Source: http://turnoff.us/geek/microservices/

What is microservices?

• A typical microservice addresses only one fairly narrow business goal or need
   
• Functionally quite simple: an incoming request is processed, and a response is generated accordingly
   
• Services are distributed and communicate with one another via http / message bus in service mesh
   
• Designed to fail gracefully, without taking down adjacent services

Why use microservices?

Services can be developed with the technology that does a given task best

Can be portable and scale independently

Dev teams stay small and agile

Why Node.js?

• Suitable for consuming data from different sources / data streams and transforming it for other applications
   
• Widely used to consume and provide APIs (e.g. REST, GraphQL)
   
• Event-based Node.js allows application to load off asynchronous tasks to
   
• Fast, lightweight with high throughput

Why Node.js?

• Highly active community and ecosystem
   
• Constant flow of new features and software being built around and for Node.js​
   
• Community-built tools, tutorials, and documentation all help to ensure that anyone developing with Node.js can find helpful solutions to common problems
   
• Knowledge of JavaScript can be applied in both backend and frontend

Ref: CLOUD NATIVE NODE.JS @ 2018 Node Summit by Chris Bailey, Chief Architect, Cloud Native Runtimes, IBM

Source: 2018 NODE.JS USER STUDY  DETAILED REPORT OF FINDINGS

Business Impact

Types of Development Work

Source: 2018 NODE.JS USER STUDY  DETAILED REPORT OF FINDINGS

Source: 2018 NODE.JS USER STUDY  DETAILED REPORT OF FINDINGS

Let's build an online reservation system

Architecture Workshop - nearForm 2018

Source: nearForm NodeSummit 2018 - Architecture Workshop

Requirements

• Limited camping site
• Seasonal - open between 6/1 ~ 8/31
• Must be reserved 3 months in advance for duration less than 7 days
• Must be reserved 4 months in advance for duration at least 7 days
• First come, first serve basis
• Collecting deposit to hold the reservation
• Allowing user to return later to complete full payment, modify, or cancel the reservation
• Must be able to handle the load & process all reservation requests

MVP
(Minimum Viable Platform)

MVP Requirements

• Register and login
• 1 booking per user.
  • While one site is being reserved by one user, it is unavailable to other users
• Account for booking limitations

Websockets

Source: nearForm NodeSummit 2018 - Architecture Workshop

Source: nearForm NodeSummit 2018 - Architecture Workshop

MVP Solution - Issues

• Mini-monolith
  • React web app is served directly from Node.js causing the eventloop to
    slow down http response time
     
  • Every web socket is handled by each single Node.js app
     
  • Single DB connection per Node.js process
     
• No back-pressure handling in case of high volume
   
• Authentication needs to be carried across services
   
• This solution scales linearly with traffic (250 instances to serve 250
  requests) without blocking!

Let's Improve

• Optimize authorization across services
   
• Solution must scale cost-effectively for peak traffic (10k req/sec)
   
• Solution must implement techniques to help scaling horizontally
   
• Implement circuit breaking and back-pressure management

Source: nearForm NodeSummit 2018 - Architecture Workshop

Let's make it better

• Scale authentication up to 10k req/sec
   
• Segregation of concerns
   
• What if I want to continue my payment later on?

Source: nearForm NodeSummit 2018 - Architecture Workshop

What if anything goes wrong?

• Application level monitoring
   
• Infrastructure level monitoring and alerting
   
• Auditing capability
   
• Request tracking

Source: nearForm NodeSummit 2018 - Architecture Workshop

Source: nearForm NodeSummit 2018 - Architecture Workshop

DevSecOps

The Evolution

Web Page

The Evolution - Go to Cloud

The Evolution - The DevOp way

Centralized Logging
System

The Evolution - microservices

Source: http://turnoff.us/geek/are-you-ready-for-microservices/

Observability?

• Where is the problem?
   
• Need a way to drill down to the precise part of a service transaction where failures or latency are occurring

MTTD

MTTR

DevOps Goals

Source: https://www.slideshare.net/EduardoSolis4/opentracing-meetup-austin-tx

• Custom logging from each service / platform?
  • Telemetry data with tags
  • Req/Res logs (JSON/XML)
  • Application log (log4j)
  • Metrics
     

• x-requestmarker=abc1234​
  

   

• Trace between dependencies?
• Async Operations?

Distributed Tracing

• A trace tells the story of a transaction or workflow as it propagates through a distributed system.
   
• Spans are logical units of work in a distributed system
   
• Relationships are the connections between Spans.

Distributed Tracing

• Not New
  • Dapper (Google's paper)
  • Zipkin (https://zipkin.io/)
     
• Standard with OpenTracing (http://opentracing.io/)
  • Vendor-neutral APIs and instrumentation for distributed tracing

Image Source: https://expediadotcom.github.io/haystack/docs/ui.html