All AWS exams

Free SAA-C03 practice test: 30 questions with answers

30 original questions for AWS Certified Solutions Architect - Associate, drawn from every topic area. Answer each one to see whether you got it right and why. No sign-up.

Free questions
30
Full bank
100 questions
Real exam
65 questions, 130 min

SAA-C03 exam guide: topics, cost and format

All 30 SAA-C03 questions with answers

The same questions as the test above, grouped by exam topic. Open any question to see the answer and explanation.

Design Secure Architectures (8 questions)

  1. Question 1

    A company hosts a web application on EC2 instances in private subnets. Users must reach the application over HTTPS, and the instances must not have public IP addresses. Which architecture meets this requirement?

    1. AAssign public IPs to every instance and open inbound port 443 to 0.0.0.0/0 on its security group
    2. BA Client VPN endpoint for each remote user, with a tunnel from each user's device terminating directly on the application instances
    3. CAn internet-facing Application Load Balancer in public subnets forwarding to instances in private subnets
    4. DAn internal Network Load Balancer in a private subnet with no route to the internet gateway, listening on TCP port 443
    Show answer and explanation
    Correct answer: C

    An internet-facing ALB in public subnets receives HTTPS traffic and forwards to private instances, so the instances keep no public IPs. Public IPs on instances and VPNs per user do not match the requirement, and a private-only load balancer cannot receive internet traffic.

  2. Question 2

    An application on EC2 needs to read objects from an S3 bucket. The security team requires that the traffic not traverse the internet and that access be limited to one bucket. Which solution fits?

    1. AA public IP on the EC2 instance and a bucket ACL granting READ to AllUsers group with S3 traffic routed over the internet gateway
    2. BA NAT gateway for outbound calls with a bucket policy granting s3:GetObject to external accounts
    3. CAn IAM role on the instance with a least-privilege policy for that bucket and a gateway VPC endpoint for S3
    4. DAn IAM user with an access key hardcoded in the application code and a publicly readable bucket
    Show answer and explanation
    Correct answer: C

    An instance role with a scoped policy plus a gateway endpoint keeps access private and limited. Stored keys, public buckets and NAT paths over the internet do not meet the requirements.

  3. Question 3

    A company must store customer records in S3 and ensure that each object is encrypted with a key that the company can rotate and disable. Which option satisfies this?

    1. AClient-side encryption with the key stored in the same bucket as the encrypted objects
    2. BSSE-KMS with a customer managed key and a key policy controlling its use
    3. CNo encryption because the bucket is private and a bucket policy already restricts access
    4. DSSE-S3 with AWS-owned keys, so the company cannot rotate, disable, or write a key policy
    Show answer and explanation
    Correct answer: B

    SSE-KMS with a customer managed key lets the company control rotation, key policy and disablement. SSE-S3 has no customer-managed key controls, co-located client keys defeat encryption, and private buckets still need encryption at rest.

  4. Question 4

    A financial application must restrict which AWS API calls can be made in member accounts of an organization, regardless of the permissions granted to IAM users in those accounts. Which control should be used?

    1. AIAM permissions boundaries attached to every user as a ceiling on grants
    2. BService control policies attached to the organization root or OUs
    3. CSecurity groups on each instance to filter inbound port access
    4. DS3 bucket policies for object read access by prefix
    Show answer and explanation
    Correct answer: B

    SCPs define the maximum permissions for accounts in the organization, regardless of local IAM grants. Permissions boundaries apply per principal, security groups control network traffic, and bucket policies apply to S3 only.

  5. Question 5

    A company wants to detect potentially malicious activity such as unusual API calls and compromised instances across all of its accounts, without deploying agents. Which service should be enabled?

    1. AAmazon GuardDuty with delegated administration across the organization
    2. BAmazon Inspector scanning EC2 hosts for software package CVEs
    3. CAWS Shield Advanced for DDoS mitigation and 24x7 SRT support
    4. DAmazon Macie classifying sensitive data in S3 buckets for PII
    Show answer and explanation
    Correct answer: A

    GuardDuty analyses CloudTrail, VPC flow logs and DNS logs for threats without agents, and it can be managed across an organization. Macie targets S3 sensitive data, Shield protects against DDoS, and Inspector finds vulnerabilities rather than threats.

