Browse AWS Certification Guides

SOA-C03 Deployment, Provisioning, and Automation Guide

AWS SOA-C03 provisioning guide covering automation, CloudFormation, CDK, deployment, and rollback decisions.

This chapter focuses on how CloudOps engineers build, roll out, and automate AWS environments without drifting into brittle manual operations. SOA-C03 is not testing whether you can describe CI/CD at a high level. It is testing whether you can provision resources repeatably, troubleshoot failed rollout paths, handle multi-account delivery, and automate the management of existing infrastructure safely.

I verified the current SOA-C03 Domain 3 task split against the AWS exam guide on May 11, 2026. AWS currently breaks this domain into two tasks:

  1. provision and maintain cloud resources
  2. automate the management of existing resources

Task 3.1 is broad enough that it needs several teachable lanes rather than one compressed page.

Current weight in the exam guide

AWS currently weights Deployment, Provisioning, and Automation at 22% of scored content.

What this domain is really testing

AWS wants to know whether you can:

  • build repeatable images and stack definitions
  • distinguish infrastructure provisioning from day-two operations automation
  • troubleshoot deployment failures by reading the first failing step
  • roll out resources consistently across accounts and Regions
  • choose deployment strategies like rolling, blue/green, or canary based on risk
  • work with both AWS-native and third-party deployment tooling responsibly

Work this domain in order

Lesson Official task fit Why it matters
3.1 Images, CloudFormation & CDK Task 3.1 Start with the native provisioning model: images, stacks, and repeatable resource definitions.
3.2 Multi-Account Provisioning, Deployment Strategies & Third-Party Tooling Task 3.1 This is the rollout lane: StackSets, AWS RAM, deployment strategies, Terraform-style tooling, and delivery tradeoffs.
3.3 Ops Automation & Deployment Troubleshooting Task 3.2 This is the day-two lane: Systems Manager, event-driven automation, patching, scheduling, and automation debugging.

Fast routing inside this chapter

If the stem is really about… Go first to… What to look for
AMIs, container images, CloudFormation, CDK, stack lifecycle, or template failures 3.1 Images, CloudFormation & CDK image hygiene, stack events, first-failure diagnosis
StackSets, AWS RAM, cross-account rollout, rolling or canary delivery, or Terraform/Git-based deployment automation 3.2 Multi-Account Provisioning, Deployment Strategies & Third-Party Tooling scope of rollout, sharing model, deployment-risk control
Systems Manager, patching, EventBridge, Lambda automation, scheduled tasks, or event-driven ops against existing resources 3.3 Ops Automation & Deployment Troubleshooting automation scope, target path, permissions, verification

Common domain traps

Trap Better thinking
“Provisioning and automation are basically the same thing.” Provisioning creates or changes resource state; operations automation manages existing resources day to day.
“CDK and CloudFormation are competing control planes.” CDK is a higher-level definition path that still synthesizes to CloudFormation.
“Multi-account rollout just means running the same command repeatedly.” AWS expects controlled mechanisms like StackSets, RAM, and governed roles.
“If a deployment fails, retry first.” Strong answers investigate stack events, permissions, quotas, subnet shape, or dependency failures before retrying.

How strong SOA-C03 answers usually work

  1. Decide whether the task is provisioning, rollout at scale, or day-two automation.
  2. Choose the service or tool that matches that lane.
  3. Troubleshoot from the first failing resource or first broken automation step.
  4. Prefer versioned, reviewable, repeatable deployment paths over console drift.
  5. Treat multi-account delivery and automated operations as explicit design choices, not afterthoughts.

In this section

Revised on Monday, June 15, 2026
2. Reliability & Continuity
4. Security & Compliance