Skip to content

stephenJeong/stephen_photoCarousel

BranchesTags

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

281 Commits
client
client
 
 
database
database
 
 
node_modules
node_modules
 
 
server
server
 
 
testing
testing
 
 
.gitignore
.gitignore
 
 
Dockerfile
Dockerfile
 
 
README.md
README.md
 
 
STYLE-GUIDE.md
STYLE-GUIDE.md
 
 
config.js
config.js
 
 
docker-compose.yml
docker-compose.yml
 
 
newrelic_agent.log
newrelic_agent.log
 
 
package-lock.json
package-lock.json
 
 
package.json
package.json
 
 
webpack.config.js
webpack.config.js
 
 

Repository files navigation

Steam Photo Carousel

Create a clone of a game page on the Steam website

Related Projects

Table of Contents

  1. Usage
  2. Development Setup
  3. Log
  4. Pro Tips

1. Usage

This service is part of a game page on the Steam website.

This service is the photo carousel component of the page that was inherited from a differnet student which will give the users the ability to scroll through different screenshots from the game.

1.1 CRUD Endpoints

  • GET /api/images/:gameId
    • returns all images that are relevant to a specific game
  • POST /api/images/:gameId
    • adds more images to the photo carousel of a specific game
  • PUT /api/images/:gameId
    • replaces an image at random in the photo carousel of a specific game
  • DELETE /api/images/:gameId
    • deletes an image in the photo carousel of a specific game

The shape of the data should be as follows:

{
  'gameId': Number,
  'gameName: String,
  'images': {
     'image1': String,
     'image2': String,
     'image3': String,
     'image4': String,
     'image5': String,
     'image6': String,
     'image7': String,
     'image8': String,
     'image9': String,
     'image10': String
   }
}

2. Development Setup

2.1 MySQL Setup

  1. For macOS users, visit https://dev.mysql.com/doc/mysql-osx-excerpt/5.7/en/osx-installation.html for instructions on downloading the .dmg file and installing it.
  2. Install mysql module through npm npm i --save mysql
  3. Create file to connect app to mySQL database Example code:
var mysql = require('mysql');
var connection = mysql.createConnection({
  host     : 'localhost',
  user     : 'root',
  password : 'secret'
});

connection.connect(function(err) {
  if (err) {
    console.error('error connecting: ' + err.stack);
    return;
  }

  console.log('connected as id ' + connection.threadId);
});

Make sure that the host, user, and password match what you have set up on your local machine. 4. Create schema file

DROP DATABASE photoCarousel;

CREATE DATABASE IF NOT EXISTS photoCarousel;

USE photoCarousel;

CREATE TABLE IF NOT EXISTS games (
  gameId INT NOT NULL AUTO_INCREMENT,
  gameName VARCHAR(100) NOT NULL UNIQUE,
  PRIMARY KEY(gameId)
);

CREATE TABLE IF NOT EXISTS images (
  id INT NOT NULL AUTO_INCREMENT,
  gameId INT NOT NULL,
  image1 VARCHAR(255) DEFAULT '',
  image2 VARCHAR(255) DEFAULT '',
  image3 VARCHAR(255) DEFAULT '',
  image4 VARCHAR(255) DEFAULT '',
  image5 VARCHAR(255) DEFAULT '',
  image6 VARCHAR(255) DEFAULT '',
  image7 VARCHAR(255) DEFAULT '',
  image8 VARCHAR(255) DEFAULT '',
  image9 VARCHAR(255) DEFAULT '',
  image10 VARCHAR(255) DEFAULT '',}
  FOREIGN KEY (gameId) REFERENCES games(gameId)
);

How to connect to MySQL database from command line:

mysql -u USERNAME -p By including -p, it'll prompt Enter Password. Enter your password and it'll direct you to the mySQL shell.

2.2 Cassandra Setup

Install HomeBrew

brew install cassandra

Install Python

brew install python

Install Java

brew cask install java

Install cql

pip install cql pip install cassandra-driver Installing cassandra-driver will take a few minutes so if it doesn't install immediately, just wait!

