Cost guardrails: budgets, alarms and the usual traps
Set up a zero-spend budget, a small monthly budget, a billing alarm, Cost Anomaly Detection and Free Tier alerts, and learn which resources quietly cost money.
Assumes: .NET 10 LTS ยท AWS SDK for .NET v4 ยท AWS CLI v2 ยท Region us-east-1 ยท Terraform AWS provider 6.73. Verified against official sources on 2026-11-17.
Learning objectives
By the end of this lesson you can:
- Create a zero-spend budget and a monthly cost budget with actual and forecasted alerts.
- Create a CloudWatch billing alarm on the
EstimatedChargesmetric in us-east-1. - Tune Cost Anomaly Detection so it alerts on a learner-sized spike, not the default USD 100.
- Opt the management account in to Free Tier usage alerts.
- Name the common cost traps (NAT Gateway, idle load balancers, RDS, public IPv4, KMS keys, Secrets Manager, log retention, EBS volumes and snapshots, forgotten Regions) and their approximate monthly cost.
Prerequisites
- Secure your AWS account โ you need the management account and the
BillingViewer/ManagementAdminpermission sets. - IAM access to billing activated (done by root in the previous lesson).
- Tools: browser; optionally AWS CLI v2 with a
learn-managementprofile (Tooling setup).
Why this matters in banking
Banks are cost-conscious in a very specific way: unexpected spend is an operational risk signal. A sudden cost jump may mean a runaway batch job, a misconfigured auto-scaling group, or an attacker mining cryptocurrency with stolen credentials. FinOps and security teams watch the same dashboards. If you learn to treat a budget alert as an incident trigger โ "something is running that I did not expect" โ you are already thinking like a bank's cloud engineer.
The idea in plain language
AWS charges for what you use, mostly by the hour or by the gigabyte, and the bill arrives at the end of the month. Nothing stops spending automatically. So you install alarms that tell you early:
- AWS Budgets compares actual (and forecasted) spend against an amount you choose and emails you.
- The billing alarm is a CloudWatch alarm on the estimated monthly charges โ a second, independent alarm.
- Cost Anomaly Detection uses machine learning to spot spend that is unusual for you, even if it is below your budget.
- Free Tier usage alerts warn you when a service passes 85% of its free allowance.
Analogy: Budgets are the monthly spending limit you set on a debit card app. The billing alarm is the bank texting you when your balance crosses a line. Anomaly detection is the fraud team noticing a payment that does not look like you. None of them freezes the card โ you still have to act.
What problem these tools solve
| Tool | What it does | Cost | Limitations |
|---|---|---|---|
| AWS Budgets | Alerts on actual or forecasted cost/usage | Free for alert-only budgets. First two action-enabled budgets free, then USD 0.10 per day each; budget reports USD 0.01 each | Data updates up to three times a day; alerts can arrive hours after the spend |
| CloudWatch billing alarm | Alarm on AWS/Billing โ EstimatedCharges | Within the CloudWatch free tier of 10 standard alarm metrics | Metric exists only in us-east-1; must enable "Receive CloudWatch billing alerts"; cannot be disabled once enabled |
| Cost Anomaly Detection | ML-based spike detection with daily/weekly summaries or individual alerts | No charge โ VERIFY on the pricing page (not stated on the pages checked) | Needs some spend history; default subscription threshold is high (USD 100 and 40%) |
| Free Tier usage alerts | Email at 85% of a service's free allowance | Free | On by default for standalone accounts, off for a management account until you opt in |
| Cost Explorer | Charts and filters of your costs | Console use is free; API calls USD 0.01 each | Data available about 24 hours after you first open it |
The honest limitation: every one of these is an alert, not a brake. Billing data lags real usage by hours. A NAT Gateway left running overnight will cost money before any email reaches you. Your real protection is the teardown habit; alerts are the safety net.
Microsoft / Azure comparison
| You know (Azure) | AWS equivalent | What's different |
|---|---|---|
| Cost Management + Billing โ Budgets | AWS Budgets | AWS Budgets also supports usage, RI and Savings Plans budgets; budget actions can apply an IAM policy or SCP automatically |
| Azure cost alerts / anomaly alerts | Cost Anomaly Detection | Created automatically when you enable Cost Explorer, with a high default threshold you should lower |
| Azure Monitor alert on a cost metric | CloudWatch billing alarm | The billing metric lives only in us-east-1 and is in USD |
| Azure free account (12 months + credits) | Free plan / Paid plan with credits + always-free offers | Free plan accounts close after 6 months; Paid plan accounts never close and bill beyond credits |
| Spending limit on Azure free/credit subscriptions | No hard spending limit on the Paid plan | Only the Free plan caps spend (by closing/limiting); otherwise you rely on alerts and actions |
Architecture
All guardrails live in the management account because it receives the consolidated bill for every member account. A forgotten resource in the sandbox still triggers them.
