Create Continuous Delivery Pipelines on AWS Using CDK

In this lecture we will:

• Use CDK CLI to create our CDK project

• Deploy our project on AWS

Prerequisite instructions:

1. Creating new AWS account: https://aws.amazon.com/free

2. Installing AWS CLI: https://docs.aws.amazon.com/cli/latest/userguide/install-cliv2.html

3. Creating IAM User (needed to set up local credentials): https://docs.aws.amazon.com/IAM/latest/UserGuide/id_users_create.html

4. Configuring AWS CLI credentials: https://docs.aws.amazon.com/cli/latest/userguide/cli-configure-files.html#cli-configure-files-methods

5. Installing NPM: https://docs.npmjs.com/downloading-and-installing-node-js-and-npm

6. Installing AWS CDK: https://docs.aws.amazon.com/cdk/latest/guide/getting_started.html#getting_started_install

In this lecture we will:

• Learn about CDK app, stack, and construct concepts

• Understand difference between L1, L2, and L3 constructs

• Understand how CDK code is converted into AWS resources

In this lecture we will:

• Learn about typical CDK project code structure

• Understand why CDK dependencies need to be under devDependencies

• Understand how CDK resource tree is created and where the entry point to our application is

In this lecture we will:

• Create new Budget CDK construct that encapsulates L1 CfnBudget construct, exposing only properties that we want to control

• Create new Billing stack with new Budget construct

• Learn how to use CDK and CloudFormation documentation

In this lecture we will:

• Write unit tests for CDK constructs that verify specific properties

• Write snapshot testing for CDK stacks

In this lecture we will:

- Discuss in high level the pipeline that we want to build in this course

- Discuss the first milestone that we are going to accomplish in this section

In this lecture we will:

- Understand what CodePipeline stages and actions are

- Discuss how actions communicate with each other via Artifacts

In this lecture we will:

• Create GitHub token to provide access for CodePipeline

• Create Pipeline with a Source action that reads from our GitHub repo

In this lecture we will:

• Add new 'Build' stage to our pipeline with 'CodeBuild' action that builds our CDK source code

• Learn how to write simple CodeBuild build specification file

• Deploy our pipeline and see it in action

In this lecture we will:

• Add new "Pipeline Update" stage to our pipeline that deploys pipeline template via CloudFormation

• Deploy changes and verify that everything works

In this lecture we will:

• Add unit test to the CDK build action and observe what happens when build fails

• Understand why storing build-specs in a separate file is better than storing them inline in CDK

• Fix pipeline CDK unit test and unblock the pipeline

In this lecture we will:

• Use jest-junit to generate junit XML test reports for jest

• Configure our build specs to extract jest test reports to make user friendly reporting in CodeBuild

In this lecture we will review our objectives for the section and go over changes that we will make to our pipeline

In this lecture we will:

• Create a new node project and write a basic express server

• Wrap this server using serverless-express helper to make it usable within lambda

• Test that our server is working locally

• Push changes to a new GitHub repository

In this lecture we will create a new service stack that contains:

• Lambda function

• API-Gateway HTTP API with Proxy integration that redirects all requests to Lambda

  
  

In this lecture we will update our pipeline to add a new source action so that it listens to changes in the service code repository

In this lecture we will:

• Add service build action to the build stage of our pipeline

• Create build specs for our service code inside service package

• Learn about restartExecutionOnUpdate property of Pipeline CDK construct that removes the need to manually restart pipeline when its changed

In this lecture we will:

• Refactor our PipelineStack to support adding multiple service stages

• Learn how to override CloudFormation input parameters when deploying service stack

• Add production stage to our pipeline

• Deploy service and verify that it works

In this lecture we will:

• Add test for the new addServiceStage method in our PipelineStack

• Learn how to write more fine-grained tests using fuzzy property matching with CDK assert library

In this lecture we will:

• Add new method to our PipelineStack for adding a billing stack to a pipeline stage

• Add unit tests for the new method

• Use the new method to add Billing stack deployment to our production stage

In this lecture we will review our current pipeline state and discuss what we want to achieve by the end of this section

In this lecture we will:

• Modify ServiceStack class to support creating multiple instances without causing name clashes

• Add a new "Test" stage to our pipeline which deploys another instance of ServiceStack

In this lecture we will write a basic integration test using axios library to call our service endpoint and verify that it returns expected response.

