SAA-C03 Design High-Performing Architectures Practice Test 6 — 15 Questions

Question 1

A media company serves on-demand video to viewers worldwide from an Amazon S3 bucket in us-east-1. Viewers in Asia and Europe report slow start times and buffering. The team wants to reduce first-byte latency for these global viewers while keeping the content in a single origin bucket and avoiding changes to the application's object URLs. Which approach BEST meets these requirements?

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

Create an Amazon CloudFront distribution with the S3 bucket as the origin and use Origin Access Control (OAC) to restrict bucket access to CloudFront.. Edge caching with CloudFront is the canonical way to lower first-byte latency for a globally distributed audience while keeping a single S3 origin. Origin Access Control keeps the bucket private and works with the existing object keys, so the application does not need URL changes. Transfer Acceleration, Cross-Region Replication, and storage-class changes address upload speed, regional copies, or cost rather than edge delivery.

Answer

Move the video objects to an S3 bucket configured with S3 Intelligent-Tiering and enable Requester Pays so viewers fetch objects directly from us-east-1.

Answer

Enable S3 Transfer Acceleration on the bucket and update the application to use the accelerated endpoint for all object requests.

Answer

Configure S3 Cross-Region Replication to copy objects into buckets in ap-southeast-1 and eu-west-1, then have the application redirect viewers to the nearest regional bucket.

Question 2

A media analytics company ingests telemetry from millions of devices into an Amazon Kinesis Data Stream with 12 shards. Consumer Lambda functions read the stream and write to Amazon S3. Peak ingest is 8 MB/second, and consumers occasionally fall behind, causing iterator age to climb above 60 seconds. The team wants to reduce consumer lag without changing the producer. Which action should they take?

Accepted Answer

Increase the number of shards in the stream so each consumer Lambda invocation processes fewer records per shard.. Consumer lag in Kinesis is ultimately bounded by per-shard throughput and shard-level parallel processing. Adding shards increases both the aggregate ingest capacity and the number of parallel Lambda consumers, which clears the backlog without touching producers. Enhanced fan-out and higher concurrency address different bottlenecks, and longer retention only preserves data rather than speeding consumption.

Answer

Raise the stream's retention period from 24 hours to 168 hours so consumers have more time to catch up.

Answer

Enable enhanced fan-out on the stream so each consumer gets dedicated 2 MiB/second read throughput per shard.

Answer

Increase the Lambda function's reserved concurrency limit so more invocations run in parallel.

Question 3

A media company serves on-demand video to viewers worldwide from an Amazon S3 bucket in us-east-1. Viewers in Asia and Europe report slow start times and buffering. The team wants to reduce first-byte latency for these global viewers while keeping the content in a single origin bucket and avoiding changes to the application's object URLs. Which approach BEST meets these requirements?

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

Amazon CloudFront with the S3 bucket as the origin. Serving media to a globally distributed audience requires caching at edge locations near viewers. CloudFront caches objects from the S3 origin at hundreds of points of presence, so European viewers fetch segments from a nearby edge instead of from us-east-1. Because the distribution presents a domain name that fronts the bucket, the application's S3 object paths continue to work.

Answer

Amazon S3 same-Region replication with a bucket policy allowing European access

Answer

S3 Cross-Region Replication to an eu-west-1 bucket

Answer

S3 Transfer Acceleration on the bucket

Question 4

A financial analytics team runs a read-heavy reporting workload against an Amazon Aurora MySQL cluster. The primary instance is being overwhelmed by long analytical queries, and the team wants to offload this read traffic without changing the application's connection logic to a different endpoint. Which two actions will BEST achieve this? (Choose two.)

Accepted Answer

Enable Aurora Auto Scaling on the cluster and attach a policy that targets the reader endpoint.. Offloading read traffic in Aurora means serving those reads from Aurora Replicas rather than the writer. Adding replicas and pointing the reporting application at the reader endpoint moves the analytical queries off the primary, and Aurora Auto Scaling adjusts the number of replicas to match demand. Together they scale reads elastically without changing connection logic or over-provisioning the writer.

Answer

Create a separate Aurora cluster and configure Aurora Global Database between the two clusters.