  6. Question 6

    A web application uses a database in a private subnet. The application's servers must reach the database, and the database must not be reachable from the internet. Which configuration is correct?

    1. ASecurity group on the database allowing inbound 3306 from 0.0.0.0/0 for the web tier
    2. BSharing the database's security group with an internet gateway attached to its VPC
    3. CPlacing the database in a public subnet with a public IP and a route to the internet gateway
    4. DSecurity group on the database allowing the database port only from the application's security group
    Show answer and explanation
    Correct answer: D

    Referencing the application security group limits database access to application instances. Open CIDR rules, public placement and internet-gateway sharing expose the database.

  7. Question 7

    A company needs to protect its web application from common exploits such as SQL injection and cross-site scripting, using managed rules at the edge. Which service fits best?

    1. AAWS Shield Standard for automatic network-layer DDoS protection on CloudFront
    2. BRoute 53 health checks probing endpoints to mark targets unhealthy
    3. CNetwork ACLs with stateless inbound rules for every known attack pattern
    4. DAWS WAF with managed rule groups associated with the CloudFront distribution or ALB
    Show answer and explanation
    Correct answer: D

    AWS WAF with managed rule groups blocks common web exploits at CloudFront or ALB. Shield Standard protects against DDoS, NACLs are not designed for application attacks, and health checks do not inspect requests.

  8. Question 8

    A company uses AWS IAM Identity Center with its corporate directory. Employees must use temporary credentials when accessing multiple AWS accounts. Which approach meets this requirement?

    1. AUse the root user for daily administration across member accounts
    2. BAssign permission sets to users or groups in IAM Identity Center for each account
    3. CCreate IAM users in every account with long-lived access keys and share the keys with the team by chat
    4. DDistribute a shared administrator password for contractors to sign in to the account console
    Show answer and explanation
    Correct answer: B

    IAM Identity Center issues temporary credentials through permission sets assigned to users and groups. Shared IAM users, root use and shared passwords violate least privilege and auditability.

Design Resilient Architectures (8 questions)

  1. Question 9

    A company runs a stateless web tier that must survive the failure of a single Availability Zone with minimal interruption. Which design is most appropriate?

    1. AA single NAT gateway in one AZ carrying outbound traffic for private subnets in every zone
    2. BA single large EC2 instance in one AZ with vertical scaling and an attached EBS root volume
    3. CInstances in one AZ with a manual restart runbook that operators follow after an outage to recover
    4. DAn Auto Scaling group spanning at least two AZs behind an Application Load Balancer
    Show answer and explanation
    Correct answer: D

    Spreading instances across AZs with a load balancer and Auto Scaling keeps the service available when one zone fails. A single instance, one-AZ setups and a single NAT gateway create single points of failure.

  2. Question 10

    An RDS database must fail over automatically to a standby in another AZ with no manual steps. Which configuration should be used?

    1. AA read replica placed in the same AZ
    2. BA larger db.r6g instance class in one AZ
    3. CMulti-AZ deployment for the DB instance
    4. DA manual snapshot taken hourly
    Show answer and explanation
    Correct answer: C

    Multi-AZ deployments maintain a synchronous standby and fail over automatically. Read replicas are for scaling reads and are not automatic failover targets, snapshots are backups, and larger instances in one AZ do not survive zone failure.

  3. Question 11

    A company needs to decouple two components so that messages are not lost when the consumer is temporarily unavailable. Which service should be used?

    1. AAmazon SQS standard queue between the producer and consumer
    2. BDirect synchronous HTTP calls from producer to consumer with no retry
    3. CA shared file on a single EC2 instance that both components poll
    4. DAn SNS topic that has no subscribed consumer yet
    Show answer and explanation
    Correct answer: A

    SQS stores messages durably until consumers process them, which prevents loss during outages. Synchronous calls fail when the consumer is down, a shared file on one instance is a single point of failure, and an unsubscribed SNS topic has no consumer to receive messages.

