Security and Governance
Scenario
Choose
Why
User needs temporary access to AWS resources
IAM role
Uses temporary credentials
Enforce permission limits across accounts
SCP
Applies at organization/account boundary
Govern many accounts quickly
AWS Control Tower
Landing zone and guardrails
Find API caller history
CloudTrail
Records API calls
Detect non-compliant resource config
Config
Tracks configuration and rules
Download AWS compliance reports
Artifact
Compliance reports and agreements
Encrypt data with managed keys
KMS
Key management service
Rotate database credentials
Secrets Manager
Secret storage and rotation
Detect compromised credentials or unusual activity
GuardDuty
Threat detection
Scan workloads for vulnerabilities
Inspector
Vulnerability management
Find PII in S3
Macie
Sensitive data discovery
Protect web app from SQL injection/XSS
WAF
Layer 7 web request filtering
Protect against DDoS attacks
Shield
DDoS protection
Compute
Scenario
Choose
Why
Need virtual machine with OS control
EC2
Full compute control
Need event-driven serverless execution
Lambda
No server management
Need simple app deployment
Elastic Beanstalk
Platform handles deployment
Need simple VPS
Lightsail
Simplified compute bundle
Need containers without managing servers
Fargate
Serverless container runtime
Need Kubernetes
EKS
Managed Kubernetes
Need AWS-native container orchestration
ECS
Integrated AWS container service
Storage
Scenario
Choose
Why
Store files, images, backups, logs
S3
Durable object storage
Persistent disk for EC2
EBS
Block storage
Shared Linux file system
EFS
Multi-instance file access
Windows file shares
FSx for Windows File Server
Managed SMB file system
Long-term archive
S3 Glacier
Low-cost archival
Hybrid on-premises storage
Storage Gateway
Connects on-premises apps to AWS storage
Move large data sets
DataSync / Snowball
Online/offline migration options
Databases
Scenario
Choose
Why
Managed MySQL/PostgreSQL/SQL Server
RDS
Managed relational database
High-performance cloud-native relational
Aurora
AWS-optimized relational database
Serverless NoSQL key-value
DynamoDB
Low-latency NoSQL
Analytics warehouse
Redshift
Columnar analytics
Cache frequent reads
ElastiCache
In-memory performance
Graph relationships
Neptune
Graph database
MongoDB-compatible document workload
DocumentDB
Managed document database
Networking
Scenario
Choose
Why
Isolate cloud network
VPC
Private network boundary
Control instance inbound/outbound traffic
Security Group
Stateful instance firewall
Control subnet traffic
Network ACL
Stateless subnet firewall
DNS and domain routing
Route 53
Managed DNS
CDN and edge caching
CloudFront
Low-latency content delivery
Dedicated private connection
Direct Connect
Private network link
Encrypted connection over internet
VPN
Encrypted tunnel
Hub for multiple VPCs
Transit Gateway
Central connectivity
6. Architecture Patterns
Pattern 1 — Highly available web application
Scenario: A business wants a web application to remain available if one data center fails.
Recommended solution:
Deploy across multiple Availability Zones.
Use Elastic Load Balancing.
Use Auto Scaling.
Store static assets in S3 and serve with CloudFront.
Use RDS Multi-AZ or DynamoDB depending on data model.
Why alternatives are wrong:
A single EC2 instance is not highly available.
Backups alone do not provide active availability.
One Availability Zone is a single point of failure.
Pattern 2 — Static website hosting
Scenario: Host static HTML, CSS, images, and JavaScript with low cost.
Recommended solution:
Amazon S3 static website hosting
CloudFront for global performance
Route 53 for DNS
ACM for TLS certificates when using CloudFront
Why alternatives are wrong:
EC2 works but adds unnecessary server management.
RDS is not for static site hosting.
EBS cannot serve objects directly as a public static website service.
Pattern 3 — Serverless event processing
Scenario: Run code when an object is uploaded.
Recommended solution:
S3 event notification
Lambda function
CloudWatch Logs for logging
Why alternatives are wrong:
EC2 requires server management.
RDS does not execute event-driven code.
CloudTrail logs API activity but does not process application logic.
Pattern 4 — Decoupled application
Scenario: One application component sends tasks to another component without tight coupling.
Recommended solution:
SQS for queues
SNS for pub/sub fanout
EventBridge for event routing
Lambda/ECS/EC2 workers for processing
Why alternatives are wrong:
Direct synchronous calls increase coupling.
CloudWatch is for monitoring, not message queuing.
Route 53 is DNS, not messaging.
Pattern 5 — Secure multi-account environment
Scenario: Company wants centralized account management and guardrails.
Recommended solution:
AWS Organizations
Organizational Units
SCPs
AWS Control Tower for landing zone setup
IAM Identity Center for workforce access
Why alternatives are wrong:
IAM groups only manage users inside one account.
Security groups control network traffic, not account governance.
CloudTrail audits activity but does not enforce account-level guardrails.
Pattern 6 — Compliance evidence
Scenario: Auditor asks for AWS SOC/ISO compliance reports.