Step-by-step implementation
Sign in to the access portal and open the management account with ManagementAdmin.
Create the zero-spend budget
Billing and Cost Management โ Budgets โ Create budget โ Use a template (simplified) โ Zero spend budget. Name it cea-zero-spend, add your email, create. It alerts as soon as spend goes above the free allowances. On the Paid plan, credits may cover early charges, so this alert tells you that credits are being consumed โ exactly what you want to know.
This also completes the "set up a cost budget" activity that earns USD 20 of credits for new customers.
Create a small monthly budget
Create budget โ Use a template โ Monthly cost budget. Name cea-monthly-5usd, amount 5.00 USD, your email. The template notifies you when you exceed, or are forecasted to exceed, the amount. Then edit the budget (or choose Custom under template settings) and add an alert at 50% of actual spend so you hear about problems earlier.
Enable billing alerts and create the billing alarm
- Billing and Cost Management โ Billing preferences โ Alert preferences โ Edit: tick Receive CloudWatch billing alerts and Receive AWS Free Tier alerts, add your email, save. Wait about 15 minutes for the metric to appear.
- Switch the console Region to US East (N. Virginia).
- CloudWatch โ Alarms โ All alarms โ Create alarm โ Select metric โ Billing โ Total Estimated Charge โ EstimatedCharges (USD).
- Statistic Maximum, period 6 hours, condition Greater than 10.
- Notification: create a new SNS topic
cea-billing-alertswith your email; confirm the subscription email. - Name the alarm
cea-billing-over-10usdand create it.
Tune Cost Anomaly Detection
Open Cost Explorer once (this enables it and, at the same time, creates a default AWS services monitor with a daily summary alert for anomalies over USD 100 and 40%). Then Cost Anomaly Detection โ Alert subscriptions โ edit the default subscription: set the threshold to absolute impact at least USD 5, keep Daily summaries, and make sure your email is a recipient. If no monitor exists yet, create one: Cost monitors โ Create monitor โ Managed by AWS โ AWS services.
Test that alerts reach you
Check your inbox for the SNS subscription confirmation and the budget creation notices. In CloudWatch the alarm should show OK (or Insufficient data for the first hours). If you want to see an email arrive, temporarily lower the alarm threshold to 0, wait for the next data point, then set it back to 10.
AWS Console instructions
Navigation used (checked 2026-11-17):
- Billing and Cost Management โ Budgets โ Create budget โ Use a template (simplified)
- Billing and Cost Management โ Billing preferences โ Alert preferences โ Edit
- CloudWatch (us-east-1) โ Alarms โ All alarms โ Create alarm โ Billing โ Total Estimated Charge
- Billing and Cost Management โ Cost Explorer (first visit enables it) and โ Cost Anomaly Detection โ Cost monitors / Alert subscriptions
AWS CLI
Run with the learn-management profile. JSON goes in files because PowerShell quoting of inline JSON is error-prone.
{
"BudgetName": "cea-monthly-5usd",
"BudgetLimit": { "Amount": "5", "Unit": "USD" },
"TimeUnit": "MONTHLY",
"BudgetType": "COST"
}[
{
"Notification": { "NotificationType": "ACTUAL", "ComparisonOperator": "GREATER_THAN", "Threshold": 50, "ThresholdType": "PERCENTAGE" },
"Subscribers": [ { "SubscriptionType": "EMAIL", "Address": "alex+aws-billing@example.com" } ]
},
{
"Notification": { "NotificationType": "FORECASTED", "ComparisonOperator": "GREATER_THAN", "Threshold": 100, "ThresholdType": "PERCENTAGE" },
"Subscribers": [ { "SubscriptionType": "EMAIL", "Address": "alex+aws-billing@example.com" } ]
}
]# Your management account ID (budgets are created per account).
