AWS Certified Solutions Architect - Associate SAA-C03 Practice Test 92
Question 910
A media company runs CPU-intensive transcoding jobs around the clock, averaging 90% CPU over each 24-hour period. Finance proposes replacing the stable-performance EC2 fleet with lower-base-price T3 Unlimited instances because it assumes “Unlimited” means all sustained CPU is included in that hourly price. The jobs cannot tolerate CPU-credit throttling. How should the architect evaluate this proposal?
- Adopt T3 Unlimited immediately; all CPU use above the T3 baseline is free once Unlimited mode is enabled.
- Model the proposed T3 baseline and billable surplus CPU credits over the sustained utilization profile, then benchmark total price and throughput against an appropriate nonburstable instance.
- Compare only the base EC2 per-hour price, treating the initial Unlimited CPU-credit surplus as permanently available for the sustained 90% workload.
- Choose T3 Standard mode instead because it guarantees sustained 90% CPU even after all accrued credits are depleted.
Explanation
Correct Answer
B:The baseline and rolling usage determine whether a burstable Unlimited instance incurs billed surplus CPU credit consumption. Comparing all-in charges and performance to steady-performance compute is the defensible way to assess the supposed savings.
Incorrect Answers
A:Unlimited mode can sustain CPU above the baseline, but surplus CPU credits can become billable when the rolling 24-hour average exceeds baseline. For around-the-clock 90% CPU, evaluate those charges and observed price/performance against a nonburstable instance before declaring savings.
C:The base hourly T3 price and initial credit availability do not capture billable excess CPU on a persistently busy Unlimited instance. The rolling 24-hour baseline and surplus-credit charging metrics must be included in the total-cost comparison.
D:Standard mode uses earned and accrued credits for bursts and returns to its baseline performance when credits are depleted. It cannot guarantee sustained high CPU beyond the baseline for this always-busy workload.
Question 911
AWS Compute Optimizer recommends downsizing a batch-enabled EC2 application server based on recent utilization. A weekly reconciliation job has a high p99 memory footprint, but custom memory metrics were not provided during the analyzed interval. The server has strict completion-time and latency targets. How should the architect apply this recommendation?
- Apply the recommended smaller instance type immediately, since Compute Optimizer recommendations guarantee the application’s tail latency and memory behavior.
- Use the EC2 CPU credit balance metric as the sole proof of available RAM, then resize without testing the weekly reconciliation workload.
- Treat the recommendation as a candidate; collect representative memory and peak-period telemetry, examine projected performance risk, and load-test the weekly job before selecting an instance size.
- Increase the attached EBS volume capacity while leaving the proposed reduced instance RAM unchanged, because additional disk automatically extends the EC2 guest’s physical memory.
Explanation
Correct Answer
C:Recommendations are useful for screening savings opportunities, but absent memory metrics and infrequent peaks limit the evidence. Collect memory and peak workload measurements, inspect performance-risk dimensions, and run a representative benchmark before accepting a cheaper instance.
Incorrect Answers
A:Compute Optimizer models resource behavior from supported telemetry and publishes performance-risk indications; it does not guarantee all application-specific peak memory or tail-latency conditions. A change without validation can fail a periodic batch job.
B:CPU credits describe burstable CPU entitlement rather than available physical memory. A memory-bound reconciliation task can fail or thrash on a smaller instance even if its CPU credit balance appears healthy.
D:EBS volume size concerns persistent block storage, not installed DRAM in the instance. Increasing disk space may help storage constraints but does not meet a job’s in-memory working-set requirement after a RAM reduction.
Question 912
An architect chooses instance families solely from average CPU utilization. A cache service has low CPU but nearly saturated RAM and high network throughput during bursts. Which metrics should be included?
- Ignore memory and network because CPU is the only EC2 resource that can constrain requests.
- Use EC2 instance name length as a proxy for memory and network limits.
- Purchase the instance with the most vCPUs without checking whether it has sufficient RAM or NIC performance.
- Evaluate CPU, peak memory, EBS I/O, network throughput and application response times together before choosing instance size.
Explanation
Correct Answer
D:Different workload bottlenecks can drive performance even when host CPU is comfortably low.
Incorrect Answers
A:Memory exhaustion or saturated NIC bandwidth can cause latency independently.
B:Naming does not reflect measured system resource utilization.
C:Extra CPU may be unused while another resource remains limiting.
Question 913
An international SaaS company considers moving its primary API from a Region near customers to a cheaper region. Latency, data residency and cross-Region exchange with an existing DB are relevant. Which cost model is appropriate?
- Compare regional compute rates together with network egress/replication, regulatory constraints and end-user latency.