Answer

Enable Multi-AZ deployment on the Aurora cluster to add a standby that serves read traffic.

Answer

Increase the instance class of the primary writer instance to a larger memory-optimized size.

Question 5

A retail company runs a containerized order-processing service on Amazon ECS with AWS Fargate. The service writes events to an Amazon Kinesis Data Streams stream, and downstream consumers read from that stream. During flash sales, producers occasionally receive ProvisionedThroughputExceededException errors, and consumers fall behind. The team wants to increase stream capacity and reduce consumer lag. (Choose two.)

Accepted Answer

Enable enhanced fan-out for the consumers using RegisterStreamConsumer. Both the write-side exceptions and the read-side lag trace back to per-shard capacity. Adding shards raises aggregate write and read capacity so producers stop exceeding limits, and enhanced fan-out gives each consumer dedicated read throughput per shard so consumers no longer compete for the shared 2 MiB per second read budget. Together they address both symptoms of the flash-sale burst.

Answer

Switch the producers to use the Kinesis Producer Library with aggregation enabled

Answer

Enable server-side encryption on the Kinesis stream with an AWS KMS customer managed key

Answer

Configure the ECS service to use the Fargate Spot capacity provider

Question 6

A media company runs a video-processing pipeline on Amazon EC2 instances in a single Availability Zone. The instances write output files to an Amazon S3 bucket in the same Region. During peak hours, the pipeline's throughput is limited, and the team wants to increase aggregate write performance to S3 without changing the application code. The instances are spread across multiple subnets in the same VPC. Which action will MOST improve the aggregate write performance to S3?

Accepted Answer

Configure an S3 gateway VPC endpoint and route S3 traffic through it.. The instances are in the same Region as the bucket, so the most effective way to improve aggregate write performance is to use an S3 gateway VPC endpoint. This keeps traffic on the AWS private network, bypassing NAT gateways and the public internet, which reduces latency and increases throughput. Other options either target long-distance transfers, increase instance capacity without addressing the network path, or add cross-Region latency.

Answer

Enable S3 Transfer Acceleration on the bucket and have the instances use the accelerated endpoint.

Answer

Move the S3 bucket to a different AWS Region closer to the instances.

Answer

Increase the size of the EC2 instances and enable enhanced networking.

Question 7

A media company runs a stateless web application on a fleet of Amazon EC2 instances behind an Application Load Balancer. The instances mount a shared NFS file system that stores user-generated images. During evening peak hours, users report that image uploads are slow, and CloudWatch metrics show the NFS file server's disk throughput is saturated. The company wants to improve upload performance while keeping the existing application code and mount path unchanged. Which solution will meet these requirements with the LEAST operational effort?

Accepted Answer

Migrate the shared file system to Amazon EFS with the Max I/O performance mode and mount it on the EC2 instances.. The saturated NFS server is the throughput bottleneck, so the fix must increase shared storage performance without changing the application's NFS mount path. Amazon EFS with Max I/O mode is purpose-built for high aggregate throughput and concurrent access, and it is a managed service that preserves the mount path, minimizing operational effort while directly relieving the upload bottleneck.

Answer

Enable Amazon EBS Multi-Attach on the existing NFS server's volume and attach it to all EC2 instances.

Answer

Add more EC2 instances to the Auto Scaling group and enable cross-zone load balancing on the load balancer.

Answer

Replace the NFS file system with an Amazon S3 bucket and modify the application to use the AWS SDK for uploads.

Question 8

A genomics research platform runs an Amazon ElastiCache for Redis cluster behind an application tier. The workload is read-dominated: the same reference-genome fragments are fetched thousands of times per minute, with a small write rate. The team wants to scale reads horizontally without changing application code that uses the Redis cluster endpoint, and they must keep a primary node for writes. Which configuration should they choose?

Accepted Answer

Enable cluster mode disabled with one primary and multiple read replicas, and configure the application to use the Redis primary endpoint for writes and the reader endpoint for reads.. Horizontal read scaling with a retained primary is achieved by cluster mode disabled plus read replicas, using the reader endpoint for read traffic and the primary endpoint for writes. The reader endpoint distributes incoming connections across replicas automatically, so no cluster-aware client or topology handling is needed, matching the low write rate and heavy repeated-read pattern.