If you run into "pip: command not found", then run: sudo easy_install pip

If you run into "ERROR: Cannot uninstall 'six'", then run: sudo pip install tld --ignore-installed six

Start/Stop Cassandra

brew services start cassandra brew services stop cassandra

Cassandra shell

You get into cassandra shell by running cqlsh

Set up new keyspace and data object in shell

A keyspace in Cassandra is like a database in other RDMS. I ran: CREATE KEYSPACE photo_carousel WITH replication = {'class': 'SimpleStrategy', 'replication_factor': '3'};

I created a UDT (user defined type) to house the 10 images for each game.

CREATE TYPE photo_carousel.images_all (
  image1 text,
  image2 text,
  image3 text,
  image4 text,
  image5 text,
  image6 text,
  image7 text,
  image8 text,
  image9 text,
  image10 text
);

Then create the table.

CREATE TABLE photo_carousel.games (
  game_id int PRIMARY KEY,
  game_name text,
  images FROZEN<images_all>
);

Connect node to Cassandra database

Queries to use with Cassandra

2.3 React Build Setup

3. Log

3.1 Set Up CRUD Endpoints

  • GET /api/images/:gameId
    • returns all images that are relevant to a specific game
  • POST /api/images/:gameId
    • adds more images to the photo carousel of a specific game
    • For this project, I decided to limit the number of images that a game can have to a maximum of 10. Since this is the case, my POST api will take the following steps to add an image to a game:
      1. Retrieve the game data from the database
      2. Check to see if the game already has 10 images
      3. If it has 10 images, then console log that an existing image must be replaced
      4. If it has less than 10 images, then add the new image to the last available image slot
  • PUT /api/images/:gameId
    • replaces an image at random in the photo carousel of a specific game
    • I decided to have it randomly replace an image because I wanted to spend more time on optimizing and scaling my database and this appeared to be the simplest solution
  • DELETE /api/images/:gameId
    • deletes an image in the photo carousel of a specific game

For my endpoints, since I need to update specific fields, I am using $set for mongoose which allows for that functionality.

3.2 Set Up Jest to Test APIs

3.3 Development Setup

3.3.1 MySQL Setup

First Issue While setting up MySQL I ran into the issue of adding an object as the value in a column. I wanted my image row to contain an object with 10 images. I wasn't sure how I can set up my data so that when I pull from the mySQL database, I'll get an object with 10 images so that I can loop through it. I initially considered using 2 tables and making use of the foreign key to connect them but then I realized that I didn't need to do that. All I needed to do was create the object, use JSON.stringify and then add it as a value like I would anything else.

I started testing my seeding script by seeding them in batches: 10,000 100,000

I used the following query to make sure the rows were successfully added to my database:

SELECT COUNT(*) FROM games;

Second Issue I received this error:

error connecting: Error: connect ETIMEDOU at Connection._handleConnectTimeout

I also increased the speed of the inserts by using extended insert (i.e. INSERT INTO user (id, name) VALUES (1, 'Ben'), (2, 'Bob');)

The issue I continue having even when using async/await is that the await is not resolving. The await is also stopping my for loop.

ANSWER! The reason that my await function wasn't resolving is because I left an error handling function in the argument

So previous code was:

await db.pool.query(queryString, queryArgs, (err, result) => {
  if (err) {
    console.log('error in insert query', err);
  } else {
    console.log('success!');
  }
})

But I changed it to: await db.pool.query(queryString, queryArgs)

I also had to include try/catch so the final code looked something like:

try {
  await db.pool.query(queryString, queryArgs);
} catch (error) {
  console.log('error in catch:', error);
}

Third Issue My server was unable to connect to mysql server.

ANSWER! After I upgraded MySQL, I didn't turn the server back on so I had to use sudo service mysqld start to start it.

Fourth Issue My version of MySQL (5.5) was unable to use the PARTITION keyword so I had to upgrade my mySQL server with the following code:

yum remove mysql55-libs
yum remove mysql55-server
yum install mysql56-server

Fifth Issue I had to recreate my mySQL server instance and I was running into an issue with user privileges so I had to run the following code to grant my app permission to access the mysql database.

