Deployment Automation

Example Google Deployment Pipeline

Concepts

• Containers vs VMs vs Bare metal
• Infrastructure as Code
• Artifacts
• Workflows
• Blue/Green

Containers vs VMs vs Bare metal

• Containers: new, lite and provider agnostic
• VMs: familiar, tested and provider agnostic
• Bare metal: for when you need all the power you can get

Infrastructure as code

• Configuration files describe servers, networks, etc
• Keep config with your code in git
• Avoid manually modifying servers
• Small changes rather than big batches

Infrastructure as code

Artifacts

• Executable files
• Compressed resources
• Images, Fonts, etc

Deployment Workflows

• Manual: login to the server, upload the artifacts, start the application
• Git Push: a git hook on a specific remote triggers the deployment
• Through CI: service monitors repository and triggers a deployment after a successful build
• Provider Tool: most hosting environments provide specific tools for deployment

Tools

• FTP & Rsync
• Git & SSH
• Ansible & Terraform
• Docker swarm & Kubernetes
• Provider specific options

Example AWS Cloudformation

  {
    "Resources" : {
      "FileBucket" : {
        "Type" : "AWS::S3::Bucket",
        "Properties" : {
          "AccessControl" : "PublicRead"
        }
      },
      "Ec2Instance" : {
        "Type" : "AWS::EC2::Instance",
        "Properties" : {
          "SecurityGroups" : [ { "Ref" : "InstanceSecurityGroup" } ],
          "ImageId" : "ami-7a11e213"
        }
      },
      "InstanceSecurityGroup" : {
        "Type" : "AWS::EC2::SecurityGroup",
        "Properties" : { "...": "..." }
      }
    }
  }
  

Example AWS Cloudformation

Example git post-recieve hook

    # hooks/post-recieve
    GIT_WORK_TREE=~/src-full-stack git checkout master -f &&
    cd ~/src-full-stack &&
    npm install &&
    npm run build &&
    echo "Deployed successfully"
  

Example kubernetes config

  apiVersion: extensions/v1beta1
  kind: Deployment
  metadata:
    name: nginx-deployment
  spec:
    replicas: 2
    template:
      metadata:
        labels:
          app: nginx
      spec:
        containers:
        - name: nginx
          image: nginx:1.8
          ports:
          - containerPort: 80
    

Blue/Green deployment

Blue/Green deployment: Process overview

1. A new deployment is triggered
2. A new environment is setup (green)
3. Old version continues serving all requests (blue)
4. Small amount of traffic is given to the new version (green)
5. New version (green) is monitored against predefined metrics
6. If the new version performs well, continue with next server, else rollback

Example Ansible deployment diagram

Benefits

• Repeatable & reliable
• Easier management
• Testable systems and processes
• Self-documenting systems and processes

Links:

• Ansible (utility)
• Terraform (utility)
• Deploying through Travis with RSync
• Deploying PHP via Ansible
• Martin Fowler on Blue/Green deployment
• Martin Folwer - Infrastructure as code (youtube)
• Infrastructure as code (youtube)
• Continuous deployment on Azure
• Continuous deployment on Google
• Continuous deployment on AWS

Course Schedule: