waldofe / microgram

A instagram-like application split into microservices.

microgram

A basic instagram-like application split into microservices.

The project uses Docker and Kubernetes.

Setup

Follow the instructions for each repo to run all services within Docker containers.

• https://github.com/oswaldoferreira/microgram-auth
• https://github.com/oswaldoferreira/microgram-feed
• https://github.com/oswaldoferreira/microgram-fe

By the end you should have the following containers running locally:

› docker ps
CONTAINER ID        IMAGE                   COMMAND                  CREATED              STATUS              PORTS                    NAMES
ad7bad56e9ca        microgram-feed:latest   "docker-entrypoint.s…"   About a minute ago   Up About a minute   0.0.0.0:8080->8080/tcp   loving_turing
26530617d31c        microgram-auth:latest   "docker-entrypoint.s…"   About a minute ago   Up About a minute   0.0.0.0:5000->5000/tcp   sad_swanson
9aa6b98d7bb7        microgram-fe:latest     "/docker-entrypoint.…"   About an hour ago    Up About an hour    0.0.0.0:4200->80/tcp     hungry_lumiere

The microgram-fe (SPA) can be accessed at port 4200.

› curl -I localhost:4200
HTTP/1.1 200 OK
Server: nginx/1.19.3
Date: Fri, 30 Oct 2020 01:34:24 GMT
Content-Type: text/html
Content-Length: 995
Last-Modified: Thu, 29 Oct 2020 20:52:14 GMT
Connection: keep-alive
ETag: "5f9b2b7e-3e3"
Accept-Ranges: bytes

Deployment

The repositories above fall short when describing the processes to setup a Kubernetes cluster, but, it provides the YML files that you'll need once you get it running.

I recommend taking a further look at GKE and EKS solutions.

Using this project, you'll endup with the following Kubernetes pod setup:

› kubectl get pods
NAME                              READY   STATUS    RESTARTS   AGE
microgram-auth-66cc68b8dd-wbqk7   1/1     Running   0          51m
microgram-fe-5f6fc7b9cf-qtz5n     1/1     Running   0          51m
microgram-feed-77c566bd87-228b7   1/1     Running   0          51m
microgram-feed-77c566bd87-m9brc   1/1     Running   0          51m

And the following K8s services:

› kubectl get services
NAME                    TYPE           CLUSTER-IP       EXTERNAL-IP                                                               PORT(S)        AGE
kubernetes              ClusterIP      10.100.0.1       <none>                                                                    443/TCP        5d10h
microgram-auth-lb-svc   LoadBalancer   10.100.158.170   a04c14c0e3b1843ae9c0ada419302b39-2052811953.sa-east-1.elb.amazonaws.com   80:30655/TCP   26h
microgram-fe-svc        LoadBalancer   10.100.213.94    a12e954cc71d24037bc1361adb33a804-1927388553.sa-east-1.elb.amazonaws.com   80:30804/TCP   3d2h
microgram-feed-lb-svc   LoadBalancer   10.100.27.58     a3af9e4b6e2094c759fae942e4781438-1541706164.sa-east-1.elb.amazonaws.com   80:32341/TCP   5d9h

Architecture

The proposed architecture can easily change, but that's what I went for on Amazon as a playground:

1. Request hits a AWS load balancer
2. Nginx serves the static content for microgram-fe (more machines could be added)
3. AWS API Gateway receives the request and decides which backend API to go to
4. Both microgram-feed and microgram-auth services sit behind AWS load balancers
5. microgram-feed depends on a Postgres instance and Amazon S3 for media storage
6. microgram-auth depends on Postgres for authentication