CREATE USER 'sdc'@'ec2-54-193-107-57.us-west-1.compute.amazonaws.com' IDENTIFIED BY 'password';
GRANT ALL PRIVILEGES ON *.* TO 'sdc'@'ec2-54-193-107-57.us-west-1.compute.amazonaws.com' WITH GRANT OPTION;
CREATE USER 'sdc'@'%' IDENTIFIED BY 'password';
GRANT ALL PRIVILEGES ON *.* TO 'sdc'@'%' WITH GRANT OPTION;
FLUSH PRIVILEGES;

3.3.2 Cassandra Setup

Errors

  • CQLSH Connection Error ('Unable to connect to any servers')

    • Make sure that you start cassandra with brew services start cassandra before you try to use cqlsh. Otherwise you will receive the above error

  • Error: Not enough bytes to read value of component

    • I was initially using a stringified version of my object to pass into the images column but it should've been a plain object.

  • Error: Target data type could not be guessed

    • I received this error and most of my research was showing that if I use a prepare statement as one of the options in my .execute function, it should solve it. The problem turned out to be that I put "prepared" instead of "prepare".

  • ReadTimeout: Error from server: code=1200 when using SELECT COUNT(*) FROM photo_carousel.games in cqlsh

    • Counting the number of records in Cassandra is a heavy operation due to its distributed nature. So this is expected unless you make some dits to the connection timeout times.

  • Parameter "game_id" not defined when trying to use executeConcurrent

    • I used the same query string and query parameters as I was using when doing the single queries but those arguments were not working in executeConcurrent.

Cassandra query setup

I initially set up my insert queries for Cassandra using concurrent execution because I thought that simultaneously running the insert queries will make the query faster. But I realized that wasn't the case (or I set it up wrong).

Insert volume Concurrent Execution Batch Inserts
500,000 9 minutes 1 minute 14 seconds
10,000,000 1 hour 20 minutes 14 minutes 50 seconds

3.4 New Relic Setup

3.5 DBMS Benchmarking

In order to start load testing each database, I created separate files to house the different queries that will be required for each database.

I also created an environmental variable for the database so that depending on which script I use to start my service, it'll know which database to use.

k6 Setup

brew tap loadimpact/k6
brew install k6

You don't have to npm install any packages.

Create a new js file which looks like this in its most basic form:

// testing/loadTests.js

import http from "k6/http";
import { check, sleep } from "k6";
export let options = {
  vus: 10,
  duration: "10s"
};
export default function() {
  let res = http.get("http://localhost:3002/api/images");
  check(res, {
    "success": (r) => r.status == 200
  });
};

Once you update the above script, run k6 run loadTests.js <- If you're not in the correct folder, make sure to adjust this.

DBMS Route RPS LATENCY THROUGHPUT ERROR RATE
Cassandra GET 1 4.03ms 41.9rpm 0.00%
Cassandra GET 10 4.79ms 498rpm 0.00%
Cassandra GET 100 5.97ms 5,450rpm 0.00%
Cassandra GET 1000 24.2ms 12,700rpm 0.00%
Cassandra POST 1 4.75ms 45.5rpm 0.00%
Cassandra POST 10 3.52ms 399rpm 0.00%
Cassandra POST 100 2.47ms 5,010rpm 0.00%
Cassandra POST 1000 14.6ms 17,100rpm 0.00%
MySQL GET 1 4.34ms 50rpm 0.00%
MySQL GET 10 3.49ms 546rpm 0.00%
MySQL GET 100 4.96ms 4,230rpm 0.00%
MySQL GET 1000 25.1ms 12,400rpm 0.00%
MySQL POST 1 6.91ms 46.4rpm 0.00%
MySQL POST 10 7.32ms 114rpm 0.00%
MySQL POST 100 5.76ms 4,650rpm 0.00%
MySQL POST 1000 38.2ms 17,400rpm 0.00%

3.6 Deployment

After creating EC2 instance on AWS do the following steps:

  1. Cd into folder with the .pem file
  2. ssh -i /users/jenn/desktop/hackreactor/rpt15-SDC/sdc.pem ec2-user@ec2-13-56-223-63.us-west-1.compute.amazonaws.com
    • replace the link after "ec2-user@" with the url of your ec2 instance
  3. sudo yum install git
  4. Set up node.js (https://docs.aws.amazon.com/sdk-for-javascript/v2/developer-guide/setting-up-node-on-ec2-instance.html)
  • curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.34.0/install.sh | bash
  • . ~/.nvm/nvm.sh
  • nvm install node
  • test that node is running properly with node -e "console.log('Running Node.js ' + process.version)"
  1. git clone <repo>
  2. cd into <repo>

Create separate instances for service, proxy, and database.

