AWS Certified Solutions Architect – Professional (SAP-C03) study guide
What the SAP-C03 exam tests, domain by domain, and how to tell the right answer from the nearly right one. Each topic links to fuller study notes and free practice questions.
Last checked against the official exam guide and AWS documentation · How we write our content
About the SAP-C03 exam
The AWS Certified Solutions Architect – Professional exam checks whether you can design complex solutions on AWS for whole organizations: many accounts, hybrid networks, strict security and compliance rules, tight budgets and demanding recovery targets. It is aimed at people with 2 or more years of hands-on experience designing and deploying cloud architecture on AWS.
SAP-C03 replaces SAP-C02. The last day to take SAP-C02 is November 16, 2026, and SAP-C03 is available from November 17, 2026. The new version adds generative AI and agents, responsible AI controls, post-quantum cryptography and more operational excellence work; see what changed from SAP-C02 to SAP-C03.
| Questions | 75: 65 scored and 10 unscored, which aren’t marked |
| Time | 180 minutes |
| Question types | Multiple choice (one answer from four) and multiple response (two or more from five or more) |
| Score | Scaled from 100 to 1,000; 720 to pass (SAP-C02 needed 750) |
| Scoring model | Compensatory: you need to pass overall, not in each domain |
| Guessing | An unanswered question counts as wrong, and there is no penalty for a wrong answer, so answer everything |
Check the official certification page for the current price, languages and booking options before you book.
What the exam really tests
Professional questions are long. They describe an organization, its current setup and a list of requirements, then offer four answers that all work to some degree. The winning answer meets every requirement, including the qualifier, with managed services and the least custom code. A few themes run through the whole exam:
- Design for the cloud. Event-driven, serverless and container architectures, stateless services and the right database for each access pattern.
- Govern many accounts. Organizations, Control Tower, SCPs and RCPs, delegated administrators, IAM Identity Center and data perimeters.
- Build AI in safely. Amazon Bedrock, agents and RAG, Guardrails, and logging of model use.
- Spend wisely. Savings Plans, Graviton, storage tiering, data transfer paths and cost allocation.
- Recover on target. DR strategies matched to RTO and RPO, multi-Region failover, backups protected from ransomware, and migration with AWS Transform MGN and DMS.
- Automate operations. Infrastructure as code, safe deployments, observability with OpenTelemetry and automated remediation.
How to read a question
- Read the last sentence first to find the qualifier: most cost-effective, least operational overhead, highest availability, most secure, fewest changes.
- List the hard constraints: accounts, Regions, RTO and RPO, compliance rules, existing tools, budget, timeline.
- Eliminate answers that use retired or closed services, break a constraint, or add servers and scripts where a managed feature exists.
- Between the last two, pick the one that meets the qualifier best, not the one with the most features.
Pace yourself: 180 minutes for 75 questions is about 2 minutes 24 seconds each, and many scenarios run to a full paragraph. Flag and move on.
How the domains add up
Architecture Design and Implementation Built for the Cloud and Security, Compliance, and Governance are the largest domains at 25% each. Operational Excellence and Automation follows at 20%, then Cost-Optimized Architecture Design and Resilience, Migration, and Business Continuity at 15% each. The domains overlap: a migration question often turns on networking, and a cost question on storage design.
Domain 1: Architecture Design and Implementation Built for the Cloud 25%
This is the largest domain at 25%, and it covers designing workloads built for the cloud: event-driven messaging with EventBridge, SQS, SNS and Kinesis, Step Functions and durable Lambda workflows, containers on ECS and EKS with VPC Lattice and Service Connect, ingress and edge patterns, multi-tenant SaaS isolation, generative AI with Amazon Bedrock, agents and RAG, SageMaker AI inference, hybrid connectivity, DNS, data sync and compute, multi-VPC networks, IPAM and IPv6, network inspection, global traffic routing, and the data layer from lakehouses and streaming analytics to database selection, storage and caching.
Event-driven messaging and cross-account event routing
Event-driven designs decouple producers from consumers with EventBridge, SQS, SNS and Kinesis. Know delivery guarantees, ordering and idempotency for each service, and how to route events across accounts and Regions with event buses and direct cross-account targets.
Study notes and 1 free question on Event-driven messaging and cross-account event routingServerless data pipelines and durable workflows (Step Functions)
Serverless pipelines chain Kinesis, Lambda, Glue and Athena to process data with no servers, and durable workflows coordinate many steps with retries and waits. Know Step Functions Standard versus Express, Lambda durable functions, and when each fits.