$acct = aws sts get-caller-identity --query Account --output text --profile learn-management
# Create the monthly budget with two notifications.
aws budgets create-budget --account-id $acct --budget file://budget-monthly.json --notifications-with-subscribers file://budget-notifications.json --profile learn-management
# List budgets to confirm.
aws budgets describe-budgets --account-id $acct --query "Budgets[].BudgetName" --profile learn-management
# SNS topic + email subscription for the billing alarm (must be us-east-1).
$topic = aws sns create-topic --name cea-billing-alerts --region us-east-1 --query TopicArn --output text --profile learn-management
aws sns subscribe --topic-arn $topic --protocol email --notification-endpoint alex+aws-billing@example.com --region us-east-1 --profile learn-management
# Billing alarm: Maximum of EstimatedCharges over 6 hours greater than 10 USD.
aws cloudwatch put-metric-alarm --region us-east-1 --profile learn-management `
--alarm-name cea-billing-over-10usd --namespace AWS/Billing --metric-name EstimatedCharges `
--dimensions Name=Currency,Value=USD --statistic Maximum --period 21600 --evaluation-periods 1 `
--threshold 10 --comparison-operator GreaterThanThreshold --treat-missing-data missing `
--alarm-actions $topic
# Cost Anomaly Detection: see existing monitors and subscriptions (one may have been created for you).
aws ce get-anomaly-monitors --profile learn-management
aws ce get-anomaly-subscriptions --profile learn-management{
"Dimensions": {
"Key": "ANOMALY_TOTAL_IMPACT_ABSOLUTE",
"Values": ["5"],
"MatchOptions": ["GREATER_THAN_OR_EQUAL"]
}
}# Lower the default subscription's threshold to an absolute USD 5 impact.
aws ce update-anomaly-subscription --subscription-arn arn:aws:ce::444455556666:anomalysubscription/00000000-0000-0000-0000-000000000000 --threshold-expression file://anomaly-threshold.json --profile learn-managementInfrastructure as Code
variable "alert_email" {
type = string
description = "Where budget and billing alerts are sent"
}
provider "aws" {
region = "us-east-1" # the billing metric only exists here
profile = "learn-management"
}
resource "aws_budgets_budget" "monthly" {
name = "cea-monthly-5usd"
budget_type = "COST"
limit_amount = "5"
limit_unit = "USD"
time_unit = "MONTHLY"
notification {
comparison_operator = "GREATER_THAN"
threshold = 50
threshold_type = "PERCENTAGE"
notification_type = "ACTUAL"
subscriber_email_addresses = [var.alert_email]
}
notification {
comparison_operator = "GREATER_THAN"
threshold = 100
threshold_type = "PERCENTAGE"
notification_type = "FORECASTED"
subscriber_email_addresses = [var.alert_email]
}
}
resource "aws_sns_topic" "billing" {
name = "cea-billing-alerts"
}
resource "aws_sns_topic_subscription" "email" {
topic_arn = aws_sns_topic.billing.arn
protocol = "email"
endpoint = var.alert_email # you must click the confirmation link in the email
}
resource "aws_cloudwatch_metric_alarm" "billing" {
alarm_name = "cea-billing-over-10usd"
namespace = "AWS/Billing"
metric_name = "EstimatedCharges"
dimensions = { Currency = "USD" }
statistic = "Maximum"
period = 21600
evaluation_periods = 1
threshold = 10
comparison_operator = "GreaterThanThreshold"
treat_missing_data = "missing"
alarm_actions = [aws_sns_topic.billing.arn]
}Complete code
The budget JSON, CLI script and IaC above are complete and runnable. There is no C# application in this lesson; the module's C# code is the verify-setup lab.
Code explained line by line
Budget JSON and notifications:
"BudgetType": "COST"โ track money (not usage hours or RI coverage)."BudgetLimit": { "Amount": "5", "Unit": "USD" }โ the monthly amount. Note the amount is a string in the CLI JSON."TimeUnit": "MONTHLY"โ resets on the first of each month."NotificationType": "ACTUAL"withThreshold: 50andPERCENTAGEโ email when real spend passes USD 2.50."NotificationType": "FORECASTED"withThreshold: 100โ email when AWS predicts you will exceed USD 5 this month. Forecasts need some history, so this alert is quiet in a brand-new account.
Billing alarm:
--namespace AWS/Billing --metric-name EstimatedCharges --dimensions Name=Currency,Value=USDโ the month-to-date estimated total for the whole account (or organisation, in the payer account).--statistic Maximum --period 21600โ the highest value in each 6-hour window; the metric only updates a few times a day, so shorter periods add nothing.--treat-missing-data missingโ gaps between data points neither trigger nor clear the alarm.--alarm-actions $topicโ publish to SNS, which emails you.
