Microsoft-to-AWS overview
A first map from the Microsoft and Azure services you know to their AWS equivalents, plus the five mindset differences that matter most.
Assumes: .NET 10 LTS ยท AWS SDK for .NET v4 ยท AWS CLI v2 ยท Region us-east-1. Verified against official sources on 2026-11-17.
Learning objectives
By the end of this lesson you can:
- Name the AWS service that most closely matches 15 Microsoft/Azure services you use today.
- Explain the five mindset differences between Azure and AWS that cause the most confusion for .NET developers.
- Sketch a typical ASP.NET Core + SQL Server application on AWS.
- Know where to find the full mapping and the glossary when you meet an unfamiliar term.
Prerequisites
- Secure your AWS account โ you have seen accounts, Organizations and IAM Identity Center in practice.
- Familiarity with Azure or on-premises Microsoft hosting (IIS, SQL Server, Azure DevOps).
Why this matters in banking
When a bank migrates from Microsoft to AWS, many of the expensive mistakes come from false friends: assuming an AWS service behaves like the Azure one with a similar name. "Key Vault = Secrets Manager" is only half true (keys live in KMS). "Resource group = stack" breaks when someone deletes a stack expecting everything to go. A precise mental map lets you ask the right questions in design reviews and spot risks before they reach production.
The idea in plain language
Most things you build on Azure have an AWS counterpart. The names differ, and so do some boundaries โ especially around identity, accounts and networking. This page gives you the 15 most important pairs and the differences that bite. Module 13 then goes service by service with side-by-side code.
Analogy: Moving from Azure to AWS is like a teller moving to a bank in another country. Accounts, payments and vaults all exist, but the forms have different names, some approvals happen at a different desk, and a few rules (like who can sign for what) are stricter. You do not relearn banking โ you relearn where things are.
What problem this map solves
It gives you a starting vocabulary so you can follow later modules and conversations with AWS-experienced colleagues. Limitation: one-line mappings hide real differences in limits, pricing, features and operational behaviour. Use them to orient yourself, never as a design decision on their own.
Microsoft / Azure comparison
The orientation map (verified for current service names on 2026-11-17):
| # | You know (Microsoft/Azure) | AWS equivalent | What's different |
|---|---|---|---|
| 1 | Tenant, management groups, subscriptions | AWS Organizations, organisational units (OUs), AWS accounts | The account is the security, quota and billing boundary; use many of them |
| 2 | Entra ID (workforce sign-in) | IAM Identity Center (can federate with Entra ID) | Permission sets become IAM roles inside each account |
| 3 | Entra ID B2C / External ID (customer sign-in) | Amazon Cognito | User pools issue JWTs for your APIs (Module 4) |
| 4 | Azure RBAC role assignments, managed identities | IAM policies and roles (task roles, Lambda execution roles) | Every API call is signed and checked against JSON policies; no network-only trust |
| 5 | Azure Policy | Service control policies (preventive) + AWS Config rules (detective) | SCPs never grant; Config evaluates resource settings after the fact |
| 6 | Virtual network (VNet), NSGs | Amazon VPC, security groups, network ACLs | Subnets are tied to one Availability Zone; NAT Gateways cost money per hour |
| 7 | IIS on Windows Server / App Service | Amazon ECS on AWS Fargate behind an Application Load Balancer, or AWS Elastic Beanstalk; Amazon EC2 for lift-and-shift | Containers on Fargate are the common modern target for ASP.NET Core |
| 8 | Azure Functions | AWS Lambda (managed dotnet10 runtime) | Packaging, cold starts and triggers differ (Module 3) |
| 9 | Windows Service / WebJobs / worker roles | ECS task on Fargate running a .NET Worker Service, or Lambda + EventBridge Scheduler | Long-running work goes to containers; Lambda runs for at most 15 minutes |
| 10 | SQL Server / Azure SQL Database | Amazon RDS for SQL Server, or Aurora PostgreSQL (optionally with Babelfish) | Licence-included vs bring-your-own; you manage less, but not nothing |
| 11 | Blob Storage | Amazon S3 | Bucket names are global; Block Public Access and bucket policies are central to security |
| 12 | Service Bus queues/topics, Event Grid | Amazon SQS, Amazon SNS, Amazon EventBridge | Several smaller services instead of one broker; FIFO and at-least-once semantics differ (Module 8) |
| 13 | Key Vault | AWS Secrets Manager (secrets) + AWS KMS (keys) + AWS Certificate Manager (TLS certs) | Three services; KMS key policies are a second permission layer |
| 14 | Application Insights / Azure Monitor | Amazon CloudWatch, AWS X-Ray, OpenTelemetry (ADOT) | Logs, metrics and traces are separate features you wire together |
| 15 | Azure DevOps Pipelines, ARM/Bicep | GitHub Actions with OIDC or AWS CodePipeline/CodeBuild; AWS CloudFormation / AWS CDK (C#) | CDK lets you write infrastructure in C#; pipelines assume roles instead of storing keys |
The full mapping, with many more services and migration notes, is in the Microsoft-to-AWS reference, and every term on this page is defined in the glossary.
