AWS Certified Solutions Architect - Associate SAA-C03 Practice Test 99
Question 980
A regional office needs a stable private connection into AWS but uses far less bandwidth than a full dedicated 10-Gbps Direct Connect port. It can obtain a supported hosted connection from a partner. What could reduce connection cost?
- Buy the largest dedicated port regardless of measured usage because a smaller hosted link cannot access AWS.
- Use two logical VIFs on an oversized port and assume they cut physical port charges in half.
- Choose connection capacity only by the number of AWS accounts, not measured bandwidth or resiliency requirements.
- Compare appropriately sized Direct Connect hosted connection offerings against dedicated ports, including partner and data transfer fees.
Explanation
Correct Answer
D:Hosted connectivity can match moderate needs without purchasing an oversized dedicated port, subject to provider availability and constraints.
Incorrect Answers
A:Hosted connections support applicable private connectivity use cases.
B:VIF count does not automatically eliminate the underlying link/port cost.
C:Account count is not a direct bandwidth demand metric.
Question 981
A hospital needs private connectivity available even if one Direct Connect circuit fails. The cheapest proposal uses one physical connection with two VIFs and no alternate route. How should the budget change?
- Include diverse physical connections or appropriate backup VPN paths that meet recovery needs, rather than optimizing only the single-circuit monthly price.
- Select the single cheapest circuit and treat two VIFs as independent physical redundancy.
- Omit monitoring and failover testing because service fees are all that matter.
- Assume a regional S3 replica can replace lost private network connectivity to application servers.
Explanation
Correct Answer
A:Resiliency has legitimate cost; VIFs on one port do not protect against that port's physical failure.
Incorrect Answers
B:Two virtual interfaces on the same Direct Connect physical circuit still share that circuit and provider path as a common failure boundary. Resilient connectivity requires independently planned connections and network paths where appropriate.
C:A backup path is useful only if route selection works during failure.
D:Data replication and network transport are separate requirements.
Question 982
An enterprise expects that paying for Direct Connect eliminates every internet egress fee, partner transport charge and AWS service data transfer rate. What should it check?
- Assume private VIF connectivity means all internet and inter-Region traffic is automatically unbilled.
- Review Direct Connect port/partner fees and applicable AWS data-transfer and service charges for the actual traffic paths.
- Ignore carrier fees because AWS owns every hosted connection's upstream billing.
- Use Direct Connect instead of logging to suppress all usage-based AWS billing records.
Explanation
Correct Answer
B:A private circuit has its own pricing model and does not make all surrounding AWS data transfer free.
Incorrect Answers
A:Transfer categories differ by path and service regardless of transport privacy.
C:A Direct Connect hosted connection can include partner or carrier pricing separate from AWS charges. Assuming AWS owns all upstream transport expenses understates the actual economic comparison for that connectivity option.
D:Physical connectivity does not remove the underlying accounting of data movement.
Question 983
A streaming startup runs inexpensive EC2 web servers but serves petabytes of downloads directly to external internet users. Finance optimizes only vCPU pricing, while network charges dominate. What should be modeled?
- Change instance purchase options to Savings Plans and assume that this also discounts unchanged internet transfer-out volumes.
- Move from EC2 to a larger Network Load Balancer while continuing to serve the same public byte volume, expecting load-balancer choice to eliminate internet egress charges.
- Model bytes sent to viewers, CloudFront/cache-hit potential and payload encoding, including CDN transfer/request prices alongside any reduced origin egress.
- Move the download service into two AWS Regions and duplicate every object without measuring extra replication or viewer-transfer costs.
Explanation
Correct Answer
C:When egress dominates, reducing delivered bytes and avoiding repeated origin fetches can have a larger effect than CPU right-sizing. Compare end-to-end delivery and CDN charges using real traffic and cacheability.
Incorrect Answers
A:Savings Plans can reduce eligible compute costs but do not universally eliminate separately billed internet transfer. The dominant bandwidth expense must be modeled independently.
B:A network load balancer changes the delivery path and its own billing dimensions; it does not inherently reduce petabytes of external transfers.
D:Multi-Region copies may help availability or locality under the right architecture, but they add replication and storage expense and do not automatically lower worldwide viewer egress bills.