Security considerations
- Send alerts to an inbox you actually read, and add a second recipient if possible (a partner, or a work-unrelated second address).
- Billing data reveals what you run; keep
BillingViewerseparate from admin roles so you can check costs without admin power. - An unexpected cost spike can be a security event (leaked credentials used for crypto-mining). If you cannot explain a spike within minutes, rotate access by signing out all sessions and check for unknown resources in every Region.
โ Learning shortcut โ not for production
The SNS topic here is not encrypted with a KMS key. Its messages contain only alarm metadata, and encrypting it properly needs a customer managed KMS key (USD 1 per month) whose key policy lets CloudWatch publish โ the AWS managed alias/aws/sns key does not allow that โ VERIFY. In a bank, alerting topics are encrypted with a customer managed key, budgets are owned per cost centre, and budget actions can automatically attach a deny policy when a threshold is crossed.
Banking use case
"Harbour Bank" (fictional) gives every development team a sandbox account with a monthly budget. At 80% of budget, a budget action attaches a restrictive SCP that blocks creating new EC2, RDS and NAT Gateway resources; the team lead is paged; existing workloads keep running. Cost Anomaly Detection runs on a managed Linked account monitor in the management account and posts to the FinOps channel, while the security operations team receives the same anomalies because a spike in GPU instance hours is a classic sign of compromised credentials.
Cost traps you will meet in this course
Prices are US East (N. Virginia), on-demand, checked 2026-11-17 on the official pricing pages. A month is taken as 730 hours. Your credits may cover these, but they still burn credits.
| Trap | How it bills | Approx. cost if forgotten | How it sneaks up on you | How to find it |
|---|---|---|---|---|
| NAT Gateway | USD 0.045 per hour + USD 0.045 per GB processed | about USD 33/month each, before data | VPC wizards offer "1 NAT Gateway per AZ" โ three AZs is about USD 99/month idle | VPC โ NAT gateways (every Region) |
| Application Load Balancer | USD 0.0225 per hour (partial hours billed as full) + USD 0.008 per LCU-hour | about USD 16/month idle | Left behind when you delete the ECS service or Beanstalk app but not the ALB | EC2 โ Load balancers |
| RDS database | Per hour while running, plus storage and backups | db.t4g.micro PostgreSQL Single-AZ USD 0.016/h โ USD 11.68/month; SQL Server Express db.t3.micro USD 0.022/h โ USD 16/month; Multi-AZ roughly doubles | "Stopped" RDS instances restart automatically after 7 consecutive days; snapshots keep billing | RDS โ Databases and Snapshots |
| Public IPv4 addresses | USD 0.005 per hour per address, in use or idle (includes Elastic IPs) | about USD 3.65/month per address | Every public subnet instance, NAT Gateway and ALB node uses them | VPC โ Public IPs / EC2 โ Elastic IPs; Amazon VPC IP Address Manager public IP insights |
| KMS customer managed keys | USD 1 per key per month (prorated hourly), +USD 1/month for each of the first two rotations; 20,000 free requests/month | USD 1โ3/month per key | Every lab that creates "its own key" adds one; keys pending deletion are not charged | KMS โ Customer managed keys (every Region) |
| Secrets Manager | USD 0.40 per secret per month + USD 0.05 per 10,000 API calls | USD 0.40/month per secret | Each RDS lab can create a managed master-user secret | Secrets Manager โ Secrets |
| CloudWatch Logs retention | USD 0.50 per GB ingested (Standard) + USD 0.03 per GB-month stored; 5 GB/month free | Grows forever | New log groups default to Never expire; a chatty Lambda in a loop can ingest GBs per day | CloudWatch โ Log groups โ Retention column |
| EBS volumes and snapshots | gp3 USD 0.08 per GB-month; snapshots USD 0.05 per GB-month | USD 2.40/month for a 30 GB volume | Terminating an instance may leave extra volumes; AMIs keep their snapshots | EC2 โ Volumes (state "available") and Snapshots |
| Forgotten Regions | Any of the above, in a Region you never look at | Anything | Console defaults to the last Region used; a tutorial used eu-central-1 | Billing โ Bills โ charges by Region; Tag Editor with All regions |
The Region-restricting SCP from the previous lesson shrinks the last trap to two Regions, and the teardown checklist catches the rest.
