Skip to content

Multi tenant SaaS starter kit with cqrs graphql microservice architecture, apollo federation, event source and authentication

License

Notifications You must be signed in to change notification settings

jozefbalun/ultimate-backend

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ULTIMATE BACKEND

(WIP): This is an enterprise scale advanced microservice pattern with GraphQL API and GRPC Microservices, based on Domain (DDD) using the command query responsibility segregation (CQRS) design pattern.

License Snyk Code Size Version Top Language Top Language

Description

This should be the go to backend base for your next scalable project. This is a proof of concept project designed to be extremely slim and scalable, with distributed data request and process handling, built from the ground up for production use. It comes with Multi-Tenancy SaaS support, following different multi-tenancy database strategy as well as different resolver patterns
to identify your tenants. The goal is to give your next big project that extra leap to awesomeness. To get started read the instructions below.

Note: Seeing alot of clone of the project which is good, but please if you can 🌟 the project as it also motivates me in improving the project. Also the docker azure CI setups is broken and will be fixed soon.

Note: Also ultimate backend is coming to rust as a complete microservice framework, if you want to be part of it and you program in rust, please write to me. Here is the repo ultimate

Features

Software features

  • âś… CQRS
  • âś… Software as a Service
  • âś… Authentication by stateful session (Password) GraphQL
  • âś… OAuth2 Authentication (Google, Github, Facebook) REST
  • âś… User
  • âś… Event Sourcing
  • âś… GraphQL API
  • âś… GRPC Microservice
  • âś… Emailing Queue
  • âś… Role Based Access Control
  • âś… Multi Tenancy
  • âś… Payment (Stripe)
  • âś… SaaS Plans (Stripe)
  • âś… Security
  • âś… Service Discovery (Default on Consul), supports ectd, Kubernetes
  • âś… React SSR Starter Kit
  • ❌ (WiP) Documentation
  • ❌ (WiP) Webhooks
  • ❌ (WiP) Support for language translation
  • ❌ (WiP) GraphQL dataloaders

Software stack

Required Optional
Store and cache Event Store (Event Source Store), Memcached (cache), Redis (Queue) and MongoDB (Database) ArangoDB (Database)
Stack and frameworks NestJS (Server Framework), NodeJS (System runtime), Typescript, Express JS, Fastify, GRPC, NestCloud and Apollo GraphQL none
Deployment and containerization Docker Kubernetes, Azure Pipeline, GitLab CI
Service Registry Consul Kubernetes and etcd

Folder Structure

Senior candidate in the folder structure is the /app folder. This folder contains all executable programs or in this case microservices

  • /app/service-access The access microservice handles access token management for each tenant in the system. It also validates current tenant credentials against a tenant specific resources. It is the gate keeper for your tenant api.
  • /app/service-account The account microservice handles user account commands and queries such as creating and validating new and current users in the system.
  • /app/service-tenant The tenant microservice handles creating new tenants as well as managing tenant specific commands and queries.
  • /app/service-role Similar to service-access, this service validates users based on roles against the entire system and tenants they belong to. This service handles only user authorization, where service-access is for tenant access keys for external api.
  • /app/service-notification The microservice performs notification tasks in the entire systems. Right now it supports emailing, but can be extended to support push notification as well as activity feeds.
  • /app/service-billing The billing microservice manages billing and payment commands. It is tightly integrated with Stripe, but can be easily replaced.
  • /app/service-project This microservice is merely an example of supporting multi-tenant database strategy. It holds no other significance.

The next important folder is the /lib folder. This folder contains all internal libraries that is depended upon by the microservices, and they are,

  • /lib/common House shared/common modules in the project.
  • /lib/core This is the core module of ultimate backend and houses most of the codebase that wires everything together.
  • /lib/repo-orm This library will be extracted in the future as it matures. It is aimed at being a simple, less cluttered NoSQL multi-tenant object relational mapper (ORM), It supports caching, MongoDB and ArangoDB, with support for FaunaDB in the works.
  • /lib/contracts Shared contracts for both typescript typings and graphql code first types, are stored here.
  • /lib/repository It holds all repositories used in the microservices, and this repository and created from the repo-rom library.
  • /lib/proto-schema All GRPC protobuf files required by each service, is stored in here and shared amongst services that depends on them.

