Run a Microservice Locally

Before the advent of microservice architecture, development teams built, deployed and ran the whole application as one large chunk of software. To test a small change in their module not merely by unit testing, the developers had to build the whole application. Therefore the builds took large amount of time. After the build, the developers deployed their version of the application into a test server. The developers ran the server either on a remote machine, or on their local computer. In the latter case, the developers had to install and operate a rather complex environment on their local computer.

In the era of microservice architecture, the developers write, build, test and run small software services. Builds are fast. With modern frameworks like Node.js there is no need to install and operate complex server environments to test a single service, since the service runs as a regular process. You do not have to deploy your service to some environment to merely test it, so you just build your service and run it immediately on your local computer.

This module covers the different aspects involved in developing a single service on a local machine. You don’t need to write code though. Instead, you build, run, and test an existing service: ratings.

The ratings service is a small web app written in Node.js that can run on its own. It performs similar actions to those of other web apps:

  • Listen to the port it receives as a parameter.
  • Expect HTTP GET requests on the /ratings/{productID} path and return the ratings of the product matching the value the client specifies for productID.
  • Expect HTTP POST requests on the /ratings/{productID} path and update the ratings of the product matching the value you specify for productID.

Follow these steps to download the code of the app, install its dependencies, and run it locally:

  1. Download the service’s code and the package file into a separate directory:

    $ mkdir ratings
    $ cd ratings
    $ curl -s https://raw.githubusercontent.com/istio/istio/release-1.8/samples/bookinfo/src/ratings/ratings.js -o ratings.js
    $ curl -s https://raw.githubusercontent.com/istio/istio/release-1.8/samples/bookinfo/src/ratings/package.json -o package.json
    
  2. Skim the service’s code and note the following elements:

    • The web server’s features:
      • listening to a port
      • handling requests and responses
    • The aspects related to HTTP:
      • headers
      • path
      • status code
  3. Node.js applications are written in JavaScript, which means that there is no explicit compilation step. Instead, they use just-in-time compilation. To build a Node.js application, then means to install its dependencies. Install the dependencies of the ratings service in the same folder where you stored the service code and the package file:

    $ npm install
    npm notice created a lockfile as package-lock.json. You should commit this file.
    npm WARN ratings No description
    npm WARN ratings No repository field.
    npm WARN ratings No license field.
    
    added 24 packages in 2.094s
    
  4. Run the service, passing 9080 as a parameter. The application then listens on port 9080.

    $ npm start 9080
    > @ start /tmp/ratings
    > node ratings.js "9080"
    Server listening on: http://0.0.0.0:9080
    
  1. Open http://localhost:9080/ratings/7 in your browser or access ratings using the curl command from a different terminal window:

    $ curl localhost:9080/ratings/7
    {"id":7,"ratings":{"Reviewer1":5,"Reviewer2":4}}
    
  2. Use the POST method of the curl command to set the ratings for the product to 1:

    $ curl -X POST localhost:9080/ratings/7 -d '{"Reviewer1":1,"Reviewer2":1}'
    {"id":7,"ratings":{"Reviewer1":1,"Reviewer2":1}}
    
  3. Check the updated ratings:

    $ curl localhost:9080/ratings/7
    {"id":7,"ratings":{"Reviewer1":1,"Reviewer2":1}}
    
  4. Use Ctrl-C in the terminal running the service to stop it.

Congratulations, you can now build, test, and run a service on your local computer!

You are ready to package the service into a container.

Was this information useful?
Do you have any suggestions for improvement?

Thanks for your feedback!