AWS
Solutions Architect Associate
SAA-C03 - Design Resilient, High-Performing, Secure & Cost-Optimized AWS
Architectures
Exam: SAA-C03
53 Questions & Verified Answers
AWS Solutions Architect Associate SAA-C03 - 53 Exam
Questions
Questions cover all four SAA-C03 domains: Design Resilient Architectures, Design High-Performing
Architectures, Design Secure Architectures, and Design Cost-Optimized Architectures.
Q01. What is the difference between High Availability and Fault Tolerance in AWS?
A. Identical
B. HA minimizes downtime by quickly recovering from failures (brief disruption acceptable);
Fault Tolerance means the system continues with NO interruption even when components fail
C. HA is cheaper than fault tolerance
D. HA only applies to EC2
Answer: B. HA = quick recovery with brief downtime; Fault Tolerance = zero interruption on
failure
HA example: Multi-AZ RDS - if primary fails, standby is promoted within ~2 minutes.
FT example: S3 - automatically replicates across 3+ AZs, no customer-visible disruption on AZ
failure.
RTO (Recovery Time Objective): max acceptable downtime. RPO (Recovery Point Objective): max
acceptable data loss.
Choosing between HA and FT: match to business requirements and budget - FT is more expensive.
Q02. A web application must survive an entire AZ failure with no manual intervention. What
architecture achieves this?
A. Deploy everything in one large EC2 instance
B. EC2 instances across at least 2 AZs behind an Application Load Balancer, with Auto Scaling
Groups spanning multiple AZs, and a Multi-AZ database
C. Use only multiple Regions
, D. Use Elastic IPs on each EC2 instance
Answer: B. EC2 across 2+ AZs behind ALB, with Auto Scaling spanning AZs, and Multi-AZ
database
ALB: distributes traffic to healthy instances across AZs automatically.
Auto Scaling Group: launches replacement instances in healthy AZs when one AZ fails.
Multi-AZ RDS: synchronous standby in a different AZ - auto-promotes in ~2 minutes on failure.
This is the standard 'AZ-resilient' architecture pattern - the foundation of most AWS HA designs.
Q03. What are the four AWS disaster recovery strategies in order from highest to lowest RTO?
A. Full, Partial, Minimal, None
B. Backup & Restore (highest RTO, cheapest), Pilot Light, Warm Standby, Multi-Site
Active/Active (lowest RTO, most expensive)
C. Hot, Warm, Cold, Frozen
D. Single-AZ, Multi-AZ, Multi-Region, Global
Answer: B. Backup & Restore → Pilot Light → Warm Standby → Multi-Site Active/Active
(cost and complexity increase)
Backup & Restore: data backed up to S3, restored from scratch on disaster. RTO: hours to days.
Pilot Light: core data infrastructure (DB) always on, compute off until needed. RTO: tens of minutes.
Warm Standby: fully functional scaled-down copy always running. RTO: minutes.
Multi-Site Active/Active: full production in 2+ Regions simultaneously. RTO: near zero. Most
expensive.
Q04. What is the purpose of Amazon SQS in a resilient architecture?
A. A chat message database
B. A managed message queue that decouples application tiers - absorbing traffic spikes,
preventing data loss when downstream systems are slow, and enabling asynchronous
processing
C. A real-time data streaming service
D. A microservices load balancer
Answer: B. Managed message queue - decouples tiers, absorbs spikes, prevents data loss,
enables async processing
Decoupling: web tier puts orders in SQS → processing tier reads at its own pace. Neither blocks the
other.
Buffering: during a traffic spike, SQS accumulates messages - no messages lost even if workers
are overwhelmed.
Standard queue: at-least-once delivery, maximum throughput, best-effort ordering.
FIFO queue: exactly-once processing, strict ordering - for financial transactions, inventory systems.
Q05. What is the difference between ALB, NLB, and Gateway Load Balancer?
A. Identical
B. ALB: Layer 7, HTTP/HTTPS content routing, host/path routing, WebSockets; NLB: Layer 4,
TCP/UDP, extreme performance (millions RPS), static IP; GWLB: for deploying third-party
network appliances (firewalls, IDS) inline
C. NLB is for internal; ALB for internet
, D. ALB supports TCP; NLB supports HTTP
Answer: B. ALB = Layer 7 HTTP/HTTPS routing; NLB = Layer 4 extreme performance + static
IP; GWLB = third-party appliances
ALB use cases: microservices with URL path routing, blue/green deployments, host-based routing.