- Choose the cheapest hourly region regardless of database geography and data residency law.
- Assume inter-Region transfer is always free whenever the two Regions belong to the same account.
- Ignore performance SLAs because cloud regions have identical global client latency.
Explanation
Correct Answer
A:A lower instance price can be outweighed by new data movement, compliance or customer-experience costs.
Incorrect Answers
B:Regional selection affects both legal constraints and network path economics.
C:Cross-Region movement may incur charges despite shared ownership.
D:Geography and carrier routing can create measurable latency differences.
Question 914
A live chat application requires p95 round-trip latency below 70 ms for European users. A distant region offers a marginal EC2 hourly discount but adds 100 ms of network round-trip. What is the appropriate decision?
- Move solely for the lowest EC2 list rate and assume network delay has no business impact.
- Keep or evaluate compute placement meeting the latency SLO, measuring total cost rather than accepting the distant region solely for minor rate savings.
- Raise the application's S3 retention period to eliminate physical signal propagation delay.
- Use extra IAM roles to reduce the round-trip distance of packets to a faraway Region.
Explanation
Correct Answer
B:An unusable user experience and violated SLO can outweigh a small direct compute discount.
Incorrect Answers
A:The added RTT directly breaches the stated experience objective.
C:S3 retention periods control how long objects are kept; they cannot reduce geographic signal propagation time or user-to-Region network latency. Regional placement, edge delivery or other routing changes address the distance component.
D:Authorization identity does not shorten physical network routes.
Question 915
Two tightly coupled microservices exchange large internal payloads hundreds of times per second across different AZs, generating significant cross-AZ transfer charges. The service can retain independent redundant replicas while reducing routine cross-AZ chatter. What should be examined?
- Put the only application replica and database copy in one AZ to eliminate all cross-AZ charges.
- Assume AZ-to-AZ data transfer is always free and no traffic analysis is needed.
- Optimize request batching/locality and same-AZ traffic patterns without sacrificing the required Multi-AZ failure tolerance.
- Increase every HTTP request's payload size to reduce transfer expenses regardless of content.
Explanation
Correct Answer
C:A design can reduce avoidable cross-zone bytes while preserving enough independent replicas for resilience.
Incorrect Answers
A:A single failure domain violates necessary availability guarantees.
B:Inter-AZ charges and latency can matter for chatty architectures.
D:Larger HTTP payloads send more bytes over the same network path and can increase metered data-transfer costs. Compression, caching and avoiding unnecessary transfers are more relevant savings measures than inflating response sizes.
Question 916
A company has EC2 Instance Savings Plans tied to one Region and family. It wants to move compute to another Region for slightly cheaper pay-as-you-go rates and expects all commitments to follow. Which calculation is needed?
- Buy a second EC2 Instance Savings Plan in the new Region before estimating how much of the old commitment will remain unused.
- Compare only the new Region on-demand instance rate and exclude ongoing Instance Savings Plan commitments from the move budget.
- Change EC2 instance cost-allocation tags to the new Region and expect the original Instance Savings Plan to re-scope automatically.
- Reassess current commitment applicability, uncovered new-region usage and stranded coverage before moving.
Explanation
Correct Answer
D:EC2 Instance Savings Plans have instance-family and Region scope. Forecast eligible usage before and after migration, measure uncovered on-demand spend in the new Region and unused old commitments, and compare total effective cost.
Incorrect Answers
A:A new location-specific commitment may cover new usage but does not erase payments on the existing plan, so total spend may increase.
B:Previously purchased commitments are still paid even if eligible usage moves away; stranded coverage affects the actual savings.
C:Tags can help attribution but do not change the family and Region applicability of a specific Savings Plan commitment.
Question 917
An organization has elastic EC2 and Fargate workloads whose daily compute costs vary with customer demand. Finance wants a cost-management alert when spending departs materially from the recent expected pattern, even if the absolute spend remains below the monthly budget. The team accepts that AWS billing data can be delayed and does not require second-by-second notification. Which AWS mechanism directly fits?
- Configure AWS Cost Anomaly Detection with an appropriate AWS services monitor and alert subscription for anomalous spending, choosing an alert threshold that fits operational significance.
- Create only an AWS Budgets monthly fixed-percentage spend alert, because a fixed 80% budget threshold independently detects every deviation from a learned daily spending baseline.
- Create a CloudWatch CPUUtilization anomaly alarm on one EC2 instance and use it as a direct measurement of account-wide discounted compute cost anomalies.
- Enable Cost and Usage Reports into S3 without a detection rule or alerting subscription, because each delivered billing row itself sends an anomaly notification.