In this lecture we will:

• Update our pipeline to create integration test action inside the test stage

• Learn how we can pass service endpoint from the service stack to the pipeline stack without creating backwards dependency

In this lecture we will learn how to send emails on failures in pipeline by reacting to state changes in CodePipeline at action, stage, or pipeline levels. We will also learn how to filter out events using EventBridge rules and forward them to emails via SNS topics.

In this lecture we will discuss why gradual deployments and canaries are useful, and what we want to achieve in this section

In this lecture we will update our CDK to the latest version, as 1.99 has a bug, which was fixed in later versions (https://github.com/aws/aws-cdk/issues/14503) that would have prevented us from creating alarm on API-G errors.

In this lecture we will learn how to use CodeDeploy deployment groups and Lambda function aliases to create gradual deployments with weighted traffic routing.

In this lecture we will:

• Configure auto-rollback on our lambda deployment group based on attached alarms

• Attach service error alarm to our deployment group

• Test our rollback by pushing bad commit to our service code

In this lecture we will discuss why we need Canaries and how we can use CloudWatch Synthetics service to create canaries that generate synthetic traffic to our service endpoints.

In this lecture we will learn how to write a basic canary lambda that validates that API returns successful (200) response using built-in Synthetics library that comes as part of CloudWatch Synthetics runtime environment.

For more details on what helpers are available for CloudWatch Synthetics canaries see links attached to this lecture

In this lecture we will create a reusable CDK construct called ServiceHealthCanary that creates a Canary, which runs every minute for 5 minutes, after each deployment.

In this lecture we will write a unit test for our ServiceHealthCanary construct that will validate that we create correct CloudFormation resource and that api endpoint environment variable is passed correctly. See attached resources for a link to the CloudFormation page for AWS::Synthetics::Canary resource

In this lecture we will:

• Use our newly created ServiceHealthCanary construct inside Service stack to create a new canary

• Deploy our changes and validate that canary is running as expected

In this lecture we will add alarm that will trigger if canary fails and will send us an email notification

In this lecture we will discuss why multi-region deployments are needed, how they work, and what we want to achieve in this section

In this lecture we will:

• Explicitly specify account and region when creating stacks

• Explicitly specify account and region in CodePipeline actions

• Bootstrap new environment using CDK

• Create cross-region support stack using CDK

In this lecture we will discuss how cross-account deployment pipelines work and what the benefits of separating stages into different accounts are

In this lecture we will:

• Bootstrap second account using new style cdk bootstrap

• Add new stage to our pipeline that deploys service to another account

• Deploy the changes and validate that cross-account deployments succeed

In this lecture we will:

• Discuss why we need to migrate to CDK v2

• Go through differences between CDK v1 and v2

In this lecture we will:

• Update our dependencies in package.json

• Update cdk.json configuration file

• Fix imports in our source code

In this lecture we will:

• Fix imports in unit tests

• Learn how to use Template and Match utilities in CDK unit tests

• Complete CDK v2 migration

In this lecture we will learn how to delete all AWS resources that were created as part of this course

Congratulations for reaching the end of this course! We've covered a lot of topics, and I hope that you will be able to apply at least some of what you learned in your work. CDK is evolving very fast, and they keep releasing new features and high level constructs all the time. Even in the few months that it took me to complete this course, CDK went from version 1.99 to 1.111. CDK version 2 is already in beta and new CDKPipelines construct, which is supposed to significantly improve pipeline management, is already in developer preview.

Moreover, as it's open sourced, we are seeing more and more 3-rd party solutions. I wouldn't be surprized if by the time you're watching this video, there are significant improvements to creating continuous delivery pipelines on CDK. Nevertheless, I think that what you've learned in this course will remain useful, as it teaches some of the fundamentals around how CDK and AWS services like CodePipeline and CodeBuild work. My hope is that as new versions and improvements emerge, you will have enough foundational knowledge to understand whether and how these changes are applicable to your use cases.

Thank you for taking your time to watch this course and congratulations again for getting to the end! Happy building!