Study notes and 1 free question on Serverless data pipelines and durable workflows (Step Functions)Containers, service mesh, images and scaling (ECS, EKS, VPC Lattice, ECR)
Microservices run on ECS or EKS, on EC2 or Fargate, and talk to each other through ECS Service Connect or VPC Lattice. Know when to pick each orchestrator, how ECR scanning and lifecycle policies keep images safe and tidy, and how containers and Lambda scale.
Study notes and 1 free question on Containers, service mesh, images and scaling (ECS, EKS, VPC Lattice, ECR)Ingress, edge functions and real-time APIs (ALB, NLB, API Gateway, CloudFront)
Traffic enters through ALB, NLB, API Gateway or CloudFront, and the right choice depends on protocol, features and latency. Know the AWS Load Balancer Controller for EKS, when to use CloudFront Functions versus Lambda@Edge, and how API Gateway WebSocket APIs power real-time apps.
Study notes and 1 free question on Ingress, edge functions and real-time APIs (ALB, NLB, API Gateway, CloudFront)Stateless and multi-tenant SaaS architectures
Cloud apps scale best when compute is stateless and state lives in DynamoDB, S3 or ElastiCache. For SaaS, know the silo, pool and bridge isolation models, how Cognito carries tenant context, and how row-level security, tenant-scoped IAM and per-tenant KMS keys enforce isolation.
Study notes and 1 free question on Stateless and multi-tenant SaaS architecturesGenerative AI and ML inference (Bedrock, SageMaker AI endpoints)
Amazon Bedrock gives serverless API access to foundation models, and SageMaker AI endpoints host your own models. Know Bedrock's multi-account and private access patterns, inference profiles and service tiers, and how to pick real-time, serverless, asynchronous or batch inference on SageMaker AI.
Study notes and 1 free question on Generative AI and ML inference (Bedrock, SageMaker AI endpoints)AI agents and RAG (AgentCore, Knowledge Bases)
Retrieval-augmented generation grounds model answers in your own data, and agents let models call tools to take actions. Know Bedrock Knowledge Bases ingestion, chunking, vector store and retrieval choices, and the Amazon Bedrock AgentCore services for running, connecting and governing agents.
Study notes and 2 free questions on AI agents and RAG (AgentCore, Knowledge Bases)Hybrid connectivity and DNS (Direct Connect, VPN, Route 53 Resolver)
Hybrid networks join data centers to AWS with Direct Connect and Site-to-Site VPN, usually through a Transit Gateway or Cloud WAN, and need DNS that resolves names both ways. Know the Direct Connect resiliency models, VPN backup and failover, and Route 53 Resolver endpoints, rules and split-horizon DNS.
Study notes and 1 free question on Hybrid connectivity and DNS (Direct Connect, VPN, Route 53 Resolver)Hybrid data sync and legacy integration (DataSync, Storage Gateway, Amazon MQ)
Hybrid data questions ask how to move or share files between on premises and AWS, and how to modernize legacy integrations. Know DataSync, Storage Gateway and Transfer Family, and Amazon MQ, API Gateway and Lambda for legacy messaging and data transformations.
Study notes and 1 free question on Hybrid data sync and legacy integration (DataSync, Storage Gateway, Amazon MQ)Hybrid and edge compute (Outposts, Local Zones, EKS and ECS Anywhere)
Some workloads must run on premises or close to users because of latency, data residency or local processing. Know when to use AWS Outposts, Local Zones, ECS Anywhere, EKS Anywhere and EKS Hybrid Nodes, and what each still depends on in the parent Region.
Study notes and 1 free question on Hybrid and edge compute (Outposts, Local Zones, EKS and ECS Anywhere)Multi-VPC and multi-account connectivity (Transit Gateway, PrivateLink, Cloud WAN)
Large organizations connect hundreds of VPCs across accounts and Regions. Know when VPC peering, Transit Gateway, AWS Cloud WAN, PrivateLink or VPC Lattice fits, how route table segmentation works, and how to share network resources with AWS RAM.
Study notes and 1 free question on Multi-VPC and multi-account connectivity (Transit Gateway, PrivateLink, Cloud WAN)VPC design, IPAM and IPv6 dual-stack
VPC design questions ask how to plan and govern IP space across an organization and how to move to IPv6. Know subnet tiers and segmentation, Amazon VPC IPAM pools and compliance, and dual-stack and IPv6-only patterns with egress-only internet gateways and DNS64 with NAT64.