Common mistakes
- Creating the billing alarm in your usual Region โ the metric is missing โ the billing metric exists only in us-east-1.
- Not confirming the SNS subscription email โ the alarm fires into the void โ check the topic's subscription status is "Confirmed".
- Relying on the default anomaly threshold โ USD 100 and 40% is far above learner spend โ lower it to USD 5.
- Assuming management accounts get Free Tier alerts โ they do not by default โ tick "Receive AWS Free Tier alerts".
- Thinking a budget stops spending โ it only notifies (unless you configure actions) โ delete resources yourself.
- Assuming "free tier eligible" means free โ limits are monthly and per service; on the Free plan only always-free offers apply โ read the limit, not the badge.
Troubleshooting
| Symptom | Likely cause | Fix |
|---|---|---|
| No Billing namespace in CloudWatch | Billing alerts not enabled, or wrong Region | Enable "Receive CloudWatch billing alerts"; switch to us-east-1; wait about 15 minutes |
| "Access denied" on Budgets or Cost Explorer | IAM access to billing not activated, or permission set lacks billing permissions | Activate IAM access as root once; use ManagementAdmin or BillingViewer |
| Budget shows USD 0 although you used resources today | Billing data delay (up to three updates per day) | Wait; check Cost Explorer the next day |
| Alarm stuck in Insufficient data | New metric, few data points | Normal for the first hours; it moves to OK after data arrives |
| Cost Explorer says data is being prepared | First visit enables it | Current-month data appears in about 24 hours |
| Credits shown but charges still appear | Some services or usage types are not covered by credits, or credits expired | Billing โ Credits shows what applies; check the service list on the Free Tier page |
Cost and cleanup
Cost: USD 0. Alert-only budgets are free, one billing alarm is within CloudWatch's free alarms, and a handful of alert emails is negligible (SNS email free allowance โ VERIFY on the SNS pricing page โ it is not listed on the page checked). Keep these guardrails for the whole course.
If you ever remove them:
aws budgets delete-budget --account-id $acct --budget-name cea-monthly-5usd --profile learn-management
aws cloudwatch delete-alarms --alarm-names cea-billing-over-10usd --region us-east-1 --profile learn-management
aws sns delete-topic --topic-arn $topic --region us-east-1 --profile learn-managementExercises
- ๐ข Guided: Create the zero-spend budget, the USD 5 monthly budget, the billing alarm and the tuned anomaly subscription. Screenshot nothing โ instead, list each guardrail with its name and threshold in your learning journal.
- ๐ก Challenge: Create a budget action on the monthly budget that, at 100% of actual spend, attaches an SCP to the Sandbox OU denying
ec2:RunInstances,rds:CreateDBInstanceandec2:CreateNatGateway. Test the SCP manually (attach and detach) rather than by spending money. - ๐ด Live problem: On day 9 of the month you receive: "AWS Budgets: cea-monthly-5usd has exceeded 50%". You have not deployed anything for three days. Find the cause using only billing tools, decide what to delete, and write a five-line incident note.
Assignment
Add infra/guardrails/ to your learning-journal repository with the Terraform (or CDK) above, parameterised by email, plus a README.md explaining each alert, its threshold and why you chose it.
Solutions
You should have four guardrails: cea-zero-spend (template), cea-monthly-5usd (50% actual, 100% forecast, plus template defaults), cea-billing-over-10usd (us-east-1, SNS email confirmed) and the anomaly subscription at USD 5 absolute. Why: they overlap on purpose. Budgets can lag; the alarm is independent; anomaly detection catches unusual patterns below your budget.
Budgets โ cea-monthly-5usd โ Actions โ Add action โ action type Service control policy, target the Sandbox OU, threshold 100% actual, approval automatic. Budget actions need an IAM role that Budgets can assume with permission to attach SCPs (organizations:AttachPolicy, organizations:DetachPolicy). The two free action-enabled budgets cover this. Why: in a sandbox, automatic "stop creating things" is better than an email you read the next morning โ but it never deletes what already runs.
Cost Explorer โ group by Service, then by Region and Usage type, daily granularity, last 10 days. A typical finding: NatGateway-Hours in eu-west-2 from a VPC created by a wizard, or EBS:VolumeUsage.gp3 from volumes left after terminating instances. Confirm in the console, delete, and verify the next day that the daily cost drops to near zero. Incident note: what alerted you, the cause, the cost so far, the fix, and the habit you will change (for example, "run the teardown checklist after every session"). Why: cost alerts are incident signals; the investigation method is the same one FinOps teams use.