How to set up mysql database on EC2

Install mysql yum install mysql-server Start mysql service mysqld start If you want to update the password mysqladmin -u root password <new password>

How to seed external database

Run the following npm script: "DB='mySql' NODE_ENV='prod' node database/mysql/seed.js && sleep 2 && DB='mySql' NODE_ENV='prod' nodemon server/index.js",

Problems testing with remote MySql database

When I was stress testing my database with k6 on my local machine, I did not have any issues connecting to my local MySql database but once I started trying to test with the remote database on AWS, I received:

Error: Handshake inactivity timeout

Solution: Using incorrect ec2 DNS. I was using the DNS for my service rather than the mysql instance. Every time I stop and start an AWS EC2 instance, the DNS changes so I need to update the url that I'm referencing each time.

Steps to take when starting new EC2 instance
  1. Update environmental variables on config.js file in root directory
  2. Copy newrelic.js from node_modules and put it in root directory
  • cp ~/node_modules/newrelic/newrelic.js ~/stephen_photoCarousel

3.6.1 Deployment Benchmarking

Benchmark

DBMS Route RPS LATENCY THROUGHPUT ERROR RATE
MySQL GET 1 68ms 60rpm 0.00%
MySQL GET 10 69ms 590rpm 0.00%
MySQL GET 100 66ms 6,000rpm 0.00%
MySQL GET 1000 5062ms 7630rpm 22.9%
MySQL POST 1 81ms 60rpm 0.00%
MySQL POST 10 65ms 600rpm 0.00%
MySQL POST 100 63ms 6,000rpm 0.00%
MySQL POST 1000 69ms 60,000rpm 0.00%

3.7 Optimization

Original Benchmark for > 1000 RPS

DBMS Route RPS LATENCY THROUGHPUT ERROR RATE
MySQL GET 1000 5062ms 7630rpm 22.9%
MySQL GET 2000 8968ms 7702rpm 43.1%
MySQL GET 5000 ms rpm ERROR OUT
MySQL GET 10000 ms rpm ERROR OUT

3.7.1 Server Side Rendering

I haven't completed setting up server side rendering yet.

Obstacles

Using import statements in node
  1. Create a separate index.js file in the root directory.
require('@babel/register');
require('./server/index.js');
  1. In package.json, for your starting script, reference this new index.js file rather than your server file. So it should be nodemon index.js
Stylesheets can't be used in server side rendering

Once I changed my server to perform server side rendering, my stylesheet stopped working. My solution is to use react styled components.

Cannot use window.location on server side rendering

Attempted to use react-router to get the URL from the client but that did not work. Seems like this method was getting the path of the file that requested it rather than the web page url.

3.7.2 Redis Cache

Benchmark after Redis

DBMS Route RPS LATENCY THROUGHPUT ERROR RATE
MySQL GET 1000 2156ms 22970rpm 0%
MySQL GET 2000 4711ms 19618rpm 5.3%
MySQL GET 5000 8757ms 3548rpm ERROR OUT
MySQL GET 10000 9738ms 2188rpm ERROR OUT

Redis is an in-memory data structure store and it supports both LRU (Least Recently Used) and LFU (Least Frequently Used) eviction policies. I am using this as one of my optimization strategies because it'll be faster to get data from memeory (when applicable) than always making a call to the database.

I will be using the allkeys-lru policy because with an app like Steam, I expect that certain games will be more popular than others and that popularity will continually change through the years. With that being the case, it makes more sense to always keep the the recent popular items in memory.