Study notes and 1 free question on VPC design, IPAM and IPv6 dual-stackNetwork inspection and filtering (Network Firewall, WAF, GWLB)
Network security layers packet filtering, stateful inspection and application-layer protection. Know where AWS Network Firewall, Gateway Load Balancer with partner appliances, AWS WAF, Shield, security groups, network ACLs and VPC Flow Logs each fit, and how centralized inspection architectures route traffic.
Study notes and 1 free question on Network inspection and filtering (Network Firewall, WAF, GWLB)Global traffic routing (Route 53 policies, Global Accelerator)
Multi-Region applications need users sent to the right Region and moved away from a failing one. Know every Route 53 routing policy and how health checks drive failover, and when AWS Global Accelerator's anycast static IPs and fast failover beat DNS.
Study notes and 1 free question on Global traffic routing (Route 53 policies, Global Accelerator)Data lakes, lakehouses and data sharing (Lake Formation, Iceberg, Clean Rooms)
Data lakes keep raw and curated data in S3 with a shared catalog, and lakehouses add table formats such as Apache Iceberg. Know the Glue Data Catalog, Lake Formation permissions and cross-account sharing, Redshift Spectrum and data sharing, S3 Tables, and AWS Clean Rooms for collaboration without sharing raw data.
Study notes and 1 free question on Data lakes, lakehouses and data sharing (Lake Formation, Iceberg, Clean Rooms)Real-time and batch analytics (Kinesis, MSK, EMR, Redshift)
Analytics questions pair a data velocity with the right engine: Kinesis or MSK for streams, Managed Service for Apache Flink for stream processing, EMR for big data jobs, Redshift for warehousing and AppFlow for SaaS data. Know the trade-offs and the serverless options of each.
Study notes and 1 free question on Real-time and batch analytics (Kinesis, MSK, EMR, Redshift)Choosing and scaling databases (Aurora, DynamoDB, global databases)
Database questions start from access patterns, consistency needs and scale, then ask how to keep the database available across AZs and Regions. Know when to pick RDS, Aurora, DynamoDB, Redshift or a purpose-built engine, and how Multi-AZ, read replicas, Aurora Global Database and DynamoDB global tables differ.
Study notes and 1 free question on Choosing and scaling databases (Aurora, DynamoDB, global databases)Storage performance and caching (EBS, EFS, FSx, ElastiCache, DAX)
Storage questions match access pattern and performance to S3, EBS, EFS or an FSx file system, and caching questions ask where to cut latency and database load. Know EBS volume types, EFS throughput modes, the four FSx options, ElastiCache strategies and DynamoDB Accelerator.
Study notes and 1 free question on Storage performance and caching (EBS, EFS, FSx, ElastiCache, DAX)
Domain 2: Security, Compliance, and Governance 25%
This domain is a quarter of the exam and covers security for whole organizations: workforce and application identity, least privilege and data perimeters, encryption with KMS, ACM, Private CA and CloudHSM (including post-quantum options), landing zones with Control Tower, SCPs and RCPs, centralized logging and threat detection, resource sharing with RAM, continuous compliance and patching, data residency and audit evidence, and new controls for generative AI with Bedrock Guardrails, AgentCore Identity and human approval steps.
Workforce federation (IAM Identity Center, SAML, SCIM)
Workforce access to many AWS accounts should come from your existing identity provider through IAM Identity Center, with SAML 2.0 or OIDC for sign-in and SCIM for provisioning. Know permission sets, attribute-based access, delegated administration and when direct IAM federation still fits.
Study notes and 1 free question on Workforce federation (IAM Identity Center, SAML, SCIM)Application authentication and authorization (Cognito, Verified Permissions, EKS Pod Identity)
Application sign-in and authorization use Amazon Cognito user pools for users and tokens, identity pools for temporary AWS credentials, and Amazon Verified Permissions for fine-grained Cedar policies. Workloads on EKS get AWS credentials through EKS Pod Identity.
Study notes and 1 free question on Application authentication and authorization (Cognito, Verified Permissions, EKS Pod Identity)Least privilege, cross-account roles, ABAC and IAM Roles Anywhere
Least privilege across accounts rests on IAM roles and temporary credentials: trust policies, external IDs, permissions boundaries, session policies and ABAC with tags. Workloads outside AWS get the same short-lived credentials through IAM Roles Anywhere.
Study notes and 2 free questions on Least privilege, cross-account roles, ABAC and IAM Roles AnywhereData perimeters and external access (Access Analyzer, endpoint policies, Verified Access)
A data perimeter allows only trusted identities, on trusted resources, from expected networks. Know the organization condition keys, VPC endpoint policies, RCPs and SCPs that build it, IAM Access Analyzer for finding external access, and AWS Verified Access for VPN-less application access.
