Advice regarding libraries and development technologies for new MP framework

[jaught]

Hey!

I've decided to build a simple 2D framework for building multiplayer/MMO games (similar scope to Agar.io), I have significant experience in building multiplayer engines for windows/linux but I have never touched web. If you have any advice on libraries and development technologies, or general advice, I would greatly appreciate it.

My current tech stack (if you think this is naive feel free to call me out, as I am indeed naive.):

• Language: Typescript
• Client
  • Platform: Browser
  • IDE: VS2017
  • Rendering: Pixi.js
  • Physics: Box2D or similar
  • Level Editor: Tiled or similar
  • Animation: Spline
• Server
  • Platform: Linux VM via google or aws
  • Run time: Node.js
  • Web Server: Apache
  • Database: MySQL
• Shared
  • Modules: UMD
  • Networking: WebSocks / Socket.io
  • Data Structures: typescript-collections
  • Logging: Log4ts (based on Log4js)

My framework will be a microservice architecture that will used to create the basis for the client engine and the server authority engine and then a bunch of microservices that clients interact with (Authenticator, Inventory, Logging, etc).

Q: Why don't I use Phaser or other for the client engine? From what I can see is that Phaser is hard and/or in optimal to run on the server side. I could use it on the client and then bunch together similar components and run them on the server but from what I can see Pixi provides a lot of runtime functionality that I will need, and I like the idea of using same physics and other libraries on both client and server.

Thanks in advance!

Edit 0:
Object pooling: Opool
Edit 1:
Typescript Optionals: https://www.npmjs.com/package/ts-optional

Node Websockets: https://www.npmjs.com/package/ws

Browser Event Emitter: eventemitter3
Edit 2:
Docker: for creating instances of node.js microservices.
Scaling: not sure yet, but simplest case will be a c++ service that will spawn docker services and join the cluster as a ring node and will limit the time slice of each microservice.