  4. Question 12

    A workload runs on EC2 instances and must scale out automatically with demand and replace unhealthy instances. Which feature should be configured?

    1. AA cron job that restarts every instance nightly to clear memory leaks and reset application state
    2. BAn Auto Scaling group with a scaling policy and health checks from the load balancer
    3. CManual launch of new instances by an operator when users complain about slow pages during peak hours
    4. DA fixed desired capacity of six instances with no health checks configured on the scaling group
    Show answer and explanation
    Correct answer: B

    An Auto Scaling group with scaling policies and health checks scales automatically and replaces unhealthy instances. Manual launches, fixed capacity without checks and nightly restarts do not meet the requirement.

  5. Question 13

    A company needs a static website to remain available during a regional outage, with low latency worldwide. Which architecture is most appropriate?

    1. AHost the content in S3 and serve it through CloudFront with an origin failover configuration
    2. BFTP access to a single EC2 server in one Region with anonymous login enabled and uploads written to the web root
    3. CA static IP on a developer's laptop running a local Python HTTP server on port 8080
    4. DA single EC2 web server in one Region behind a fixed Elastic IP serving static files
    Show answer and explanation
    Correct answer: A

    S3 with CloudFront provides global caching and origin failover options for resilience. A single server in one Region and laptop-based hosting are not resilient or global.

  6. Question 14

    Which THREE are elements of a resilient multi-AZ architecture? (Choose three.)

    1. AA single NAT gateway in one zone for all subnets
    2. BA single subnet covering the entire VPC
    3. CSubnets in at least two Availability Zones
    4. DA load balancer distributing traffic across zones
    5. EAuto Scaling groups that span zones
    Show answer and explanation
    Correct answer: C, D, E

    Multiple zone subnets, cross-zone load balancing and zonal Auto Scaling provide resilience. A single NAT gateway in one zone and a single subnet remove resilience.

  7. Question 15

    A company is designing a disaster recovery plan with a recovery point objective of one minute for a relational database and a recovery time objective of a few minutes. Which approach best fits?

    1. AAurora Global Database or a cross-Region replica with automated promotion procedures
    2. BWeekly backups stored on tape in an offsite vault, with cartridges rotated by courier and restored by hand
    3. CA manual restore from logs after an incident is reported by an on-call engineer
    4. DDaily snapshots copied to another Region each night for 30 days
    Show answer and explanation
    Correct answer: A

    Continuous cross-Region replication with promotion meets low RPO and RTO. Daily snapshots, weekly tapes and manual restores cannot meet a one-minute RPO.

  8. Question 16

    An application uses an ECS service and must keep running if an entire Availability Zone fails. Which configuration should be used?

    1. ARunning the service during night hours with the desired count scaled to zero during the daytime
    2. BTasks spread across at least two AZs with the service scheduler and a load balancer
    3. CA task pinned to a specific instance in one AZ using a placement group
    4. DA single task in one AZ with automatic restart on failure via ECS
    Show answer and explanation
    Correct answer: B

    Spreading tasks across AZs with the scheduler and load balancer maintains capacity during zone failures. Single tasks and pinned instances in one AZ create single points of failure.

Design High-Performing Architectures (7 questions)

  1. Question 17

    A web application experiences high read traffic for product pages that rarely change. The team wants to reduce latency and database load. Which approach is most effective?

    1. AAdd more application servers behind the load balancer without caching so reads still hit the database
    2. BIncrease the database's provisioned write capacity and IOPS so more writes can be served per second
    3. CStore product pages in local storage on each user's device with no server copy
    4. DAdd a caching layer such as ElastiCache or CloudFront caching for the product responses
    Show answer and explanation
    Correct answer: D

    Caching product responses reduces latency and database read load. Write capacity does not address reads, more servers without caching still hit the database, and device storage is not a web architecture.