Study notes and 2 free questions on Data perimeters and external access (Access Analyzer, endpoint policies, Verified Access)Encryption at rest, key policies and post-quantum cryptography (AWS KMS)
Encryption at rest on AWS centers on AWS KMS: customer managed keys, key policies and grants, automatic rotation, multi-Region keys, custom and external key stores, and per-tenant keys. SAP-C03 adds post-quantum cryptography: ML-DSA signing keys, ML-KEM keys and hybrid post-quantum TLS.
Study notes and 2 free questions on Encryption at rest, key policies and post-quantum cryptography (AWS KMS)Certificates, private CA and CloudHSM
Encryption in transit relies on AWS Certificate Manager for public certificates, AWS Private CA for internal ones, and TLS 1.3 security policies on load balancers and CloudFront. AWS CloudHSM gives single-tenant, FIPS-validated HSMs when a regulation demands keys you alone control.
Study notes and 1 free question on Certificates, private CA and CloudHSMLanding zones, OUs, SCPs and RCPs
A landing zone is the multi-account foundation: AWS Organizations with an OU structure, AWS Control Tower for guardrails and baselines, Account Factory for automated provisioning, and SCPs and RCPs as the hard limits no local admin can bypass.
Study notes and 2 free questions on Landing zones, OUs, SCPs and RCPsCentralized security logging (organization trails, Security Lake, GuardDuty)
Security data from every account should land in central, protected accounts: an organization CloudTrail trail, Config aggregators, GuardDuty and Security Hub with delegated administrators, Amazon Security Lake in OCSF format, and Bedrock model invocation logs for AI workloads.
Study notes and 2 free questions on Centralized security logging (organization trails, Security Lake, GuardDuty)Cross-account resource sharing (AWS RAM)
AWS Resource Access Manager shares resources such as subnets, transit gateways, Route 53 Resolver rules, prefix lists, IPAM pools and Private CAs across accounts. Know when to share through RAM, when to use a resource policy, and how VPC sharing differs from peering.
Study notes and 1 free question on Cross-account resource sharing (AWS RAM)Compliance monitoring, vulnerability scanning and patching
Continuous compliance means finding drift and fixing it automatically: AWS Config rules and conformance packs with remediation, Security Hub CSPM standards, Amazon Inspector for vulnerabilities in EC2, containers and Lambda, and Patch Manager patch policies across the organization.
Study notes and 1 free question on Compliance monitoring, vulnerability scanning and patchingData protection, sovereignty and residency controls
Data protection and residency questions pair encryption, access control and audit logging with controls that keep data in approved Regions: Region-deny SCPs and controls, restricted replication, geographic Bedrock inference profiles, and key stores that keep key material under your control.
Study notes and 1 free question on Data protection, sovereignty and residency controlsAudit trails, compliance evidence and threat modeling (AWS Artifact)
Regulated workloads need tamper-resistant audit trails, compliance evidence an auditor will accept, and a documented risk assessment. Know CloudTrail and Bedrock invocation logs with Object Lock, AWS Artifact reports and agreements, Config and Security Hub CSPM as continuous evidence, and threat modeling.
Study notes and 1 free question on Audit trails, compliance evidence and threat modeling (AWS Artifact)Responsible AI controls (Bedrock Guardrails, AgentCore Identity, human approval)
Responsible AI on SAP-C03 means filtering what generative AI takes in and puts out with Amazon Bedrock Guardrails, controlling what agents can reach with Bedrock AgentCore Identity and IAM, and keeping a human in the loop for risky actions with Step Functions callbacks.
Study notes and 2 free questions on Responsible AI controls (Bedrock Guardrails, AgentCore Identity, human approval)
Domain 3: Cost-Optimized Architecture Design 15%
This domain is new as a stand-alone 15% domain in SAP-C03. It covers choosing purchase options (Savings Plans, Reserved Instances, Spot and capacity reservations), cost-optimized compute with Graviton, rightsizing and metric-based scaling, generative AI costs (Bedrock service tiers, batch inference, prompt caching and cross-Region inference), data transfer and network charges (NAT gateways, VPC endpoints, Transit Gateway, AZ-aware design), CloudFront and replication costs, tagging, cost categories and chargeback, cost reporting with CUR 2.0, Budgets and Cost Anomaly Detection, storage tiering and EBS volume types, database capacity modes and Database Savings Plans, and cost-optimized backup and observability retention.
Purchase options (Savings Plans, Reserved Instances, Spot)
Purchase-option questions match a usage pattern to the cheapest fitting commitment: On-Demand, Savings Plans, Reserved Instances, Spot or Capacity Reservations. Know which plan covers which services, when you need reserved capacity rather than a discount, and how to layer options.
