Skip to main content
In a CI/CD pipeline, AWS CodePipeline orchestrates the workflow from source to build and test. The final step—deployment—is implemented by AWS CodeDeploy. This fully managed service automates code delivery and updates across multiple compute platforms.

​
What Is AWS CodeDeploy?

AWS CodeDeploy deploys application revisions (code, scripts, web assets, multimedia) to these targets:
  • Amazon EC2 instances
  • On-premises servers
  • AWS Lambda functions
  • Amazon ECS services
  • AWS Fargate tasks
The image shows a list of deployment targets for AWS CodeDeploy, including Amazon EC2 instances, on-premise instances, Lambda, and ECS, alongside the AWS CodeDeploy logo.
At the source stage of your pipeline, you can store your application bundle in:
  • Amazon S3
  • AWS CodeCommit
  • GitHub
  • Bitbucket
The image shows a DevOps lifecycle diagram with stages like Source, Build, Test, Deploy, and tools such as AWS S3, AWS CodeCommit, GitHub, and Bitbucket.
As a fully managed service, CodeDeploy handles infrastructure provisioning, scaling, and maintenance. Key benefits include:
  • Automatic scaling from a single instance to tens of thousands
  • Built-in deployment strategies to minimize downtime
  • Monitoring and rollback capabilities via AWS Console, CLI, or SDKs
The image is a promotional graphic for AWS CodeDeploy, highlighting features such as being fully managed, scalable, and minimizing downtime.

​
Deployment Strategies

AWS CodeDeploy offers four primary strategies:
StrategyDescriptionBest For
In-PlaceUpdate the application on existing instances directly.Simple updates; minimal infrastructure changes.
RollingDeploy to a subset of instances in batches, allowing validation per batch.When gradual rollout and testing are required.
ImmutableLaunch new instances with the updated application, then swap over traffic.Zero-downtime requirements; high reliability.
Blue/GreenMaintain two environments (“blue” and “green”) and switch traffic between them.Instant rollback and seamless cutover.
The image shows three types of deployment: In Place, Rolling, and Immutable, each represented by a distinct icon.

​
Blue/Green Deployments

In a blue/green deployment, you maintain:
  • Blue: The active production environment.
  • Green: A staging environment with the new release.
After validating the green environment, you shift traffic using DNS, load balancers, or Auto Scaling. If any issues occur, switch back to blue instantly. Otherwise, complete the traffic migration to green.
The image depicts a blue-green deployment architecture using AWS services, including EC2 instances, Route 53 DNS, Elastic Load Balancing, and Auto Scaling.

​
Traffic Shifting Configurations

ConfigurationBehavior
CanaryShift a small percentage of traffic to green, validate, then shift rest.
LinearShift equal percentages at regular intervals until complete.
All-at-onceMove 100% of traffic to green in a single step.
The image illustrates three types of blue/green deployment methods: Canary, Linear, and All-at-once, each represented by a distinct icon.
You can integrate health checks or manual approval actions before, during, or after traffic shifts to ensure deployment safety.

​
Managing Deployments with AppSpec

To coordinate deployments, CodeDeploy uses:
  1. CodeDeploy Agent
    Installed on each target to execute deployment tasks.
Ensure the CodeDeploy Agent runs with sufficient IAM permissions and is kept up to date.
  1. AppSpec File (appspec.yml)
    A YAML manifest that maps source files to destination paths and defines lifecycle event hooks.
version: 0.0
os: linux
files:
  - source: /web/build/
    destination: /var/www/html/
hooks:
  BeforeInstall:
    - location: scripts/backup.sh
      timeout: 300
  AfterInstall:
    - location: scripts/configure.sh
  ApplicationStart:
    - location: scripts/start_server.sh
The image shows a split design with "Appspec.yml" on the left and "AWS CodeDeploy" on the right, featuring their respective icons.

​
References

Watch Video