NCorreia100/Street-Breasy-Optimization

Optimization of Back-end to reduce latency and increase availability

Street Breasy Optimization Project

Street Breasy Project Introduction

This repository contains the source for a single component of a real estate listing page built with a microservice architecture (SoA). The component is designed as a full stack service and includes a Node.Js API, a MongoDB database, a Nginx content-caching proxy and a Node.js load-balancer proxy. The API provides different image links to the client depending on a listing id number.

Index of Contents

• Project subfolders
• Requirements and Dependencies
• Environment Setup
• API Documentation
• Back-end Performance Optimization

Project Subfolders

• Image Carousel Component API
• Load Balancer Proxy
• [NginX Content-Caching Proxy] (https://github.com/NCorreia100/Street-Breasy-Optimization/tree/master/nginx-proxy)
• Database Scripts

Requirements

External API Key

• This project makes use of the MapBox API to access static map images. To get an API key visit: https://www.mapbox.com/signup

Key Dependencies

• Node.js with Express.js web framework
• MongoDB database
• React front end framework
• Eric Meyer's CSS Reset

Environment Setup

1. Create a server instance to run the project
2. On the remote server, allow traffic through TCP protocol on ports 22 and 80 for SSH and HTTP and ICMP (for ping)
3. VPN to the remote server
4. On the terminal, once at the desired directory, run the command git clone https://github.com/HackberryPie/listing-page-load-balancer.git followed by cd listing-page-load-balancer.
5. Run the command npm install to download all required modules
6. On a different terminal window run the command mongod --dbpath [path] in which path refer to path to a directory of your choice as data storage
7. To start the server run the command npm start.
8. To make an API request, on your browser's address bar, enter your remote server's URL (or IP) followed by a single / and an interger between 1 and 10000000
9. Optionally, you may use a tool to start the node server such as pm2 or docker.
10. Optionally you may also select a different port to receive incoming HTTP requests. To do so, modify the value of GATEWAY_PORT. Default is 80.

API Documentation

API Documentation

Back-end Performance Optimization

Optimization goals

The optimization goals for this project were as follows:

• Decrease latency for a complete page load to not exceed 1 second
• Increase server capacity to sustain at least 2'000 RPS (requests per second)
• Diminuish error rate (such as connection time out and port in use)
• Document metrics akin to each implementation

Optimization tools / techniques

• Webpack to compress all code into a single file with the required modules
• Babel to transpile JSX code to JavaScript and eariler versions of JavaScript (such as ES6 and ECMAScript2018) to ES5
• Compression.js to apply gzip compression to reduce the response body size
• Morgan.js to log API requests and timing to complete the full request-response cycle
• Artillery.js for submiting large waves of requests (and get error rate)
• Newrelic.js to create reports on the performance of the server (latency and request-rate)
• Google's DevTools Performance Analyzer (page load performance)

##Optimization Metrics - https://docs.google.com/document/d/1vXppWpZGpDYjkNtJU35qRpkVJGpV_mEjU3HsvI0lV3M/edit

Author

• Nuno Correia