  2. Question 18

    A company needs to transfer 80 TB of data from an on-premises data center to S3 over a limited network connection within a few weeks. Which service is most appropriate?

    1. AAWS Snowball Edge devices for offline transfer
    2. BUpload the files over a 10 Mbps link without estimating time
    3. CCopy the data to USB drives and mail them internally
    4. DEmail the files to AWS support
    Show answer and explanation
    Correct answer: A

    Snowball Edge devices move large datasets efficiently when the network is limited. Slow uploads over small links, email attachments and internal USB shipments are impractical.

  3. Question 19

    A workload needs the lowest possible network latency between EC2 instances that communicate heavily with each other, within a single AZ. Which placement option should be used?

    1. AA cluster placement group
    2. BA spread placement group
    3. CInstances in different Regions
    4. DInstances with public IPs only
    Show answer and explanation
    Correct answer: A

    A cluster placement group places instances close together within an AZ for low latency and high throughput. Spread groups separate instances for resilience, cross-Region placement adds latency, and public IPs do not reduce latency.

  4. Question 20

    A data analytics team queries large amounts of data stored in S3 in an ad hoc manner and wants to avoid managing servers. Which service fits best?

    1. AAmazon EFS for file sharing
    2. BAmazon Athena
    3. CAn EC2 instance with a custom query tool
    4. DAmazon RDS for a single small instance
    Show answer and explanation
    Correct answer: B

    Athena runs serverless SQL queries directly on S3 data, with no infrastructure to manage. A small RDS instance, custom EC2 tools and EFS do not fit ad hoc queries over S3 without management.

  5. Question 21

    Which TWO improve the read performance of a database that serves heavy read traffic? (Choose two.)

    1. ARemove indexes to reduce overhead on reads
    2. BIncrease write capacity only
    3. CMove all reads to the primary instance
    4. DAdd read replicas and direct read traffic to them
    5. EAdd an in-memory cache such as ElastiCache for frequently read data
    Show answer and explanation
    Correct answer: D, E

    Read replicas and caching offload reads and reduce latency. Write capacity does not help reads, removing indexes slows queries, and moving all reads to the primary adds load.

  6. Question 22

    A company needs to serve large video files to users across many countries with low latency. Which service provides this?

    1. AA single EC2 instance in one Region
    2. BAmazon CloudFront with S3 as the origin
    3. CDial-up access over analog phone lines at 56kbps
    4. DFTP servers in each office for staff
    Show answer and explanation
    Correct answer: B

    CloudFront caches content at edge locations close to users, with S3 as the origin. A single Region server adds latency for distant users, and FTP and dial-up are not appropriate.

  7. Question 23

    A machine learning team needs high-throughput, low-latency access to a shared file system from many EC2 instances in multiple AZs. Which storage option fits best?

    1. AAmazon Glacier
    2. BAmazon EFS with mount targets in each AZ
    3. CInstance store on each instance
    4. DA single EBS volume attached to one instance
    Show answer and explanation
    Correct answer: B

    EFS provides shared file storage across AZs for many instances. A single EBS volume attaches to one instance, instance store is ephemeral, and Glacier is archival.

Design Cost-Optimized Architectures (7 questions)

  1. Question 24

    A company runs a steady production workload on EC2 for a full year and wants the lowest compute cost with flexibility across instance families. Which purchasing option should be used?

    1. ACompute Savings Plans
    2. BOn-Demand instances for all workloads
    3. CSpot instances for the production database
    4. DDedicated hosts for every instance
    Show answer and explanation
    Correct answer: A

    Compute Savings Plans provide lower prices for a committed hourly spend with flexibility across families and Regions. On-Demand is more expensive, Spot can be interrupted, and dedicated hosts are used for specific licensing needs.

  2. Question 25

    A batch processing job can be interrupted and restarted, and the company wants to minimize EC2 costs. Which purchasing option fits best?

