Split a monolith

In the previous tutorial we did some basic canary releasing on two versions of a monolithic application. Very nice, but Vamp isn’t called the Very Awesome Microservices Platform for nothing. The next step is to split our monolithic Sava application into separate services. In this tutorial we will:

• define a new service topology
• learn about Vamp environment variables and service discovery

Requirements: Docker machine should have access to at least 3GB memory

Define a new service topology

To prove our point, we are going to slightly “over-engineer” our services solution. This will also help us demonstrate how we can later remove parts of our solution using Vamp. For now, we’ll split the monolith into a topology of one frontend and two separate backend services. After our engineers are done with coding, we can catch this new topology in the following blueprint. Please notice a couple of things:

• We now have three clusters: sava, backend1 and backend2. Each cluster can have multiple services for canary release and traffic filtering
• The sava cluster has explicit dependencies on the two backends. Vamp will make sure these dependencies are checked and rolled out in the right order
• We have used environment variables to connect the dynamically assigned ports and hostnames of the backend services to the “customer facing” sava service
• The gateway port has been changed to 9060 so it doesn’t collide with the monolithic deployment

name: sava:1.2
gateways:
  9060: sava/webport
clusters:
  sava:  # cluster 1
    services:
      breed:
        name: sava-frontend:1.2.0
        deployable: magneticio/sava-frontend:1.2.0
        ports:
          webport: 8080/http
        environment_variables:
          BACKEND_1: http://$backend1.host:$backend1.ports.webport/api/message
          BACKEND_2: http://$backend2.host:$backend2.ports.webport/api/message
        dependencies: # Vamp will check dependencies are available
          backend1: sava-backend1:1.2.0
          backend2: sava-backend2:1.2.0
      scale:
        cpu: 0.2
        memory: 64MB
        instances: 1
      health_checks:
        initial_delay: 10s
        port: webport
        timeout: 5s
        interval: 10s
        failures: 10
  backend1:  # cluster 2
    services:
      breed:
        name: sava-backend1:1.2.0
        deployable: magneticio/sava-backend1:1.2.0
        ports:
          webport: 8080/http
      scale:
        cpu: 0.2
        memory: 64MB
        instances: 1
  backend2:  # cluster 3
    services:
      breed:
        name: sava-backend2:1.2.0
        deployable: magneticio/sava-backend2:1.2.0
        ports:
          webport: 8080/http
      scale:
        cpu: 0.2
        memory: 64MB
        instances: 1

Deploy this blueprint using either the UI or a REST call - let’s name it after the blueprint this time sava-new. Once it’s deployed, you can check out the new topology in the Vamp UI through the Gateways page or the Deployments page

You can also use Vamp as a reverse proxy to access the exposed gateways:

• http://localhost:8080/proxy/gateways/sava-new%2F9060/
• http://localhost:8080/proxy/gateways/sava-new%2Fsava%2Fwebport/

Learn about environment variables and service discovery

If you were to check out the Docker containers using docker inspect, you would see the environment variables that we set in the blueprint.

> docker inspect 66e64bc1c8ca
...
"Env": [
    "BACKEND_1=http://192.168.65.2:33021/api/message",
    "BACKEND_2=http://192.168.65.2:33022/api/message",
    "PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin"
],
...

Host names and ports are configured at runtime and injected in the right parts of your running deployment. Your service/app should pick up these variables to configure itself. Luckily, this is quite easy and common in almost all languages and frameworks.

Remember, there is no “point-to-point” wiring. The exposed host and port are actually service endpoints. The location, amount and version of containers running behind that service endpoint can vary. Learn more about how Vamp does service discovery.