Question 984
A desktop compliance tool polls an S3-hosted 50-MiB report every ten minutes. Reports change once or twice daily, and the client retains the last successful copy and its ETag. Direct downloads dominate transferred bytes, but the tool must detect a new version promptly. Which change can reduce redundant download volume?
- Replace each poll with an unconditional full GetObject download, then compare checksums after receiving the entire 50-MiB payload.
- Set the client to never ask S3 for new versions, because a retained local copy proves that the report cannot change.
- Copy the report into a new S3 object key before each poll, ensuring that object versions and network downloads increase together.
- Issue a conditional GetObject request using If-None-Match with the cached ETag, retrieving a body only when the object has changed.
Explanation
Correct Answer
D:S3 GetObject supports If-None-Match: if the current ETag matches, the service returns 304 Not Modified without the object body. Request charges and small response metadata still apply, but repeated large downloads can be avoided.
Incorrect Answers
A:A full GetObject retrieves the object body even when it matches the client copy, wasting repeated transfer. Comparing checksums locally cannot avoid the bytes already delivered.
B:Eliminating change detection violates the requirement to see new report versions promptly. A local copy needs an update signal or validation before it can be trusted indefinitely.
C:Creating a fresh object on each polling cycle adds S3 PUT/COPY work and may amplify storage and replication charges. It does not avoid repeated downloading of unchanged report content.
Question 985
A content platform already caches objects effectively in CloudFront and has measured the CDN bill by geographic location. Most paid subscribers are in North America and Europe. Viewers elsewhere can tolerate somewhat higher latency, and the service uses eligible pay-as-you-go CloudFront pricing. Which distribution setting should the team evaluate to reduce expensive edge-delivery charges?
- Evaluate a more restricted CloudFront price class and measure the change in delivery charges and latency for users outside the included edge locations.
- Reduce object TTLs to zero, because a cache miss automatically routes every request to the cheapest global edge location.
- Create separate identical distributions for every global country, because multiple distribution IDs automatically qualify each request for a lower pricing tier.
- Change the S3 origin bucket region and assume the origin region alone chooses the price class of each viewer-facing edge request.
Explanation
Correct Answer
A:CloudFront price classes limit the edge-location set used for pricing and can reduce delivery charges when some geographically distant viewers can accept added latency. Test performance because the default global edge set prioritizes proximity.
Incorrect Answers
B:Shorter cache lifetimes increase origin revalidation or fetches but do not configure the distribution price class. The cache hit ratio and the edge price class are separate decisions.
C:Duplicating CloudFront distributions introduces configuration and operational work without automatically selecting lower-priced geographic edge locations. Price class is an explicit distribution setting.
D:Changing the origin Region may affect origin cost and latency, but it does not by itself select the regional pricing class for CloudFront viewer-facing traffic.
Question 986
A financial report treats every inbound byte uploaded from customers as AWS internet data-transfer-out revenue loss, inflating a migration estimate. What should finance verify?
- Assume all uploads to AWS are charged identically to outgoing internet downloads for every service.
- Check the precise service pricing for data transfer in versus out and for any specialized ingest or cross-Region path.
- Assume all migration traffic is free across every AWS service, partner and network path.
- Use only EC2 vCPU list prices as the complete migration data-movement cost model.
Explanation
Correct Answer
B:AWS often prices inbound transfer differently from outbound internet traffic; the actual source/destination path matters.
Incorrect Answers
A:Ingress and egress are different pricing dimensions and may have separate rates or allowances.
C:Specific transfer tools and external providers may charge even where a basic inbound rate is zero.
D:Compute rates do not price bytes entering or leaving networks.
Question 987
A SaaS app has twelve HTTP microservices requiring host/path routing under one domain. A proposal buys a dedicated public ALB for each service despite sufficient listener rule capacity on a shared ALB. What cheaper consolidation could work?
- Require a separate ALB for every URL path because ALB cannot route based on HTTP host or path.
- Use an NLB alone for HTTP path inspection without another layer.
- Use an appropriately configured shared ALB with host/path rules and separate target groups while preserving isolation/limits.
- Place all microservices on one unmonitored EC2 instance as the only cost savings measure.
