AWS IV: Operations Automation
Running AWS by code instead of by hand: CloudFormation stacks, change sets, rollbacks and drift, StackSets across Regions; Systems Manager Automation runbooks, Patch Manager, maintenance windows and State Manager; EventBridge rules that react to what happens in the account; AWS Config rules with automatic remediation; Secrets Manager rotation and ACM certificates; and the security findings of GuardDuty, Security Hub, Inspector and Access Analyzer worked like incidents - against a simulated account with real outputs and real errors.
In plain words
Imagine a big building where every repair used to be done by whoever was on duty, from memory, with a torch. Now the building gets a set of written recipes for every repair, a robot that follows them, sensors that notice a door left open and ring the right bell, an inspector who walks the corridors every day with a checklist, a safe for the keys that changes its own combination, and a guard who reports strangers.
On AWS those are CloudFormation and Systems Manager (the recipes and the robot), EventBridge (the sensors and bells), AWS Config (the inspector), Secrets Manager and ACM (the safe), and GuardDuty with its colleagues (the guards).
Why it matters on call
An account run by hand drifts: the console fix nobody wrote down, the patch someone forgot, the password that never changed, the certificate that expired on a Saturday. This chapter is about the account fixing itself and changing safely - stacks that roll back instead of half-applying, change sets that show a replacement before it happens, runbooks with rate control and approvals, patch windows in the right time zone, rules that remediate what they find, secrets that rotate, and findings that page a person. Each comes with the incident it prevents, because that is how you will meet them at work and on the SOA-C03 exam.
Lessons
- 37.1CloudFormation: templates, stacks and events
- 37.4Change sets, replacement, and protecting what matters
- 37.7When a stack fails: rollbacks, UPDATE_ROLLBACK_FAILED and drift
- 37.10Exports, pipelines and StackSets: CloudFormation at scale
- 37.12Systems Manager: managed instances, Run Command and Automation runbooks
- 37.15Patch Manager, maintenance windows and State Manager
- 37.19EventBridge: rules, patterns, targets and the auto-remediation loop
- 37.24AWS Config: recording, rules, remediation and conformance packs
- 37.29Secrets and certificates: Parameter Store, Secrets Manager rotation, KMS and ACM
- 37.33Security findings: GuardDuty, Security Hub, Inspector, Access Analyzer and the advisors
Open this chapter in the terminal Free, in your browser - a real Ubuntu terminal to try it in, with missions that check your work.
Questions people ask
Do I need AWS I-III before this chapter?
It helps a lot. The chapter runs in the same simulated account and uses what they taught: IAM roles and trust policies, the CLI with --query, SNS topics, CloudWatch alarms and CloudTrail. Where a lab needs something older, such as a VPC or an instance profile, the setup builds it and the lesson explains the part you touch.
Why CloudFormation if the course already teaches Terraform?
Because you will meet it anyway: landing zones, Control Tower, vendor products, the CDK and SAM all produce CloudFormation stacks, and the CloudOps exam assumes it. The chapter compares the two honestly - state, previews, rollback, drift - and the incidents show what happens when both tools touch the same resource.
Is everything automatic at the end of this chapter?
No, and it should not be. Automatic remediation is right where there is exactly one acceptable state, such as a public bucket in an account that must never have one. Where exceptions exist the chapter uses manual remediation, approvals in runbooks and maintenance windows, so a person decides and the automation does the typing.
Would the labs cost money on a real account?
Some would. Config charges per configuration item and rule evaluation, Security Hub and GuardDuty bill per event analysed after their free trials, Secrets Manager charges per secret per month, and the instances cost what instances cost. Turn the security services on in a sandbox for a week, set a budget first, and delete stacks and secrets when you are done.
What is simulated and what is real here?
The commands, options, output formats and error messages follow the AWS CLI v2 and the current documentation. The account behind them is simulated (simulator): stacks run on the lab clock, the GuardDuty findings and Inspector CVEs are lab data, and other member accounts of the organization are modelled rather than real. What you type works the same way against a real account.