Study notes and 1 free question on Purchase options (Savings Plans, Reserved Instances, Spot)Cost-optimized compute and rightsizing (Graviton, Compute Optimizer)
Cost-optimized compute means choosing the right processor, size and execution model, then scaling on real metrics. Know Graviton, Compute Optimizer and Cost Optimization Hub, Step Functions Standard versus Express, Lambda memory tuning, scheduled and target tracking scaling, and Fargate Spot.
Study notes and 1 free question on Cost-optimized compute and rightsizing (Graviton, Compute Optimizer)AI workload costs (Bedrock tiers, batch inference, caching, cross-Region inference)
Generative AI bills grow with tokens, so SAP-C03 asks how to cut cost per request within latency and residency rules. Know the Bedrock service tiers, batch inference, prompt caching, model choice, cross-Region inference profiles and SageMaker AI endpoint options.
Study notes and 2 free questions on AI workload costs (Bedrock tiers, batch inference, caching, cross-Region inference)Data transfer and network costs (NAT, endpoints, Transit Gateway, AZ-aware design)
Network cost questions hinge on which hop is billed: NAT gateway processing, cross-AZ and inter-Region transfer, Transit Gateway data processing and PrivateLink endpoint hours. Know when gateway endpoints, VPC peering, AZ-aware routing and keeping traffic in one Region remove the charge.
Study notes and 1 free question on Data transfer and network costs (NAT, endpoints, Transit Gateway, AZ-aware design)Content delivery and replication costs (CloudFront, Origin Shield, CRR)
CloudFront cuts latency and cost when it serves from cache instead of the origin. Know cache policies, Origin Shield and origin failover, and how to replicate data across Regions with S3 Cross-Region Replication and DynamoDB global tables without paying for copies nobody reads.
Study notes and 1 free question on Content delivery and replication costs (CloudFront, Origin Shield, CRR)Tagging, cost categories and chargeback
Chargeback needs consistent tags, enforced by policy, and a way to split shared costs. Know cost allocation tag activation, tag policies, SCP and Config rule enforcement, cost categories with split charges, split cost allocation data for containers, and Bedrock spend attribution.
Study notes and 1 free question on Tagging, cost categories and chargebackCost reporting, budgets and anomaly detection
Cost visibility starts with detailed billing data and ends with alerts that reach the right owner fast. Know Cost Explorer, CUR 2.0 through Data Exports with Athena and QuickSight, AWS Budgets with budget actions, and Cost Anomaly Detection monitors and alert subscriptions.
Study notes and 1 free question on Cost reporting, budgets and anomaly detectionStorage tiering and volume costs (S3 classes, lifecycle, gp3)
Storage cost questions match data access patterns to S3 storage classes, lifecycle rules and EFS tiers, and match block workloads to the right EBS volume type. Know minimum storage durations, retrieval times, when Intelligent-Tiering beats lifecycle rules, and why gp3 usually replaces gp2 and many io1 or io2 volumes.
Study notes and 1 free question on Storage tiering and volume costs (S3 classes, lifecycle, gp3)Database capacity modes and reserved capacity
Database cost questions match a traffic shape to a capacity mode and a commitment. Know DynamoDB on-demand versus provisioned with auto scaling and reserved capacity, RDS reserved instances, Aurora Serverless v2, Aurora I/O-Optimized versus Standard, and the new Database Savings Plans.
Study notes and 1 free question on Database capacity modes and reserved capacityBackup retention and observability costs
Backups and telemetry grow forever unless you set retention. Know AWS Backup lifecycle to cold storage, cross-Region copy costs, Data Lifecycle Manager and EBS Snapshots Archive, plus the CloudWatch levers: log retention, the Infrequent Access log class, filtering, metric aggregation and archiving logs to S3.
Study notes and 1 free question on Backup retention and observability costs
Domain 4: Resilience, Migration, and Business Continuity 15%
This domain covers moving workloads to AWS and keeping them running. It includes discovery, portfolio assessment and migration waves, rehosting with AWS Transform MGN, database and data migration with DMS and DataSync, replatforming and refactoring with the strangler fig pattern, and mainframe, .NET and database modernization. On the continuity side it covers DR strategies against RTO, RPO and cost, AWS Elastic Disaster Recovery, multi-Region failover with Route 53, Global Accelerator and Application Recovery Controller, multi-cluster containers, Resilience Hub assessments and Fault Injection Service experiments, backup governance and ransomware protection, and fault-tolerant, self-healing design.