Recommended solution:
Why alternatives are wrong:
CloudTrail shows API history, not AWS compliance reports.
Config tracks resource compliance, not AWS audit report downloads.
IAM manages permissions, not compliance documentation.
Pattern 7 — Cost control
Scenario: A team wants to know when monthly cost exceeds a threshold.
Recommended solution:
Why alternatives are wrong:
Pricing Calculator estimates planned workloads before deployment.
Cost Explorer analyzes cost trends but is not primarily the budget alert tool.
CloudWatch monitors service metrics, not billing thresholds in the same way.
Pattern 8 — Hybrid connectivity
Scenario: Company needs a dedicated low-latency private connection to AWS.
Recommended solution:
Why alternatives are wrong:
Site-to-Site VPN uses the internet.
CloudFront is a CDN, not private hybrid connectivity.
Transit Gateway connects networks but does not itself create the dedicated physical link.
7. Exam Traps
Misleading wording patterns
Wording
Think
“Who accessed this resource?”
CloudTrail
“CPU exceeded 80%”
CloudWatch
“Resource changed from compliant to non-compliant”
Config
“Download compliance reports”
Artifact
“Protect against SQL injection”
WAF
“Protect against DDoS”
Shield
“Find sensitive data in S3”
Macie
“Scan EC2 for vulnerabilities”
Inspector
“Detect suspicious account behavior”
GuardDuty
“Estimate cost before migration”
Pricing Calculator
“Analyze last month’s cost”
Cost Explorer
“Alert when spending exceeds $X”
Budgets
“Static website objects”
S3
“Persistent EC2 disk”
EBS
“Shared file system”
EFS
“Archive for years”
S3 Glacier
“DNS”
Route 53
“Cache content near users”
CloudFront
Wrong-but-plausible answer patterns
Choosing CloudWatch for audit history : wrong when the question asks who made an API call. Use CloudTrail.
Choosing CloudTrail for performance metrics : wrong when the question asks CPU, memory, alarms, logs, or dashboards. Use CloudWatch.
Choosing IAM policy when the scenario needs organization-wide guardrails : use SCPs.
Choosing EC2 when the scenario emphasizes no server management : use Lambda or Fargate.
Choosing RDS for key-value scale : use DynamoDB.
Choosing DynamoDB for SQL joins and relational schema : use RDS/Aurora.
Choosing S3 for EC2 boot disks : use EBS.
Choosing EBS for shared file access : use EFS.
Choosing VPN when the question says dedicated private connection : use Direct Connect.
Choosing Cost Explorer for alerts : use AWS Budgets.
Elimination strategy
When stuck, classify the requirement:
Security?
Identity → IAM
Encryption keys → KMS
Secrets → Secrets Manager
Threat detection → GuardDuty
Web filtering → WAF
DDoS → Shield
Monitoring/governance?
Metrics/alarms/logs → CloudWatch
API calls → CloudTrail
Resource config/compliance → Config
Best practice checks → Trusted Advisor
Storage?
Object → S3
Block → EBS
Shared file → EFS/FSx
Archive → Glacier
Compute?
Server control → EC2
Serverless function → Lambda
Containers → ECS/EKS/Fargate
Simple app platform → Elastic Beanstalk
Cost?
Estimate → Pricing Calculator
Analyze → Cost Explorer
Alert → Budgets
Recommendations → Trusted Advisor/Compute Optimizer
8. Quick Memory Rules
Rules of thumb
S3 stores objects, EBS stores blocks, EFS stores files.
CloudTrail tracks API calls; CloudWatch watches performance; Config checks configuration.
WAF filters web requests; Shield handles DDoS.
GuardDuty detects threats; Inspector finds vulnerabilities; Macie finds sensitive data.
KMS manages keys; Secrets Manager manages secrets.
RDS is relational; DynamoDB is NoSQL; Redshift is analytics; ElastiCache is cache.
Route 53 routes DNS; CloudFront delivers cached content.
Direct Connect is dedicated; VPN is encrypted internet.
SQS queues; SNS publishes; EventBridge routes events; Step Functions orchestrates workflows.
Pricing Calculator predicts; Cost Explorer analyzes; Budgets alerts.
Fast service mapping
If you see...
Think...
“least privilege”
IAM policy
“temporary credentials”
IAM role
“centralized account management”
AWS Organizations
“prevent actions across accounts”
SCP
“landing zone”
Control Tower
“audit API calls”
CloudTrail
“metric alarm”
CloudWatch
“configuration drift”
Config
“compliance reports”
Artifact
“PII in S3”
Macie
“SQL injection”
WAF
“DDoS”
Shield
“serverless function”
Lambda
“managed Kubernetes”
EKS
“serverless containers”
Fargate
“object lifecycle”
S3 lifecycle policies
“data warehouse”
Redshift
“DNS failover”
Route 53
“low latency global content”
CloudFront
“large offline transfer”
Snowball
“hybrid storage”
Storage Gateway
9. Final Revision Notes
Highest-yield review points
Understand the shared responsibility model deeply.