NLB use cases: ultra-high throughput applications, static IP for firewall whitelisting, TCP/UDP non-
HTTP apps.
GWLB use cases: route all traffic through a firewall or IDS appliance for inspection before
forwarding to targets.
Q06. What is Amazon ElastiCache and when should you use Redis vs Memcached?
A. A CDN service
B. A managed in-memory caching service - Redis: rich data structures, persistence, pub/sub,
replication, Multi-AZ; Memcached: simpler, multi-threaded, horizontally scalable - use Redis for
most production use cases
C. A database service
D. A content storage service
Answer: B. Redis = rich data structures, persistence, replication, pub/sub - use for most
production; Memcached = simple multi-threaded for basic caching
ElastiCache Redis use cases: session management, leaderboards, pub/sub messaging, caching
with persistence.
ElastiCache Memcached: simple key-value caching when you don't need persistence or advanced
data types.
DAX (DynamoDB Accelerator): DynamoDB-specific in-memory cache - microsecond reads, drop-in
compatible.
Q07. What are S3 storage classes and how do you choose between them?
A. Only one storage class exists
B. Standard (frequent access, multi-AZ, most expensive), Standard-IA (infrequent, retrieval fee,
multi-AZ), One Zone-IA (infrequent, single AZ, 20% cheaper), Intelligent-Tiering (auto-moves
between tiers), Glacier Instant/Flexible/Deep Archive (archival, progressive latency and cost
reduction)
C. Hot, Warm, Cold
D. SSD, HDD, Tape
Answer: B. Standard → Standard-IA → One Zone-IA → Intelligent-Tiering → Glacier
(Instant/Flexible/Deep Archive) - trading access speed for cost
Standard: frequent access, no retrieval fee - web app assets, frequently downloaded files.
Standard-IA: less than once per month access, multi-AZ resilience - backups, disaster recovery.
One Zone-IA: reproduced data (thumbnails, processed data) - data loss acceptable if AZ fails.
Glacier Deep Archive: years-long retention, 12-48 hour retrieval - compliance archives, regulatory
records.
Q08. A company needs millions of writes per second globally with <10ms latency. Which
database?
A. Amazon RDS MySQL Multi-AZ
B. Amazon DynamoDB with Global Tables
Solutions Architect Associate
SAA-C03 - Design Resilient, High-Performing, Secure & Cost-Optimized AWS
Architectures
Exam: SAA-C03
53 Questions & Verified Answers
AWS Solutions Architect Associate SAA-C03 - 53 Exam
Questions
Questions cover all four SAA-C03 domains: Design Resilient Architectures, Design High-Performing
Architectures, Design Secure Architectures, and Design Cost-Optimized Architectures.
Q01. What is the difference between High Availability and Fault Tolerance in AWS?
A. Identical
B. HA minimizes downtime by quickly recovering from failures (brief disruption acceptable);
Fault Tolerance means the system continues with NO interruption even when components fail
C. HA is cheaper than fault tolerance
D. HA only applies to EC2
Answer: B. HA = quick recovery with brief downtime; Fault Tolerance = zero interruption on
failure
HA example: Multi-AZ RDS - if primary fails, standby is promoted within ~2 minutes.
FT example: S3 - automatically replicates across 3+ AZs, no customer-visible disruption on AZ
failure.
RTO (Recovery Time Objective): max acceptable downtime. RPO (Recovery Point Objective): max
acceptable data loss.
Choosing between HA and FT: match to business requirements and budget - FT is more expensive.
Q02. A web application must survive an entire AZ failure with no manual intervention. What
architecture achieves this?
A. Deploy everything in one large EC2 instance
B. EC2 instances across at least 2 AZs behind an Application Load Balancer, with Auto Scaling
Groups spanning multiple AZs, and a Multi-AZ database
C. Use only multiple Regions
, D. Use Elastic IPs on each EC2 instance
Answer: B. EC2 across 2+ AZs behind ALB, with Auto Scaling spanning AZs, and Multi-AZ
database
ALB: distributes traffic to healthy instances across AZs automatically.
Auto Scaling Group: launches replacement instances in healthy AZs when one AZ fails.
Multi-AZ RDS: synchronous standby in a different AZ - auto-promotes in ~2 minutes on failure.
This is the standard 'AZ-resilient' architecture pattern - the foundation of most AWS HA designs.
Q03. What are the four AWS disaster recovery strategies in order from highest to lowest RTO?