Five mindset differences
- Accounts, not resource groups, are your boundary. In Azure you often put many environments in one subscription with resource groups. In AWS, separate accounts per environment (and per team) are normal and cheap.
- Identity is checked on every call. There is no "it's inside the VNet so it's trusted". Every request to an AWS API is signed (SigV4) and evaluated against identity policies, resource policies, SCPs and sometimes KMS key policies. Most early
AccessDeniederrors come from forgetting one of these layers. - Regions and Availability Zones are explicit. You choose the Region for almost every call, and high availability means placing resources in multiple AZs yourself.
- Some services are global, most are Regional. IAM, Organizations, Route 53 and CloudFront are global; S3 bucket names are global but buckets live in a Region; almost everything else is per Region. That is why the Region SCP needed exemptions.
- Hourly billing rewards teardown. Many resources bill per hour whether used or not. IaC plus teardown discipline is how AWS teams keep non-production costs low.
Architecture
A typical ASP.NET Core + SQL Server application, side by side:
Security-boundary view: in the AWS version, the container never holds secrets in configuration. Its task IAM role grants permission to read one secret and use one KMS key, and the database sits in private subnets reachable only from the ECS security group. Module 3 and Module 5 build this.
Step-by-step: use the map
Find your current stack in the table
List the five Microsoft services your team uses most. Find each in the table above or in the full mapping.
Read the "What's different" column
For each, write one question you would ask before migrating (for example, "Who owns the KMS key policy?").
Look up unfamiliar terms
Use the glossary for terms such as AZ, ARN, SCP or task role.
AWS Console instructions
Not applicable โ this is a conceptual lesson. Later modules give console steps for each service.
AWS CLI
A quick way to see the global-vs-Regional difference for yourself (read-only, free):
# IAM is global: no Region needed, same answer from anywhere.
aws iam list-roles --max-items 3 --profile learn-sandbox
# SQS is Regional: each Region has its own (separate) list.
aws sqs list-queues --region us-east-1 --profile learn-sandbox
aws sqs list-queues --region eu-west-2 --profile learn-sandboxInfrastructure as Code
Not applicable here. Module 10 compares ARM/Bicep with CloudFormation, CDK (C#) and Terraform in depth.
Complete code
Not applicable โ the module's runnable code is the verify-setup lab. Module 13 contains side-by-side Azure and AWS C# examples.
Security considerations
- Do not map Azure's network-perimeter trust onto AWS. Design for IAM on every call and private networking as an additional layer.
- Key Vault access policies or RBAC become two controls in AWS: the IAM policy and the KMS key policy. Both must allow an operation.
- Federating IAM Identity Center with Entra ID keeps one joiner-mover-leaver process โ a common audit requirement in banks.
โ Learning shortcut โ not for production
One-line equivalences are learning aids. A real migration needs a service-by-service assessment of features, limits, data residency, encryption, logging and licensing, reviewed by architecture, security and compliance. Module 13 covers that process.
Banking use case
"Harbour Bank" (fictional) runs its customer-statements API on App Service with Azure SQL and Key Vault. The migration design maps it to ECS on Fargate, RDS for SQL Server in Multi-AZ, Secrets Manager with rotation, and a customer managed KMS key. In design review, the security architect asks who can use the KMS key (no Azure equivalent of a key policy in their old design), and the data-protection officer asks which Region holds customer data โ the answer, eu-west-2, is enforced by an SCP. Neither question would have surfaced from a simple name mapping.
Common mistakes
- Treating a CloudFormation stack like a resource group โ deleting the stack deletes only what it created โ design stacks deliberately.
- Assuming Secrets Manager also holds encryption keys โ that is KMS โ plan both services and their policies.
- One account for dev, test and prod โ habit from single-subscription setups โ use separate accounts.
- Ignoring the Region parameter โ resources appear "missing" โ always check which Region you are in.
Troubleshooting
| Symptom | Likely cause | Fix |
|---|---|---|
| "My queue is gone" | Looking in a different Region | Check the Region selector / --region |
AccessDenied although the role "has admin" | SCP, resource policy or KMS key policy denies | Read the full error; check each policy layer |
| Can't find an AWS equivalent for a feature | Feature lives in a different service, or doesn't exist | Search the full mapping and the official docs; ask in design review |
Cost and cleanup
Cost: USD 0. The CLI commands are read-only and free. Nothing to clean up.