Knowledge check
Answer: us-east-1. Billing metric data for the whole account is published only there, regardless of where resources run.
Answer: No. A budget only notifies, unless you configure a budget action (apply an IAM policy or SCP, or target specific EC2/RDS instances). Even actions do not delete existing resources.
Answer: About 0.045 ร 720 โ USD 32, plus a public IPv4 address at USD 0.005/hour (โ USD 3.60) and any data processing at USD 0.045/GB.
Answer: Free Tier usage alerts are automatic for standalone accounts but not for an Organizations management account; you must opt in under Billing preferences โ Alert preferences.
Interview questions
- "How do you prevent surprise cloud bills in a development environment?" โ Separate accounts, budgets per account with actions, anomaly detection, mandatory tags, IaC so environments can be destroyed and rebuilt, scheduled shutdown of non-production, and a review of the top cost drivers each week.
- "A developer says the AWS bill doubled. Where do you start?" โ Cost Explorer grouped by service, account, Region and usage type over daily granularity; then correlate with deployments and CloudTrail; check for leaked credentials if usage is unfamiliar.
Certification practice
A company wants to be notified when its forecasted monthly AWS spend will exceed USD 1,000. Which service should it use?
A. AWS Cost Explorer B. AWS Budgets C. AWS Pricing Calculator D. AWS Trusted Advisor
Answer: B. AWS Budgets supports forecasted cost alerts. Cost Explorer visualises and forecasts but does not send threshold notifications; the Pricing Calculator estimates costs before deployment; Trusted Advisor gives optimisation checks, not budget alerts.
A development VPC has private subnets in three Availability Zones, each with its own NAT Gateway. Developers use the environment only during office hours, and outbound traffic is minimal. What is the MOST cost-effective change that keeps outbound internet access for private subnets during the day?
A. Replace the NAT Gateways with a single NAT Gateway and delete it outside office hours with automation, recreating it each morning B. Move all instances to public subnets with public IPv4 addresses C. Replace the NAT Gateways with VPC gateway endpoints D. Keep three NAT Gateways for high availability
Answer: A. One NAT Gateway, created only when needed (with IaC), removes most of the fixed hourly cost in a non-critical environment. B increases exposure and public IPv4 charges. C only covers S3 and DynamoDB, not general internet access. D is the most expensive and unnecessary for development.
Completion checklist
- Zero-spend and USD 5 monthly budgets exist and I received the emails
- Billing alerts enabled;
cea-billing-over-10usdalarm in us-east-1 with a confirmed SNS email - Cost Anomaly Detection subscription threshold lowered to USD 5
- Free Tier alerts opted in for the management account
- I can name five cost traps and roughly what each costs per month
Key takeaways
- Budgets, billing alarms and anomaly detection are alerts, not brakes โ teardown is the real control.
- Put guardrails in the management account so they cover every member account.
- The billing metric lives in us-east-1; enable it before creating the alarm.
- Lower the default Cost Anomaly Detection threshold (USD 100 and 40%) to a learner-sized amount.
- The biggest silent costs are hourly: NAT Gateways, load balancers, databases and public IPv4 addresses.
Further reading
- Managing your costs with AWS Budgets and budget templates
- AWS Budgets pricing
- Create a billing alarm
- Getting started with Cost Anomaly Detection and enabling Cost Explorer
- Tracking your Free Tier usage
- Pricing pages: VPC (NAT Gateway, public IPv4), Elastic Load Balancing, RDS for PostgreSQL, KMS, Secrets Manager, CloudWatch, EBS
- AWS Well-Architected โ Cost Optimization Pillar
Independent learning resource โ not affiliated with, endorsed or sponsored by Amazon Web Services. AWS and related marks are trademarks of Amazon.com, Inc. or its affiliates. All banking data in examples is fictional. Sample architectures are for learning and are not automatically compliant with GDPR, PCI DSS or any banking regulation.
Secure your AWS account
Lock down the root user, sign in through IAM Identity Center, isolate labs in a sandbox account and restrict Regions with a service control policy.
Tooling setup on Windows
Install and configure the .NET 10 SDK, AWS CLI v2 with IAM Identity Center profiles, an AWS Toolkit for your IDE, SAM, CDK, Terraform, a container runtime, Git, Python and an API client.