an Amazon DynamoDB Accelerator (DAX) layer and migrate scene metadata storage to DynamoDB, presenting the architectural change as a caching modernization initiative to the finance committee.
- D. Switch the rendering workload from the r5.4xlarge instance to a p3.2xlarge GPU-accelerated instance to reduce render times, and inform the finance committee that faster renders will lower the total compute hours billed each month.
View answer & explanation
Correct answer: B. Audit the application's cache key design and TTL configuration using ElastiCache slow logs and CloudWatch cache hit/miss metrics to identify why metadata is not being retrieved from cache; remediate by correcting key naming inconsistencies or extending TTLs, and present the root-cause findings with before/after hit ratio targets as the governance deliverable for both stakeholder groups.
[Domain 6 - Cost and Performance Optimization, Subtopic 6 - Cost and Performance - Area 1 - Instance types] Core Concept: Optimize performance for operational workloads. centers on instance types as defined in the official (ISC)² exam outline. Correct Answer (B): Audit the application's cache key design and TTL configuration using ElastiCache slow logs and CloudWatch cache hit/miss metrics to identify why metadata is not being retrieved from cache; remediate by correcting key naming inconsistencies or extending TTLs, and present the root-cause findings with before/after hit ratio targets as the governance deliverable for both stakeholder groups. Why the other options are incorrect: Option A: Upgrading the ElastiCache node size increases memory capacity but does not resolve a hit ratio problem caused by key design flaws or TTL misconfiguration; a larger node will cache more data that is never retrieved correctly, wasting cost without improving performance. Option C: Migrating to DynamoDB and DAX is a significant architectural change that introduces new data modeling requirements and costs; it does not address the root cause of the existing Redis misconfiguration and would be difficult to justify to a finance committee without evidence that the current caching approach is fundamentally unsuitable. Option D: Switching to a GPU-accelerated p3.2xlarge instance may improve raw render computation speed but does not address the caching layer's low hit ratio, which is the identified performance problem; additionally, GPU instances are significantly more expensive and would likely increase the finance committee's billing concerns rather than resolve them.