Answer

Enable cluster mode disabled with a single shard and add read replicas, then direct reads to the cluster endpoint, which load-balances read traffic across replicas.

Answer

Enable cluster mode enabled with multiple shards, and let the application read from and write to the configuration endpoint so Redis distributes reads across shards automatically.

Answer

Enable cluster mode enabled with a single shard and multiple read replicas, and use the cluster's configuration endpoint with a cluster-aware client that supports reading from replicas.

Question 9

A financial analytics platform ingests millions of small JSON records per hour into Amazon Kinesis Data Streams. A fleet of consumer applications reads the records, aggregates them, and writes results to Amazon S3. The team observes that consumers are frequently throttled with ReadProvisionedThroughputExceeded errors, and that a few shards are much hotter than others because the partition key is a customer ID with skewed distribution. The team wants to improve read throughput and reduce hot-shard throttling with minimal changes to the consumer code. Which two actions should the team take? (Choose two.)

Accepted Answer

Increase the number of shards in the Kinesis data stream to distribute the load more evenly.. ReadProvisionedThroughputExceeded on Kinesis data streams occurs when consumers exceed the per-shard read limit or when multiple consumers share a shard. Adding shards increases total read capacity and spreads the skewed partition key across more hash buckets, while enhanced fan-out gives each consumer a dedicated 2 MiB/sec per shard, eliminating contention. Together they resolve both the throttling and the hot-shard problem with minimal code changes.

Answer

Enable server-side encryption on the Kinesis data stream with an AWS KMS customer managed key.

Answer

Switch the consumer applications to use the Kinesis Client Library (KCL) with the default configuration.

Answer

Increase the retention period of the Kinesis data stream from 24 hours to 168 hours.

Question 10

A financial services company runs a real-time analytics application on Amazon EC2 instances that read from an Amazon DynamoDB table. The table has a partition key of 'AccountId' and a sort key of 'TransactionTimestamp'. The application performs many queries for a specific account over a range of timestamps. During peak loads, the application experiences throttling on read operations. The team wants to reduce throttling and improve read performance without changing the table schema. Which solution will meet these requirements?

Accepted Answer

Increase the provisioned read capacity units (RCUs) for the table and enable auto scaling.. The application is experiencing throttling due to insufficient read capacity. Increasing provisioned RCUs and enabling auto scaling directly provides more read throughput and automatically adjusts to workload changes. This maintains the existing table schema and query patterns. Other options either do not address the capacity issue, introduce unnecessary complexity, or may not be cost-effective for sustained high loads.

Answer

Switch the table to on-demand capacity mode.

Answer

Enable DynamoDB Accelerator (DAX) for the table and direct read queries to the DAX cluster.

Answer

Create a global secondary index (GSI) with a partition key of 'TransactionTimestamp' and a sort key of 'AccountId'.

Question 11

A company runs a web application on Amazon EC2 instances in an Auto Scaling group behind an Application Load Balancer. The application stores session state in a MySQL database on a single Amazon RDS for MySQL instance. During peak traffic, the database CPU reaches 100%, and users experience slow response times. The company wants to improve performance and reduce database load without changing the application's session handling logic. Which solution will meet these requirements?

Accepted Answer

Migrate the session state to Amazon ElastiCache for Redis and update the application to use it.. Session state stored in a relational database creates a high volume of small reads and writes that saturate CPU during peak traffic. Moving session state to ElastiCache for Redis, an in-memory key-value store, offloads that traffic and provides sub-millisecond latency, reducing database load and improving response times. The application's session handling logic remains the same; only the session store backend changes, which fits the requirement.

Answer

Enable Multi-AZ deployment for the RDS instance and increase the instance size.

Answer

Enable Amazon RDS Performance Insights and tune the slowest SQL queries.

Answer

Create a read replica of the RDS instance and point the application's session reads to the replica.

Question 12

A company runs a high-traffic e-commerce website on AWS. The website uses an Amazon RDS for MySQL database. During flash sales, the database experiences heavy read traffic, causing performance degradation. The company wants to offload read traffic from the primary database and improve read scalability. The application requires strong consistency for some queries, but can tolerate eventual consistency for others. Which solution should a solutions architect implement to improve read performance while meeting the consistency requirements?