Discovery, portfolio assessment and migration waves
Large migrations start with discovery, dependency mapping and a strategy for each application, grouped into waves. Know the 7 Rs, what discovery data you need, how AWS Transform now handles discovery, assessment and wave planning, and how to sequence waves so dependent systems move together.
Study notes and 1 free question on Discovery, portfolio assessment and migration wavesRehosting and data migration (AWS Transform MGN, DMS, DataSync)
Rehosting moves servers with continuous block-level replication, while databases and files move with purpose-built tools. Know AWS Transform MGN test and cutover launches, AWS Transform for VMware, DMS full load plus CDC, schema conversion with DMS Schema Conversion or SCT, and DataSync for file data.
Study notes and 1 free question on Rehosting and data migration (AWS Transform MGN, DMS, DataSync)Replatforming, refactoring and modernization (strangler fig)
Modernization questions ask how far to change an application after or instead of a rehost. Know replatforming to RDS, Aurora, ECS and Lambda, refactoring a monolith with the strangler fig pattern, and modernizing mainframes, .NET apps and commercial databases with AWS Transform.
Study notes and 1 free question on Replatforming, refactoring and modernization (strangler fig)DR strategies, RTO/RPO and Elastic Disaster Recovery
Disaster recovery strategy is a trade between RTO, RPO and cost: backup and restore, pilot light, warm standby or multi-site active/active. Know which AWS replication feature delivers each RPO and how AWS Elastic Disaster Recovery gives low RPO and RTO for servers at low standby cost.
Study notes and 2 free questions on DR strategies, RTO/RPO and Elastic Disaster RecoveryMulti-Region failover (ARC, Route 53, Global Accelerator, multi-cluster containers)
Multi-Region failover moves traffic reliably when a Region or AZ is impaired. Know Route 53 health checks and failover routing, Global Accelerator, Application Recovery Controller (routing controls, zonal shift, Region switch), data-layer failover and multi-cluster EKS or ECS.
Study notes and 2 free questions on Multi-Region failover (ARC, Route 53, Global Accelerator, multi-cluster containers)Resilience assessment and testing (Resilience Hub, Fault Injection Service)
A DR plan only counts once it has been assessed and tested. Know how AWS Resilience Hub measures an application against RTO and RPO targets and recommends fixes, how AWS Fault Injection Service runs controlled chaos experiments with stop conditions, and how to measure real failover and recovery times.
Study notes and 2 free questions on Resilience assessment and testing (Resilience Hub, Fault Injection Service)Backup governance and ransomware protection (AWS Backup, Object Lock)
Backups must be governed centrally and protected from the people and malware that might delete them. Know AWS Backup organization policies, delegated administration, cross-account and cross-Region copies, Backup Vault Lock, logically air-gapped vaults, S3 Object Lock, EBS snapshot lock and restore testing.
Study notes and 1 free question on Backup governance and ransomware protection (AWS Backup, Object Lock)Fault-tolerant and self-healing architectures
Fault-tolerant design removes single points of failure and lets the system heal itself. Know Multi-AZ patterns, health checks that replace failed parts, queues, retries with backoff and jitter, circuit breakers, bulkheads, cell-based architecture and static stability.
Study notes and 1 free question on Fault-tolerant and self-healing architectures
Domain 5: Operational Excellence and Automation 20%
This domain is worth 20% and covers running workloads with automation: CloudFormation, the CDK and StackSets across accounts and Regions, policy as code with CloudFormation Hooks, CDK aspects, Config rules and SCPs, secrets in IaC, cross-account CI/CD with CodePipeline, CodeBuild and CodeDeploy, pipeline security scanning, artifact signing and Bedrock evaluations, deployment strategies with automatic rollback, observability with CloudWatch, OpenTelemetry, Container Insights and Application Signals SLOs, multi-account log aggregation and alerting, AI and user experience monitoring, Systems Manager runbooks and Service Quotas, and safe remediation, auto scaling and automatic recovery.
Infrastructure as code and multi-account deployment (CloudFormation, CDK, StackSets)
Know when to use declarative CloudFormation or CDK code that synthesizes to it, how to deploy safely with change sets and rollback triggers, and how StackSets roll one baseline out to many accounts and Regions with service-managed permissions and blast-radius controls.
Study notes and 1 free question on Infrastructure as code and multi-account deployment (CloudFormation, CDK, StackSets)Policy as code and secrets in IaC (Hooks, CDK aspects, Secrets Manager)
Policy as code checks infrastructure at every stage: CDK aspects at synthesis, CloudFormation Hooks before provisioning, SCPs as hard limits and Config rules with remediation for drift. Secrets stay out of templates through dynamic references.