Other folders such as iac which simply means Infrastructure as Code is teraform setup for creating a kubernetes cluster with just a simple command for running the microservices. scripts contains helper bash scripts for CI/CD.

Installation

$ yarn install

Configuration

Before starting the services, please create the appropriate consul (Default service registry Consul) kv store config for all the services. You can find the example config
in the folders of each service called config.example. The consul config key of say the account service should be
ultimatebackend/config/io.ultimatebackend.srv.account and just paste the config.yaml content in the consul store for that key in yaml and save.
You will need to set your sendgrid api key, so the backend can send emails on signup etc. If using stripe for payments you'll also need to put your public and private keys there too.
You can opt in for etcd or kubernetes as service registry.

Usage

With Docker locally

$ docker-compose --project-directory=. -f docker-compose.dev.yml up --build

Note: I've seen some issues with consul docker image and so would recommend setting up consul manually before running this command

Without Docker locally

Consul, Mongodb, redis, memcached, and eventstore all need to be started first as our microservices need to connect to them.

Start consul locally

consul agent --dev  

For help installing consul on your local machine, visit Consul Website

Start mongodb locally

mongod  

If you have docker installed

docker run -d -p 27017:27017 mongo  
docker run -d -p 1113:1113 -p 2113:2113 eventstore/eventstore  
docker run -d -p 6379:6379 redis  

Otherwise, you can install and run redis and eventstore locally if you choose.

Running the microservices

You should start the microservices in any other. Example

  
# Generate protobuf typescript definitions and please fix the path for timestamps import (You should do this only when you update the protobuf files)  
# in `lib/proto-schem`  
$ sh ./gen-ts.sh  
  
# Now build the proto-scheme lib (You should do this only when you update the protobuf files)  
$ nest build proto-schema  
  
# Start the account service  
$ npx nest start service-account  
  
# Start the access service  
$ npx nest start service-access  
  
# Start the role service  
$ npx nest start service-role  
  
# Start the graphql api  
$ npx nest start api-admin  
  

Note: You don't need all services running plus you can start them in any order.

Alternative method of running services

If you find the nest cli using too much memory running each service, you can build them first and then run them:

Build each service:

npx nest build service-account  
npx nest build service-notification  
npx nest build service-billing  
npx nest build service-project  
npx nest build service-tenant  
npx nest build service-access  
npx nest build service-role  

Each service is built and written into dist/apps directory from where you can directly run each service with nodejs. Running each service with npx nest start uses three orders of magnitude more memory than this method so you will use a lot less memory!

Run each service in a separate terminal:

node dist/apps/service-account/main.js  
node dist/apps/service-notification/main.js  
node dist/apps/service-billing/main.js  
node dist/apps/service-project/main.js  
node dist/apps/service-tenant/main.js  
node dist/apps/service-access/main.js  

With the databases and the services running you can now start the gateways as mentioned above.

yarn start:dev api-admin  

Get started by registering a user on the admin gateway

In the graphql playground running at http://localhost:4000//graph you can register a user:

mutation register {  
  account {  
    register(input:{  
        firstname: "Alice"  
        lastname: "Bob"  
        email: "[email protected]"  
        password: "supersecretpassword"  
    }) {  
      token  
    }  
  }  
}  

All going well you should have received the following reply as well as an email with a verification code as well as a jwt token that contains both email and verification code for your frontend

{  
    "data": {  
        "account": {  
            "register": {  
                "token": "gtdjghdrtd65edjhc-chxgfxjtgzrkzxyzxtrs45wi6kydch"  
            }  
        }  
    }  
}  

Quick Tips

Generating dynamic mongo filter GraphQL type.