Memorize CloudTrail vs CloudWatch vs Config .
Memorize S3 vs EBS vs EFS vs Glacier .
Know when to use EC2, Lambda, ECS, EKS, Fargate .
Know database choices: RDS, Aurora, DynamoDB, Redshift, ElastiCache .
Know security services: IAM, KMS, Secrets Manager, WAF, Shield, GuardDuty, Inspector, Macie, Artifact .
Know cost tools: Pricing Calculator, Cost Explorer, Budgets, Trusted Advisor .
Know support plans and what TAM means.
Know global infrastructure: Regions, AZs, edge locations .
Last-day revision list
Read all service selection tables.
Practice 50 mixed questions.
For every wrong answer, ask: “What clue did I miss?”
Review cost tools and support plans.
Review security service differences.
Review storage/database differences.
Sleep; do not overload with deep professional-level details.
10. Exam-Day Checklist
Must-know topics
Official domains and their relative weights
AWS shared responsibility model
IAM users, groups, roles, policies, MFA
Organizations, OUs, SCPs, Control Tower
CloudTrail, CloudWatch, Config, Trusted Advisor
WAF, Shield, GuardDuty, Inspector, Macie
KMS, Secrets Manager, ACM, Artifact
EC2, Lambda, ECS, EKS, Fargate, Elastic Beanstalk
S3, EBS, EFS, FSx, Glacier
RDS, Aurora, DynamoDB, Redshift, ElastiCache
VPC, security groups, NACLs, Route 53, CloudFront
VPN, Direct Connect, Transit Gateway
SQS, SNS, EventBridge, Step Functions
Athena, Glue, Kinesis, EMR, QuickSight
Pricing Calculator, Cost Explorer, Budgets
On-Demand, Reserved, Savings Plans, Spot
AWS Support plans and AWS Marketplace
Final confidence checklist
Before taking the exam, you should be able to answer these quickly:
What does AWS manage vs what does the customer manage?
Which service audits API calls?
Which service monitors CPU and logs?
Which service checks configuration compliance?
Which storage service is object, block, file, or archive?
Which database fits relational, NoSQL, warehouse, or cache?
Which tool estimates cost before deployment?
Which tool analyzes actual spend?
Which tool sends budget alerts?
Which support plan includes a TAM?
Which service is for WAF filtering vs DDoS protection?
Which network service is DNS vs CDN vs private connectivity?
Appendix A — High-Frequency Services Detected in the Question Bank
The source question bank most frequently referenced these topics, so they deserve extra review:
Rank
Service / Concept
Approximate Mentions
1
IAM
261
2
EC2
222
3
AWS Artifact
186
4
Route 53
183
5
Amazon EBS
174
6
Amazon S3
151
7
CloudFront
148
8
Polly
132
9
VPC
117
10
AWS WAF
109
11
Lambda
97
12
AWS Shield
91
13
SNS
91
14
AWS Budgets
91
15
SQS
86
16
AWS Config
84
17
S3 Glacier
81
18
Amazon EFS
81
19
AWS Organizations
75
20
Amazon RDS
72
21
Snowball
72
22
Amazon Macie
71
23
Well-Architected
68
24
Direct Connect
64
25
DataSync
63
26
AWS KMS
62
27
Redshift
62
28
Cost Explorer
62
29
ElastiCache
60
30
AWS Pricing Calculator
59
Appendix B — Mini Mock Reasoning Examples
Example 1
A company wants to know which user deleted an S3 bucket.
Correct thinking: this is API activity/audit history.
Best answer: AWS CloudTrail
Why not CloudWatch: CloudWatch monitors metrics/logs/alarms, not primarily API caller history.
Example 2
A company needs to store old compliance records for seven years at the lowest cost and does not need immediate access.
Correct thinking: archive storage.
Best answer: S3 Glacier
Why not EBS: EBS is block storage for EC2, not low-cost long-term archive.
Example 3
A team wants an alert when monthly AWS spend exceeds $5,000.
Correct thinking: budget threshold alert.
Best answer: AWS Budgets
Why not Cost Explorer: Cost Explorer helps analyze cost trends, but Budgets is the alerting tool.
Example 4
A company wants to protect an application from SQL injection.
Correct thinking: layer 7 web request filtering.
Best answer: AWS WAF
Why not Shield: Shield protects from DDoS, not application-layer rule filtering in the same way.
Example 5
A startup wants to run code only when an image is uploaded to S3 and does not want to manage servers.
Correct thinking: serverless event-driven compute.
Best answer: AWS Lambda
Why not EC2: EC2 requires server provisioning and management.
Appendix C — Fast Wrong Answer Diagnostics
When an answer seems plausible, reject it if:
It solves a different layer of the stack.
It is operational when the requirement is governance.
It is governance when the requirement is monitoring.
It is monitoring when the requirement is auditing.
It is relational when the workload is key-value.
It is block storage when the need is object storage.
It is a future estimate tool when the need is actual cost analysis.
It is a cost analysis tool when the need is budget alerting.
It adds unnecessary complexity for a foundational cloud scenario.
End of Course