    1. AOn-Demand instances running all the time
    2. BDedicated hosts
    3. CReserved instances with a three-year term for a job that runs once a month
    4. DSpot instances with checkpointing and a fallback strategy
    Show answer and explanation
    Correct answer: D

    Spot instances are inexpensive and suit interruptible, restartable work when checkpointing exists. Reserved instances for monthly jobs, always-on On-Demand and dedicated hosts increase costs.

  3. Question 26

    Infrequently accessed backup data in S3 must be retained for seven years and retrieved only rarely. Which storage class minimizes cost while meeting retrieval needs?

    1. AEBS gp3 volumes attached to EC2 instances for the archive with daily snapshots taken
    2. BS3 Standard for every retained backup object across the full seven years of retention
    3. CS3 Glacier Flexible Retrieval or Glacier Deep Archive, with a lifecycle policy
    4. DA local RAID 6 array in the office server room with on-site copies of each backup
    Show answer and explanation
    Correct answer: C

    Glacier classes provide low-cost archival storage with retrieval options suited to rare access. S3 Standard is more costly for long retention, EBS is not archival storage, and local arrays do not meet the managed durability need.

  4. Question 27

    Which TWO are effective cost-saving practices for data stored in Amazon S3? (Choose two.)

    1. ALifecycle policies that transition older objects to lower-cost storage classes
    2. BProvisioning IOPS on the S3 bucket
    3. CKeeping every object version forever with no expiration
    4. DStoring rarely accessed data in S3 Standard
    5. EA lifecycle rule that removes incomplete multipart uploads
    Show answer and explanation
    Correct answer: A, E

    Lifecycle transitions and cleanup of incomplete uploads reduce storage costs. Unlimited versions, Standard for rarely accessed data and provisioned IOPS do not fit S3's model or cost goals.

  5. Question 28

    A company's analytics jobs run for a few hours each night on an EMR cluster. The jobs tolerate interruption. Which purchasing option reduces the compute cost the most?

    1. ADedicated hosts for every node, with host-level tenancy and per-socket licensing
    2. BOn-Demand instances for every node around the clock in each cluster
    3. CReserved instances for a three-year term regardless of actual usage in each Region
    4. DSpot instances for the task nodes, with On-Demand core nodes for stability
    Show answer and explanation
    Correct answer: D

    Spot task nodes reduce cost for interruptible work, and On-Demand core nodes keep the cluster stable. Always-on On-Demand, dedicated hosts and unused three-year reservations increase cost.

  6. Question 29

    A company stores log files in S3 and must keep them for 90 days for troubleshooting, then delete them. Which feature automates this?

    1. AStoring logs in memory with a rolling log buffer
    2. BAn S3 lifecycle rule that expires objects after 90 days
    3. CManually deleting files every quarter by an operator after audit
    4. DKeeping logs forever to be safe, with no expiry rule
    Show answer and explanation
    Correct answer: B

    Lifecycle expiration removes objects automatically after the retention period. Manual deletion, indefinite storage and in-memory storage do not automate the policy.

  7. Question 30

    An organization wants to reduce data transfer costs for workloads that send large amounts of data from EC2 to users on the internet. Which service helps reduce these costs?

    1. AUsing larger instance types so that data transfer charges per gigabyte fall across the whole fleet, with fewer hosts overall
    2. BPlacing users in one AZ so that internet egress is billed at the lower intra-AZ rate for the Region's workloads and origin servers
    3. CRoute 53 private hosted zones that resolve internal names inside the VPC with weighted records per tier
    4. DAmazon CloudFront, which serves cached content from edge locations with lower transfer rates for many workloads
    Show answer and explanation
    Correct answer: D

    CloudFront can reduce delivery costs for many internet-facing workloads by caching at the edge. Private DNS zones, AZ placement and instance size do not reduce internet transfer costs.

Want more SAA-C03 questions?

The full bank has 100 questions with answers and explanations, as a PDF, an online practice exam, or both.

Full question bank · 100 questions
$75$104

Both formats together. The best value.

Instant access after payment, with 90 days of updates. Checkout is on our store, AIExamCerts.