Accepted Answer

Configure the application to send strong consistency queries to the primary database and eventually consistent queries to a read replica.. The best solution is to route strong consistency queries to the primary database and eventually consistent queries to a read replica. This offloads read traffic while preserving the required consistency for critical queries. Other options either do not address consistency requirements or require significant application changes without guaranteeing strong consistency.

Answer

Use Amazon ElastiCache for Redis to cache the results of frequent read queries.

Answer

Migrate the database to Amazon Aurora MySQL and create Aurora Replicas, then direct read-only queries to the replicas.

Answer

Create an RDS read replica and configure the application to direct read-only queries to the replica.

Question 13

A company runs a real-time analytics application that ingests data from thousands of IoT devices. The data is written to an Amazon Kinesis Data Stream and then processed by an AWS Lambda function that writes aggregated results to Amazon DynamoDB. The Lambda function is configured with a batch size of 100 records and a maximum batching window of 5 seconds. During peak hours, the Lambda function is throttled, and the iterator age metric increases. The company wants to improve processing throughput and reduce iterator age without changing the downstream DynamoDB table. Which solution will meet these requirements?

Accepted Answer

Increase the number of shards in the Kinesis data stream and enable parallelization factor on the Lambda event source mapping.. Lambda throttling and rising iterator age indicate that the event source mapping cannot keep up with the incoming record rate. Increasing the number of shards raises the stream's total throughput, and enabling the parallelization factor allows multiple Lambda invocations per shard, up to 10, which multiplies processing concurrency. Together they increase throughput and reduce iterator age without modifying the downstream DynamoDB table.

Answer

Switch the Lambda function to use a larger memory allocation and increase the maximum batching window to 60 seconds.

Answer

Enable enhanced fan-out on the Kinesis data stream and configure the Lambda function as a consumer.

Answer

Increase the Lambda function's reserved concurrency and reduce the batch size to 10 records.

Question 14

A financial analytics team runs a nightly batch job that reads approximately 20 TB of Parquet files from Amazon S3, joins them with a 50 GB reference dataset, and writes aggregated results back to S3. The job currently runs on a fleet of Amazon EC2 instances and takes nine hours, and the team wants the transformation to complete faster without managing cluster infrastructure. Which solution BEST improves performance while minimizing operational overhead?

Accepted Answer

Run the job on AWS Glue with Apache Spark, tune the number of DPUs, and enable the Glue Data Catalog for the reference dataset.. AWS Glue is the managed, serverless Spark service that removes cluster provisioning while still allowing the worker count to be tuned for a large distributed join. Combining it with the Glue Data Catalog gives Spark partition and column statistics on the Parquet data, cutting the scanned bytes. The other approaches either reintroduce infrastructure management, cap parallelism, or cannot handle the reference dataset join at this scale.

Answer

Create an Amazon Redshift cluster and use COPY to load the Parquet files before running the aggregation.

Answer

Move the Parquet files to Amazon EFS and run the join on a larger EC2 instance type with more memory.

Answer

Use S3 Batch Operations to run a Lambda function over each Parquet object and aggregate the results.

Question 15

A financial analytics platform runs on Amazon EC2 instances in a private subnet and queries an Amazon RDS for MySQL Multi-AZ DB instance. The workload is read-heavy, and the team wants to offload read traffic without changing application code. What should a solutions architect recommend?

Accepted Answer

Create a read replica and update the application's connection string to point to the read replica endpoint.. A read replica is the correct solution because it creates a separate endpoint for read-only queries, allowing the application to offload SELECT traffic from the primary DB instance. This improves performance without changing the application's data model or requiring a migration. The primary instance remains available for writes, and the replica can be scaled independently. This is a standard AWS pattern for read-heavy relational workloads.

Answer

Increase the size of the primary DB instance and enable storage autoscaling to handle the read load.

Answer

Enable Multi-AZ on the primary DB instance and direct all read queries to the standby.

Answer

Migrate the database to Amazon DynamoDB and use global secondary indexes to serve analytical queries.