AWS CDK

LINKS

• AWS CDK Official: cdk docs
• Teacher Repository: cdk alexhddev

PRE REQUISITES

• NodeJS + npm
• AWS CDK - As global npm dependency
• A text editor: VS Code
• AWS CLI
• Optional: Python + PIP
• Optional: VSCode AWS toolkit extension

Installing NodeJS

Ver la documentación Oficial

# Descarga e instala nvm:
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.3/install.sh | bash

# en lugar de reiniciar la shell
\. "$HOME/.nvm/nvm.sh"

# Descarga e instala Node.js:
nvm install 22

# Verify the Node.js version:
node -v # Should print "v22.20.0".

# Verifica versión de npm:
npm -v # Debería mostrar "10.9.3".

Installing CDK

sudo npm i -g aws-cdk
cdk --version

IAM User

• New IAM User as Administrator
• Create an access Key

# Configure and Authenticate
aws configure
# Testing connection
aws s3 ls

CDK AND CLOUDFORMATION INTRO

AWS CloudFormation

• Service behind CDK
• Infrastructure as Code
• Templates JSON / YAML
• Stack: Collection of AWS resources managed as a single Unit
• Stack Operations: Create, update, delete
• Functions: get, ref, import… (Intrinsic)

CloudFormation Template ---------> Stack

CDK Intro

Initialize in an empty directory

cdk init app --language python

For Python we can start creating our virtual environment

python3 -m venv .venv
# Then Activate that venv
source .venv/bin/activate
# Install requirements
pip install -r requirements.txt

Finally the synthesis step

cdk synth

CDK Environment

• Required step for deployment: Bootstrap CDK Toolkit ----> IAM role for CDK CDK Toolkit ---> S3 Bucket

# Create the resources on AWS
cdk bootstrap [Account Number]/[Region]
# Deploy
cdk deploy

CDK Code

• How does CDK knows where to look for CDKCode: cdk.json
• cdk.out: Stores the output for the CloudFormation Template

CDK Stacks: Abstraction over CloudFormation Stacks Extend the Stack class (object)

class PyStarterStack(Stack):

    def __init__(self, scope: Construct, construct_id: str, **kwargs) -> None:
        super().__init__(scope, construct_id, **kwargs)

• Stack properties:
  • Scope (app)
  • id
  • Optional properties

CDK Constructs:

• Abstraction over AWS resources
• Mos AWS services have their own constructs
• Properties:
  • Scope (stack)
  • id
  • Required/optional properties

Levels of Constructs:

• L1: Low level constructs - cfn (CloudFormation) resources. We must configure all required properties (For new services that are still not migrated)
• L2: AWS resources with a higher-level. CDK provides additional functionality like defaults, boiler plate and type safety for many parameters (Used most of the time)
• L3: Patterns. Combine multiple types of resources and help with common tasks in AWS. Example LambdaRestApi (Matter of preference)

CDK Commands

CDK INTERMEDIATE

CDK IDs

• Construct IDs: Ourselves

bucket = s3.Bucket(self, "PyBucket",...)

• Logical IDs: Construct ID + Suffix from CDK

"Resources": {
  "PyBucket5C1531F2": {
    "Type": "AWS::S3::Bucket",
    "Properties": {
    }
  }
}

• Physical ID: Resource name. Used for reference inside other services (ARN) pystarterstack-pybucket5c1531f2-1rjjdg.....

How CloudFormation tracks resources:

• By the Logical ID
• When Logical ID changes:
  • Old resource is removed
  • Creates a new one
• Potential problems:
  • Accidental deletion of stateful resources (S3, DynamoDB)
  • Resource already exists error

CloudFormation Intrinsic Functions

Key to infrastructure as code Class Fn And can be found on the file PyStarterStack.template.json

Example: arn:aws:cloudformation:us-west-2:123456789012:stack/mystack-mynestedstack-sggfrhxhum7w/f449b250-b969-11e0-a185-5081d0136786

from aws_cdk import (
  Fn #CloudFromation Intrinsic functions
)

# Create a Private Method receive self and return string using Intrinsic functions
    def __initialize_suffix(self):
        short_stack_id = Fn.select(2, Fn.split('/', self.stack_id))
        suffix = Fn.select(4, Fn.split('-', short_stack_id))
        return suffix

Note: When using self we mean that it is this same stack

Handling Multiple Stacks

• Due to complexity
• Some resources may be sensitive
• Some can take a long time to deploy
• Solution: Cross stack references, shared resources

Recommendations:

• Separate stacks for resources with state
• Separate stacks for IAM roles, policies, secrets
• Separate stacks for resource with complex initialization (VPCs, DNS)
• AWS Recommendation: Organize stacks by life cycle and ownership

Cross-stack references:

• CF intrinsic function: Fn:ImportValue

SERVERLESS REST API

Pre Requisites:

• We need a rest client: REST Client from VS Code (Extensions)

AWS Lambda and CDK

There are 3 levels of difficulty we can face:

• aws-cdk-lib.aws_lambda module: Generic: Function
  • CDK and lambda code can be written in different languages
  • Lambda runtime contain AWS SDK. No need to bundle
  • Other dependencies: Building required. CDK uses Docker
• TypeScript: NodeJsFunction
  • Requires esbuild packages. If not present it uses Docker
• Python: aws_lambda_python_alpha - If other dependencies are required
  • Still experimental
• Other languages: Code.fromAsset

AWS Lambda Architecture: Multiple, Monolithic

Multiple

Monolithic

The AWS SDK

• Interacts with AWS
  • Console: Using UI
  • CLI: Using commands aws s3 ls
  • Code: Using AWS SDKs JavaScript, Python, etc
    • Code can be run from EC2 instances
    • Lambda
    • Other AWS services
    • Our computer:
      • JavaScript: aws-sdk-js-v3
      • Python: boto3

The AWS SDK - DynamoDB

• Write queries for DynamoDB (putltem, getltem)
• Run queries from AWS lambda
• Lambda needs to have access to the table
  • table name
  • IAM role for different operations
• Required:
  • DynamoDB client - Outside the Lambda body
• Use client to make request

# Use it outside the Handler
table_name = os.environ.get("TABLE_NAME")
dynamodb = boto3.resource("dynamodb")
table = dynamodb.Table(table_name)

Documentation: aws dynamodb We need to install:

pip install boto3

TESTING REST CLIENT

We can use that extension:

# requests.http
# We get the url from the Deploy result during Lambda creation

@url = https://covr3k2a3h.execute-api.eu-west-1.amazonaws.com/prod

GET {{url}}/empl?id=kjasdnchfg

POST {{url}}/empl
content-type: application/json

{
  "name": "John",
  "possition": "engineer"
}
#Output: It will return the ID of the Item