Study notes and 2 free questions on Policy as code and secrets in IaC (Hooks, CDK aspects, Secrets Manager)CI/CD and multi-account pipelines
CodePipeline, CodeBuild and CodeDeploy release from a tooling account into workload accounts and Regions. Know cross-account roles, customer managed KMS keys for artifacts, per-Region artifact stores and promoting one immutable artifact through every stage.
Study notes and 1 free question on CI/CD and multi-account pipelinesPipeline security scanning, signing and AI evaluation
Security belongs inside the pipeline: Amazon Inspector and ECR scanning find vulnerable packages, AWS Signer signs images and Lambda code so only trusted artifacts deploy, and Amazon Bedrock Evaluations scores AI outputs before release. Know basic versus enhanced scanning and where to place each gate.
Study notes and 2 free questions on Pipeline security scanning, signing and AI evaluationDeployment strategies with rollback (blue/green, canary, rolling)
Pick a deployment strategy by how fast you must roll back and how much risk you can take: all at once, rolling, canary, linear or blue/green. Know the CodeDeploy configurations for EC2, Lambda and ECS, ECS circuit breakers and built-in blue/green, Auto Scaling instance refresh, and alarm-driven automatic rollback.
Study notes and 1 free question on Deployment strategies with rollback (blue/green, canary, rolling)Metrics, logs, tracing and SLOs (CloudWatch, OpenTelemetry, Application Signals)
Standardize metrics, logs and traces with AWS Distro for OpenTelemetry and CloudWatch, monitor containers with Container Insights, and track service level objectives with CloudWatch Application Signals. The X-Ray SDKs are in maintenance mode, so new instrumentation should use OpenTelemetry.
Study notes and 2 free questions on Metrics, logs, tracing and SLOs (CloudWatch, OpenTelemetry, Application Signals)Multi-account log aggregation and alerting
Centralize logs from every account with CloudWatch Logs subscriptions to destinations, account-level subscription policies or organization log centralization, and archive them cheaply in S3. Alert with CloudWatch alarms, composite alarms and SNS, and hand off to Systems Manager for action.
Study notes and 1 free question on Multi-account log aggregation and alertingMonitoring AI workloads and user experience (Synthetics, RUM)
AI workloads need their own signals: Bedrock invocation metrics and logs, AgentCore observability for agent traces, and SageMaker Model Monitor for drift. User experience is measured from outside with CloudWatch Synthetics canaries and from real browsers with CloudWatch RUM.
Study notes and 2 free questions on Monitoring AI workloads and user experience (Synthetics, RUM)Operational automation and runbooks (EventBridge, Systems Manager, Service Quotas)
Operational responses should run themselves: CloudWatch alarms and AWS Health events go to EventBridge, which starts Systems Manager Automation runbooks, Lambda or Step Functions. Know Patch Manager patch policies, maintenance windows, OpsCenter, and Service Quotas monitoring for capacity planning.
Study notes and 1 free question on Operational automation and runbooks (EventBridge, Systems Manager, Service Quotas)Safe remediation, auto scaling and automatic recovery
Circuit breakers, rate limits, bulkheads and retries with backoff and jitter keep automation safe. Know target tracking, scheduled and predictive scaling, and automatic recovery with ELB health checks, ECS task replacement and Lambda retries and destinations.
Study notes and 2 free questions on Safe remediation, auto scaling and automatic recovery
An 8-week study plan
This plan assumes about an hour on weekdays and a longer, hands-on session at the weekend. Time follows the domain weights: design and security are 25% each, so they get the most weeks; operations (20%) gets a full week; cost and resilience (15% each) get a week apiece. If you already architect multi-account AWS environments at work, six weeks may be enough.