A. Full, Partial, Minimal, None
B. Backup & Restore (highest RTO, cheapest), Pilot Light, Warm Standby, Multi-Site
Active/Active (lowest RTO, most expensive)
C. Hot, Warm, Cold, Frozen
D. Single-AZ, Multi-AZ, Multi-Region, Global
Answer: B. Backup & Restore → Pilot Light → Warm Standby → Multi-Site Active/Active
(cost and complexity increase)
Backup & Restore: data backed up to S3, restored from scratch on disaster. RTO: hours to days.
Pilot Light: core data infrastructure (DB) always on, compute off until needed. RTO: tens of minutes.
Warm Standby: fully functional scaled-down copy always running. RTO: minutes.
Multi-Site Active/Active: full production in 2+ Regions simultaneously. RTO: near zero. Most
expensive.
Q04. What is the purpose of Amazon SQS in a resilient architecture?
A. A chat message database
B. A managed message queue that decouples application tiers - absorbing traffic spikes,
preventing data loss when downstream systems are slow, and enabling asynchronous
processing
C. A real-time data streaming service
D. A microservices load balancer
Answer: B. Managed message queue - decouples tiers, absorbs spikes, prevents data loss,
enables async processing
Decoupling: web tier puts orders in SQS → processing tier reads at its own pace. Neither blocks the
other.
Buffering: during a traffic spike, SQS accumulates messages - no messages lost even if workers
are overwhelmed.
Standard queue: at-least-once delivery, maximum throughput, best-effort ordering.
FIFO queue: exactly-once processing, strict ordering - for financial transactions, inventory systems.
Q05. What is the difference between ALB, NLB, and Gateway Load Balancer?
A. Identical
B. ALB: Layer 7, HTTP/HTTPS content routing, host/path routing, WebSockets; NLB: Layer 4,
TCP/UDP, extreme performance (millions RPS), static IP; GWLB: for deploying third-party
network appliances (firewalls, IDS) inline
C. NLB is for internal; ALB for internet
, D. ALB supports TCP; NLB supports HTTP
Answer: B. ALB = Layer 7 HTTP/HTTPS routing; NLB = Layer 4 extreme performance + static
IP; GWLB = third-party appliances
ALB use cases: microservices with URL path routing, blue/green deployments, host-based routing.
NLB use cases: ultra-high throughput applications, static IP for firewall whitelisting, TCP/UDP non-
HTTP apps.
GWLB use cases: route all traffic through a firewall or IDS appliance for inspection before
forwarding to targets.
Q06. What is Amazon ElastiCache and when should you use Redis vs Memcached?
A. A CDN service
B. A managed in-memory caching service - Redis: rich data structures, persistence, pub/sub,
replication, Multi-AZ; Memcached: simpler, multi-threaded, horizontally scalable - use Redis for
most production use cases
C. A database service
D. A content storage service
Answer: B. Redis = rich data structures, persistence, replication, pub/sub - use for most
production; Memcached = simple multi-threaded for basic caching
ElastiCache Redis use cases: session management, leaderboards, pub/sub messaging, caching
with persistence.
ElastiCache Memcached: simple key-value caching when you don't need persistence or advanced
data types.
DAX (DynamoDB Accelerator): DynamoDB-specific in-memory cache - microsecond reads, drop-in
compatible.
Q07. What are S3 storage classes and how do you choose between them?
A. Only one storage class exists
B. Standard (frequent access, multi-AZ, most expensive), Standard-IA (infrequent, retrieval fee,
multi-AZ), One Zone-IA (infrequent, single AZ, 20% cheaper), Intelligent-Tiering (auto-moves
between tiers), Glacier Instant/Flexible/Deep Archive (archival, progressive latency and cost
reduction)
C. Hot, Warm, Cold
D. SSD, HDD, Tape
Answer: B. Standard → Standard-IA → One Zone-IA → Intelligent-Tiering → Glacier
(Instant/Flexible/Deep Archive) - trading access speed for cost
Standard: frequent access, no retrieval fee - web app assets, frequently downloaded files.
Standard-IA: less than once per month access, multi-AZ resilience - backups, disaster recovery.
One Zone-IA: reproduced data (thumbnails, processed data) - data loss acceptable if AZ fails.
Glacier Deep Archive: years-long retention, 12-48 hour retrieval - compliance archives, regulatory
records.
Q08. A company needs millions of writes per second globally with <10ms latency. Which
database?
A. Amazon RDS MySQL Multi-AZ
B. Amazon DynamoDB with Global Tables
