What is not a microservice?

Async operations with queue (and other) systems

Having 50 different queue consumers (for notifications, logging, reconciliation services, e-mailing etc) doesn’t mean you have 50 microservices

Multiple programs running on a single machine and communicating over HTTP

dashersw

What is a Microservice?

dashersw

If you are breaking down the fulfillment of a client request into multiple collaborating services that run in their own memory space, then you are doing microservices.

THE BASELINE

Microservices

The quality of your approach depends on how well you apply Domain Driven Design and other best practices.

You have to —

Embrace change

Give autonomy to teams, to services

Automate testing, configuration, deployment, monitoring and more

Favor cattle over pets

dashersw

An example e-commerce app

dashersw

Auto-discovery

dashersw

Auto-discovery

dashersw

Zero-configuration

dashersw

Highly-redundant

dashersw

Fault-tolerant

dashersw

Self-healing

dashersw

An example in Node.js

const cote = require('cote'),
    models = require('./models');

const userResponder = new cote.Responder({ name: 'user responder' });

userResponder.on('list', req => models.User.find(req.query));
userResponder.on('get', req => models.User.get(req.id));
userResponder.on('create', req => models.User.create(req.user));

const cote = require('cote'),
    router = express.Router();

const userRequester = new cote.Requester({ name: 'user requester' });

router.get('/users/:id', (req, res, next) => {
    userRequester.send({type: 'get', id: req.params.id})
        .then(user => res.send(user))
        .catch(next);
});

router.get('/users', (req, res, next) => {
    userRequester.send({type: 'list'})
        .then(users => res.send(users))
        .catch(next);
});

module.exports = router;

cote

https://github.com/dashersw/cote

• Zero dependency: Microservices with only Node.js
• Zero-configuration: no IP addresses, ports or routing to manage
• Decentralized: No fixed parts, no "manager" nodes, no single point of failure
• Auto-discovery: Discovery without a central bookkeeper
• Fault-tolerant: Don't lose any requests when a service is down
• Scalable: Horizontally scale to any number of machines
• Performant: Process thousands of messages per second
• Humanized API: Simple to get started with a reasonable API

dashersw