Exercises
- ๐ข Guided: For your current (or last) work project, fill in a three-column table: Microsoft service โ AWS service โ one migration question.
- ๐ก Challenge: Redraw the Mermaid diagram for an application that uses Azure Functions with a Service Bus trigger and Cosmos DB. Which AWS services replace each part?
- ๐ด Live problem: A manager asks, "Can we just lift our App Service app to AWS in a week?" Write a five-bullet answer listing what you would need to know first.
Assignment
Add docs/my-stack-map.md to your learning-journal repository with the table from the guided exercise and your redrawn diagram. Revisit it after Module 13 and note what you got wrong.
Solutions
Example rows: "Azure SQL โ RDS for SQL Server โ Do we need licence-included or BYOL, and Multi-AZ?"; "Key Vault โ Secrets Manager + KMS โ Who administers KMS key policies?"; "App Service โ ECS on Fargate โ Are we ready to containerise, and who patches base images?" Why: the question column turns a vocabulary list into a migration risk list.
Azure Functions โ Lambda (dotnet10); Service Bus trigger โ SQS queue as a Lambda event source (or EventBridge for event routing); Cosmos DB โ DynamoDB. Add a dead-letter queue for failed messages, an IAM execution role for the function, and CloudWatch Logs. Why: serverless patterns map closely, but retries, batching and idempotency are configured differently (Module 8).
- Data classification and residency โ which Region is allowed for customer data? 2. Identity โ how do staff and customers authenticate, and is IAM Identity Center federated with Entra ID? 3. Dependencies โ databases, queues, secrets, certificates, and their AWS targets. 4. Security and compliance sign-off โ logging, encryption keys, network design, change approval. 5. Operations โ monitoring, backups, DR targets and who is on call. Why: "lift in a week" is possible technically for simple apps, but a bank's controls and approvals are the real critical path.
Knowledge check
Answer: AWS Secrets Manager (secrets such as passwords and API keys) and AWS KMS (encryption keys). AWS Certificate Manager covers TLS certificates.
Answer: An AWS account, grouped into organisational units within AWS Organizations.
Answer: False. Every AWS API call is authenticated and authorised by IAM, regardless of network location.
Interview questions
- "How does AWS IAM differ from Azure RBAC?" โ IAM uses JSON policies attached to identities and resources, evaluated together with SCPs, permission boundaries and session policies; explicit denies win. Azure RBAC assigns roles at scopes. Concepts overlap, but AWS adds resource-based policies and organisation-level guardrails that you must reason about together.
- "What would you use instead of Azure Functions and Service Bus on AWS?" โ Lambda with SQS (or SNS/EventBridge for fan-out and routing), with DLQs, idempotency and IAM execution roles.
Certification practice
A company is moving a .NET application from Microsoft Azure. Which AWS service should it use to store database passwords with automatic rotation?
A. AWS Key Management Service (AWS KMS) B. AWS Secrets Manager C. AWS Certificate Manager (ACM) D. Amazon S3 with server-side encryption
Answer: B. Secrets Manager stores secrets and supports rotation. KMS manages encryption keys, not passwords. ACM manages TLS certificates. S3 is object storage and has no rotation feature for secrets.
Which AWS service runs containers without requiring the customer to manage servers or clusters of EC2 instances?
A. Amazon EC2 B. AWS Fargate C. Amazon Lightsail D. AWS Outposts
Answer: B. Fargate is serverless compute for containers (used with ECS or EKS). EC2 requires managing instances; Lightsail provides simple virtual servers; Outposts runs AWS infrastructure on premises.
Completion checklist
- I can name the AWS equivalent for each of the 15 rows without looking
- I can explain the five mindset differences
- I bookmarked the full mapping and the glossary
Key takeaways
- Most Microsoft services have an AWS counterpart, but boundaries differ โ especially accounts, identity and networking.
- Key Vault splits into Secrets Manager, KMS and ACM; Service Bus splits into SQS, SNS and EventBridge.
- Every AWS API call is checked by IAM; network location never replaces identity.
- Use the map to orient yourself, then verify details before any design decision.
Further reading
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.
Teardown discipline: find anything still costing money
Tag every lab resource, destroy it with the same tool that created it, and run a cross-Region checklist so nothing keeps billing.
Lab: verify your setup with C#
Run a .NET 10 console app that uses your IAM Identity Center profile to call STS and S3, and prints the account, caller ARN and Region.