| Week | Focus | Do this |
|---|---|---|
| 1 | Baseline and cloud-native design | Take the free practice exam in timed mode. Then event-driven messaging (SQS, SNS, EventBridge, Kinesis ordering), Step Functions and serverless pipelines, ECS and EKS with scaling, VPC Lattice, ingress and edge (ALB, NLB, API Gateway, CloudFront), multi-tenant SaaS isolation. |
| 2 | Networking and hybrid | Transit Gateway, PrivateLink and Cloud WAN, VPC design with IPAM and IPv6, Network Firewall, WAF and Gateway Load Balancer, Direct Connect and VPN resilience, Route 53 Resolver hybrid DNS, Route 53 routing policies and Global Accelerator, Outposts and Local Zones, DataSync and Storage Gateway. |
| 3 | Data and AI | Database selection and scaling (Aurora, DynamoDB, global databases), storage and caching, data lakes with Lake Formation and Iceberg, Kinesis, MSK, EMR and Redshift, Amazon Bedrock integration, agents with AgentCore, RAG with Knowledge Bases, SageMaker AI endpoints. |
| 4 | Security: identity and encryption | IAM Identity Center with SAML and SCIM, Cognito and Verified Permissions, EKS Pod Identity, cross-account roles, permissions boundaries, ABAC, IAM Roles Anywhere, data perimeters and Access Analyzer, Verified Access, KMS (rotation, key stores, post-quantum keys), ACM, Private CA, CloudHSM. |
| 5 | Security: governance and compliance | Organizations, OUs, Control Tower and Account Factory, SCPs and RCPs, organization trails, Config aggregators, GuardDuty, Security Hub CSPM, Security Lake, RAM sharing, Inspector and Patch Manager, data residency controls, AWS Artifact, threat modeling, Bedrock Guardrails and human approval with Step Functions. |
| 6 | Resilience and migration | Discovery and migration waves, AWS Transform MGN, DMS and SCT, replatforming and strangler fig refactoring, DR strategies against RTO and RPO, Elastic Disaster Recovery, Application Recovery Controller, AWS Backup with Vault Lock and Object Lock, Resilience Hub and Fault Injection Service. |
| 7 | Operations and automation | CloudFormation, CDK and StackSets, CloudFormation Hooks, multi-account pipelines, deployment strategies with rollback, CloudWatch, OpenTelemetry and Application Signals SLOs, cross-account log aggregation, Synthetics and RUM, EventBridge and Systems Manager automation, auto scaling and safe remediation. |
| 8 | Cost and exam practice | Savings Plans, Reserved Instances and Spot, Graviton and Compute Optimizer, Bedrock cost levers, data transfer and NAT costs, tagging and cost categories, Budgets and Cost Anomaly Detection, S3 and EBS tiering. Then full timed practice exams; review every wrong answer. |
How to use practice questions well
- Practice the length. Do full 75-question sets in 180 minutes; reading stamina is half the challenge.
- Find the qualifier. Many wrong answers work but cost more, take more effort to run or miss one constraint.
- Read the wrong options. They are usually real services solving a neighboring problem, or the right service set up the wrong way.
- Don’t memorize questions. The real exam won’t reuse them.
- Aim for 80% or more across a few full timed practice exams before you book.
Get hands-on
In a personal account with a budget alert set: create an organization with a few OUs and an SCP and RCP, set up IAM Identity Center, share a subnet through RAM, build a Transit Gateway between two VPCs, call a Bedrock model with a guardrail attached, and run a Fault Injection Service experiment against an Auto Scaling group. Delete everything afterward.
Frequently asked questions
Should I take SAP-C02 or SAP-C03?
SAP-C02 can be taken until November 16, 2026, and SAP-C03 from November 17, 2026. If your exam date is after that, study for SAP-C03. The core is similar, but SAP-C03 adds generative AI, agents, responsible AI, post-quantum cryptography and more operations content, and the passing score drops from 750 to 720. See what changed from SAP-C02 to SAP-C03.
Do I need the associate certification first?
No, AWS has no prerequisites. Most candidates find Solutions Architect – Associate (SAA) knowledge a big help, because this exam assumes it and goes much further into multi-account, hybrid and migration design.
How much AI and ML do I need?
Architecture, not data science. Know how to integrate Amazon Bedrock, build agents and RAG, host models on SageMaker AI endpoints, control costs, and apply guardrails and logging. Model training and fine-tuning are out of scope, and individual AI services such as Comprehend or Textract are no longer on the service list.
Which services are no longer right answers?
AWS Application Migration Service is now AWS Transform MGN. OpsWorks has reached end of life. Incident Manager, Audit Manager and CloudTrail Lake are closed to new customers, and the X-Ray SDKs and daemon are in maintenance mode in favor of AWS Distro for OpenTelemetry. The exam guide also lists out-of-scope services such as Service Catalog, WorkSpaces, App Runner and AWS Batch.
Are the practice questions the real exam questions?
No. Every question on this site is original, written from the public exam guide and checked against AWS documentation. Using real exam questions breaks the AWS Certification Program Agreement.
Test yourself
The free SAP-C03 practice exam has 75 questions with an explanation for every option. No sign-up needed.
Want to be sure you're ready? The 5 full-length practice sets add 375 new questions, none of them in the free exam. One payment of $14.99, no subscription. See the practice sets