You can generate dynamic graphql filter by decorating the fields you want to be
avaiable in your generated filter type with the @Filterable() decorator like below

import { Filterable } from '@ultimatebackend/core';  
  
@ObjectType()  
export class Tenant extends Node {  
  
  @Filterable()  
  @Field({ nullable: true })  
  name: string;  
  
  @Filterable()  
  @Field({ nullable: true })  
  normalizedName: string;  
}  

After adding decorator to the fields you can now add generate your graphql input type.

Note: Enums and sub-types not supported at the moment.

import { FilterMongo } from '@ultimatebackend/contracts';  
  
@InputType()  
export class TenantFilterInput extends FilterMongo(Tenant, { simple: true }) {}  

Multi-tenant Database for services.

You can enable multi-tenant database support in your micro-service by adding this

MongoModule.registerAsync({  
 useClass: MongoMultiTenantConfigService,  
})

code block to the service app module. For example;

import { Module } from '@nestjs/common';  
import { MongoModule } from '@juicycleff/repo-orm/database';  
import { MongoMultiTenantConfigService } from '@ultimatebackend/core/mutiltenancy';  
  
@Module({  
  imports: [
    // ...
    MongoModule.registerAsync({  
      useClass: MongoMultiTenantConfigService,  
    }),
    // ...
  ],  
})  
export class AppModule {}  

A good example is the service-project microservice.

Next you must enable multi-tenancy in the main.ts file of the api-admin service or any other api type microservice you create in your project.

import { NestFactory } from '@nestjs/core';  
import { NestCloud } from '@nestcloud/core';  
import { AppModule } from './app.module';  
import { enableMultiTenancy, TenantDatabaseStrategy } from '@ultimatebackend/core/mutiltenancy';  
  
async function bootstrap() {  
  const app = NestCloud.create(await NestFactory.create(AppModule));  
  
  app.use(enableMultiTenancy({  
    enabled: true,  
    tenantResolver: {  
      resolverType: 'Header',  
      headerKeys: {  
        tenant: 'x-tenant-id',  
        apiKey: 'x-tenant-key',  
      },  
      requiresToken: true,  
    },  
    databaseStrategy: TenantDatabaseStrategy.DataIsolation,  
  }));  
  
  // ..... code continues  
}  
  

Access token with scopes

Access tokens scopes just a combination of action and resource identifier. For example, take this mutation;

  @UseGuards(GqlAuthGuard)
  @Resource({ name: 'billing', identify: 'billing:card', roles: ['customer'], action: 'update' })
  @ResolveField(() => Card)
  async create(@Args('input') input: CreateCardInput, @Context() ctx: GqlContext): Promise<Card> {
    // @ts-ignore
    const result = await this.service.billing.createCard({...input}, setRpcContext(ctx)).toPromise();
    return result.card;
  }

Your access token scope will be update_billing:card and so your mutation to create an access token should look like this

mutation {
  accessToken {
    create(input: {
      name: "my-superb-token",
      scopes: ["update_billing:card"]
    }) {
      token
      id
      name
      active
    }
  }
}

That's all you need to know when creating access token which should be used together with your tenant normalized name to access the API without a logged in user.

More docs updates coming.

Test

# unit tests  
$ yarn run test  
  
# e2e tests  
$ yarn run test:e2e  
  
# test coverage  
$ yarn run test:cov  

Financial Contributors

Become a financial contributor. Your funds go to; people in need [Contribute]

Backers

Sponsors

Support this project with your organization. Your logo will show up here with a link to your website. [Contribute]

My Other Projects

Here are some of my projects, show some love and start them if you find them helpful :)

Special Thanks

This project wouldn't be possible without these two awesome projects, NestJS (Server Framework) and NestCloud, please make sure to Star them.

License

This project is MIT licensed.

About

Multi tenant SaaS starter kit with cqrs graphql microservice architecture, apollo federation, event source and authentication

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • TypeScript 98.8%
  • Other 1.2%