Explanation
Correct Answer
C:HTTP-aware routing can multiplex many services onto fewer load balancers, reducing unnecessary fixed ALB charges.
Incorrect Answers
A:ALB listener rules explicitly support those application conditions.
B:NLB does not offer ALB-style HTTP content routing.
D:That may save some spend but creates severe resilience and scaling limitations.
Question 988
An ecommerce site uses CloudFront pay-as-you-go pricing. Every viewer request needs a simple deterministic URL rewrite and one response-header adjustment. The logic runs without network calls, external libraries, request bodies, or origin events, and it receives tens of millions of requests daily. Which AWS edge execution approach should be assessed first for suitable capability and cost?
- Use an origin-request Lambda@Edge function to perform each viewer response-header adjustment, because origin-request events always run for every cache hit.
- Run the same simple viewer request logic in Lambda@Edge with unnecessary external SDK imports, despite no requirement for the longer runtime or origin access.
- Move URL rewriting to an S3 bucket lifecycle policy, which executes arbitrary JavaScript against every viewer HTTP request.
- Use CloudFront Functions on supported viewer request/response events, validating feature limits and comparing per-request pricing with Lambda@Edge.
Explanation
Correct Answer
D:CloudFront Functions runs lightweight JavaScript at viewer-facing events without network or filesystem access. For simple rewrites and headers at high request volume, it is a purpose-fit alternative to heavier Lambda@Edge processing; review current pricing and limits.
Incorrect Answers
A:Origin-request triggers do not run for every cached viewer response, and they are not a substitute for viewer-response header handling. This event selection fails the specified per-viewer behavior.
B:Lambda@Edge can support viewer-event processing, so it is technically possible, but its broader runtime and billing model may be unnecessary for short, isolated high-volume transformations. Compare rather than assuming it is cheapest.
C:S3 lifecycle rules transition or expire stored objects; they do not run viewer-side JavaScript on HTTP requests or modify response headers in real time.
Question 989
An application uses an ALB that handles many persistent connections, high rule evaluation volume and large payload transfer. Finance prices only the ALB fixed hourly charge and ignores capacity-unit charges. What should it estimate?
- Include ALB running hours and measured ALB LCU drivers: new and active connections, processed bytes and qualifying rule evaluations.
- Forecast ALB cost from fixed load-balancer hours plus EC2 instance-hours, but omit connection, byte and rule-evaluation dimensions.
- Price only ALB processed bytes and discard active-connection, new-connection and listener-rule metrics.
- Use Network Load Balancer flow metrics to price the Application Load Balancer without observing its HTTP listener-rule evaluations.
Explanation
Correct Answer
A:ALB pricing combines hourly operation with load-balancer capacity units based on the applicable maximum usage dimension. A workload with many persistent connections, bytes and nondefault rules needs all relevant ALB measurements rather than a flat hourly forecast or NLB flow proxy.
Incorrect Answers
B:Target compute and ALB running hours do not account for metered ALB LCU consumption when connection, traffic or rule load is material.
C:ALB capacity-unit pricing considers multiple usage dimensions, and the bill can be driven by a dimension other than bytes.
D:NLB and ALB use different capacity dimensions; NLB flow counts do not substitute for ALB-specific rules and connection measurements.
Question 990
A small HTTP app uses an NLB in front of an ALB for historical reasons. It has no static IP, PrivateLink, special protocol or other requirement demanding both layers. Finance wants to simplify. What should be considered?
- Keep both balancers because adding layers always lowers costs while preserving identical functionality.
- Remove the unnecessary layer after validating HTTP routing, client connectivity and availability requirements, retaining a single suitable ALB where possible.
- Replace the ALB with an NLB and expect URL path rules to continue unchanged.
- Replace both with an unprotected single EC2 endpoint without reviewing resilience obligations.
Explanation
Correct Answer
B:Stacking NLB and ALB incurs additional load balancer resources without benefit when the app needs only ALB HTTP features.
Incorrect Answers
A:Extra layers can increase hours, capacity charges and complexity.
C:NLB does not perform application-layer HTTP path routing.
D:Removing required balancing can create an unacceptable single point of failure.
