Skip to Content
Cloud Engineering Academy
3. .NET Application Development on AWS

Module 3 cheat sheet

One-page revision summary for .NET Application Development on AWS: services, Azure mapping, CLI commands, limits, decision rules, exam traps and cost traps.

DVA-C03SAA-C03

Assumes: .NET 10 LTS · AWS SDK for .NET v4 · AWS CLI v2 · Lambda dotnet10 · Region us-east-1 · facts as stated in the Module 3 lessons; items marked VERIFY there are not repeated as certain here. Verified against official sources on 2026-11-18.

A printable summary of Module 3. Detail lives in the linked lessons.

Key services

ServiceWhat it is for
AWS SDK for .NET v4Typed clients; credentials from the default chain, never from config files (SDK)
ECS on Fargate + ALBDefault host for a steady HTTP API (Fargate)
Elastic BeanstalkManaged EC2 + ALB + ASG; .NET 10 on Amazon Linux 2023 or Windows/IIS (Beanstalk)
Lambda (dotnet10)Spiky or event-driven code, or a whole ASP.NET Core app (Lambda)
ECS Express ModeFastest container-to-HTTPS path; replaces App Runner for new accounts (hosting)
Amazon ECRPrivate image registry: immutable tags, scan on push, lifecycle (ECR)
Parameter Store / AppConfigSettings, and gradual feature-flag rollout (config)
Secrets ManagerCredentials with rotation and cached reads (secrets)
CognitoToken issuer; your API validates the JWT (auth)
SQS / EventBridge SchedulerQueued and scheduled work (background)
CloudWatch LogsStructured JSON logs, queried with Logs Insights (logging)
AWS Transform for .NETAgentic porting of .NET Framework code (porting)

.NET / Azure to AWS

Microsoft / AzureAWS
IIS on a VMEC2 behind an ALB (Linux + Kestrel is cheaper)
App ServiceElastic Beanstalk or ECS Express Mode
Container Apps / AKSECS on Fargate / EKS
Azure FunctionsLambda
Container RegistryECR
Managed identityIAM role on the task, instance or function
App Configuration / Key VaultParameter Store + AppConfig / Secrets Manager
Application Insights loggingCloudWatch Logs (+ X-Ray, OpenTelemetry)
Worker / WebJobsWorker Service on Fargate, Lambda, EventBridge Scheduler
Upgrade Assistant.NET Upgrade Assistant + AWS Transform for .NET

Most useful CLI commands

TaskCommand
Who am I?aws sts get-caller-identity --profile learn-sandbox
Sign in with SSOaws sso login --profile learn-sandbox
ECR loginaws ecr get-login-password | docker login --username AWS --password-stdin ACCOUNT.dkr.ecr.us-east-1.amazonaws.com
Build and push arm64docker buildx build --platform linux/arm64 -t REPO:TAG --push .
Scan findingsaws ecr describe-image-scan-findings --repository-name NAME --image-id imageTag=TAG
Roll the ECS serviceaws ecs update-service --cluster C --service S --force-new-deployment
Wait for stabilityaws ecs wait services-stable --cluster C --services S
Tail logsaws logs tail LOG_GROUP --follow
Read parametersaws ssm get-parameters-by-path --path /app/ --with-decryption
Read a secretaws secretsmanager get-secret-value --secret-id NAME
Invoke a functionaws lambda invoke --function-name NAME out.json
CDK deploy and destroycdk deploy STACK · cdk destroy STACK
Find leftoversaws resourcegroupstaggingapi get-resources --tag-filters Key=Project,Values=cea-lab

Key limits and defaults

ItemValue
Lambda runtimedotnet10, deprecation scheduled 2028-11-14; dotnet8 deprecated 2026-11-10
Lambda memory / timeout128 to 10,240 MB (CPU scales with memory) / 900 s
Lambda payload, sync6 MB each way
Lambda default concurrency1,000 per Region (soft quota)
ALB health check defaults30 s interval, 5 s timeout, 5 healthy / 2 unhealthy thresholds, path /, code 200
ALB deregistration delay300 s
ECS health check grace periodDefault 0; set it past start-up time
ECS container health check30 s interval, 5 s timeout, 3 retries, startPeriod off
Smallest Fargate task0.25 vCPU with 512 MiB; 20 GB ephemeral storage included
SDK v4 retriesStandard mode; MaxErrorRetry 3 means 1 attempt plus 3 retries
SQS visibility timeoutDefault 30 s, maximum 12 hours; long poll up to 20 s
Parameter Store value size4 KB standard, 8 KB advanced
Secrets ManagerUSD 0.40 per secret per month; staging label AWSCURRENT is the default read
ECR basic scanningAt most one scan per image per 24 h; enhanced (Inspector) rescans continuously

Decision rules

If...Then...
New steady HTTP APIECS on Fargate in private subnets behind an ALB
Spiky, event-driven, short workLambda
Want containers with no infrastructure designECS Express Mode
.NET Framework or IIS app, not ported yetEC2 Windows or Beanstalk Windows; plan a port
Org standard is KubernetesEKS (control plane bills hourly)
Secret, rotatedSecrets Manager with caching; not an environment variable
Plain setting or flagParameter Store or AppConfig
Slow or scheduled workQueue plus worker, or EventBridge Scheduler; not a timer in the web process
Health endpointLiveness with no dependencies; readiness may check the database
Per-call deadlineTighter than SDK defaults for user-facing APIs

Common exam traps

  • Explicit config beats the default chain: a stale key in appsettings overrides the task role.
  • Task role vs execution role: the execution role pulls the image and writes logs; the task role is what your code uses.
  • Singleton clients: register SDK clients as singletons; transient clients exhaust sockets.
  • Beanstalk creates real resources (EC2, ALB, ASG) you can see and pay for.
  • Liveness must not check the database, or a database blip makes ECS replace every task.
  • ALB fails open when all targets are unhealthy, but ECS still replaces the tasks.
  • SnapStart applies to published versions only, not $LATEST.
  • Lifecycle policies can delete an image a running task needs.
  • A JWT must be validated for issuer, token_use, client ID and scopes, not just signature.
  • At-least-once delivery: SQS consumers must be idempotent.

Cost traps

TrapAvoid it by
Idle ALB and NAT gateway after a labTear down the same day; check with the tag query
Fargate left runningcdk destroy; Fargate bills per second with a one-minute minimum
Beanstalk load-balanced environmentUse single-instance for labs, then terminate
Secrets Manager secrets for every settingParameter Store for non-secrets
Enhanced ECR scanning on every repoEnable where the risk justifies it
CloudWatch Logs ingestionLog levels, retention, no payload dumps
EKS or ElastiCache for a labDesign-only; use ECS and local Valkey in Docker

Labs 3.4, 3.5 and the project at a glance

TopicRemember
One binary, three hostsAddAWSLambdaHosting(LambdaEventSource.HttpApi) is a no-op outside Lambda; handler string is the assembly name; publish -r linux-arm64 --self-contained false (Lab 3.4)
Reading a cold startInit Duration in the REPORT line (Logs Insights field @initDuration); SnapStart shows Restore Duration and needs a published version
Config layer orderappsettings, environment, Parameter Store, AppConfig: later wins (Lab 3.5)
Parameter pathThe configured prefix is stripped and / becomes :; the path is also the IAM boundary
Secret cacheSecretsManagerCache calls DescribeSecret and GetSecretValue once per TTL; grant both on one secret ARN
Rotating a pseudonymisation keyChanges every derived reference; carry a key id in the output
Host comparisonMeasure with one probe, one network, several rounds; calculate cost with a stated model and a break-even (project)

Independent learning resource — 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.

On this page