Explanation
Correct Answer
A:AWS Cost Anomaly Detection models cost patterns and detects unexpected spend after billing data is processed. Service-level monitors and alert subscriptions support daily or relevant alerting; the mechanism is not real-time and can lag observed usage.
Incorrect Answers
B:AWS Budgets is appropriate for thresholds against planned spending or forecasts, but crossing 80% of a monthly budget does not necessarily indicate an unusual deviation from normal daily cost behavior.
C:CloudWatch CPU utilization anomalies are infrastructure metrics, not account-level dollar-cost anomalies. Discounts, data transfer, fleet composition and usage prices cannot be inferred from one instance CPU metric.
D:Cost and Usage Reports provide detailed billing records for analysis, but exporting rows to S3 does not itself configure statistical anomaly detection or recipients for alerts.
Question 918
A company has steady always-on API hosts plus unpredictable nightly ETL jobs. Finance needs to separate baseline compute usage from bursts before buying commitments. What history should it inspect?
- Purchase a long-term commitment equal to the single busiest hour of the year.
- Analyze historical EC2 utilization and costs in Cost Explorer or detailed billing views by workload and time pattern.
- Estimate baseline demand only from S3 object creation counts with no EC2 usage records.
- Exclude idle instance hours from the analysis even though they still incur charges.
Explanation
Correct Answer
B:Commitment sizing should reflect a durable baseline rather than temporary or interruptible peaks.
Incorrect Answers
A:Peak usage may not recur enough to consume committed spending.
C:Storage events do not fully describe compute billable hours.
D:Idle instance-hours are real cost and can dominate a misleading baseline.
Question 919
A finance team needs to attribute EC2 instance hours, Reserved Instance and Savings Plans coverage, account-level usage and inter-Availability-Zone transfer charges to applications. The existing monthly Billing dashboard omits the necessary hourly usage-type detail. Which AWS data source should be analyzed for granular cost attribution?
- Aggregate EC2 instance state changes in CloudTrail and estimate all discounts and network transfer charges from instance IDs alone.
- Use a high-level monthly Billing dashboard total without enabling resource identifiers, usage-type dimensions or detailed export fields.
- Export AWS Cost and Usage Reports or Data Exports CUR 2.0 line items with the required usage, account, resource and commitment dimensions and analyze them by billing period.
- Use AWS Config resource-compliance snapshots as the official detailed source of billed Savings Plans usage, effective rates and data-transfer line items.
Explanation
Correct Answer
C:CUR and CUR 2.0 provide itemized AWS usage and charge data, including lineItem usage-type dimensions and, where applicable, reservation and Savings Plans fields. With appropriate options and analysis, the report can explain the resource, account and transfer costs masked by monthly summary totals.
Incorrect Answers
A:CloudTrail records API actions but not the comprehensive billed usage, effective commitment rates, or metered inter-AZ transfer charge categories. It cannot reconstruct accurate payer-account spend from lifecycle events alone.
B:A monthly aggregate can show the total but removes the hour, operation and usage-type detail required for commitment and data-transfer attribution. It cannot reliably separate idle instance hours from other charges without richer dimensions.
D:AWS Config describes resource inventory and configuration compliance rather than AWS metered billing line items. It does not replace the Cost and Usage Report for effective commitment charges or transfer-cost breakdowns.
Question 920
Two product teams share a central platform account. Finance has categorized the platform support and data-transfer charges as SharedPlatform but wants an internal showback view allocating that shared category 70% to Team A and 30% to Team B. The real AWS bill and underlying CUR records must not be rewritten. Which AWS Billing capability matches?
- Activate an EC2 resource tag on Team A instances and assume it automatically splits platform-level support charges in the requested fixed percentages.
- Change the payer account’s payment instrument so AWS issues separate provider invoices for the 70% and 30% portions.
- Create two AWS Budgets with different alarm thresholds and expect those thresholds to reallocate SharedPlatform historical billing line items.
- Define an AWS Cost Categories split-charge rule using SharedPlatform as the source, both team categories as targets, and a fixed 70/30 allocation; inspect allocated figures in the cost-category details view.
Explanation
Correct Answer
D:Cost Categories supports fixed-percentage split-charge rules for shared source values. The allocated display is available on the cost-category details page; AWS notes that split results do not change CUR, Cost Explorer or the underlying bill.
Incorrect Answers
A:Ordinary resource cost-allocation tags help attribute eligible resource charges but do not automatically distribute a shared support category between two recipients at an arbitrary fixed ratio.
B:Changing the payer payment method does not construct a business-unit allocation report or change who incurred the shared platform line items in AWS billing data.
C:Budgets monitor or alert on spending, but their thresholds do not split a shared cost-category value or modify AWS Cost and Usage Report line items.