Setup Redis

  1. Run the following commands in terminal:
sudo yum -y install gcc make # install GCC compiler
cd /usr/local/src
sudo wget http://download.redis.io/redis-stable.tar.gz
sudo tar xvzf redis-stable.tar.gz
sudo rm -f redis-stable.tar.gz
cd redis-stable
sudo make distclean
sudo make
sudo yum install -y tcl

  1. Run a test to make sure Redis is set up correctly sudo make test If successful, you should see a smiley face like, \o/ and a message saying that "All tests passed without errors!"

  2. Copy Redis server and CLI

sudo cp src/redis-server /usr/local/bin/
sudo cp src/redis-cli /usr/local/bin/

  1. Copy over config file sudo cp /usr/local/src/redis-stable/redis.config /etc/redis/redis.conf

  2. Update config file

vim /etc/redis/redis.conf

bind 127.0.0.1 -> #bind 127.0.0.1
dir ./ -> dir /etc/redis
daemonize no -> daemonize yes
protected-mode yes -> protected-mode no
pidfile /var/run/redis.pid -> pidfile /etc/redis/redis.pid
logfile '' -> logfile /etc/redis/redis_log
#maxmemory <bytes> -> maxmemory 700
#maxmemory-policy noeviction -> maxmemory-policy allkeys-lfu

  1. Start the server: redis-server /etc/redis/redis.conf

  2. Check if Redis is working redis-cli ping

Set up redis on separate instance

Obstacles

Error connecting to redis: ReplyError: Ready check failed: DENIED Redis is running in protected mode because protected mode is enabled, no bind address was specified, no authentication password is requested to clients. I tried to disable protected mode in the config file and restart the server (redis-cli shutdown) but I received another error message saying that there are errors trying to shutdown.

Resolution I was able to shutdown the server by using: redis-cli shutdown nosave

I then re-started the server redis-server /etc/redis/redis.conf

In order to check if the config file loaded, you can use redis-cli -p 6379 info server

Benchmark after adding Redis on separate instance

Strategy Route RPS Latency Throughput Error Rate
Original GET 1000 5062ms 7630rpm 22.9%
Redis GET 1000 4369ms 23595rpm 0.0%
Original GET 2000 8968ms 7702rpm 43.1%
Redis GET 2000 8487ms 16717rpm 29.3%
Original GET 5000 error error error
Redis GET 5000 error error error
Original GET 10000 error error error
Redis GET 10000 error error error

3.7.3 MySQL Partitions

Benchmark after MySQL Partitions

Strategy Route RPS Latency Throughput Error Rate
Original GET 1000 5062ms 7630rpm 22.9%
Redis GET 1000 4365ms 23942rpm 0.0%
MySQL Partition GET 1000 4369ms 23595rpm 0.0%
Original GET 2000 8968ms 7702rpm 43.1%
Redis GET 2000 8064ms 14738rpm 37.7%
MySQL Partition GET 2000 8487ms 16717rpm 29.3%
Original GET 5000 error error error
Redis GET 5000 error error error
MySQL Partition GET 5000 error error error
Original GET 10000 error error error
Redis GET 10000 error error error
MySQL Partition GET 10000 error error error

1000 RPS performed similar with partitions but 2000 performed worse.

Add partitions to existing table

ALTER TABLE games PARTITION BY RANGE (game_id) (PARTITION p0 VALUES LESS THAN (1000000),PARTITION p1 VALUES LESS THAN (2000000), PARTITION p2 VALUES LESS THAN (3000000), PARTITION p3 VALUES LESS THAN (4000000), PARTITION p4 VALUES LESS THAN (5000000), PARTITION p5 VALUES LESS THAN (6000000), PARTITION p6 VALUES LESS THAN (7000000), PARTITION p7 VALUES LESS THAN (8000000), PARTITION p8 VALUES LESS THAN (9000000), PARTITION p9 VALUES LESS THAN (10000000), PARTITION p10 VALUES LESS THAN (11000000));

Query to see partitions SELECT * FROM information_schema.partitions WHERE table_name = 'games';

Query example to select rows from specific partition SELECT * FROM games WHERE game_id BETWEEN '1' AND '999999';

3.7.4 Custom Load Balancer

Benchmark after Load Balancer - 1st attempt

Strategy Route RPS Latency Throughput Error Rate
Original GET 1000 5062ms 7630rpm 22.9%
Redis GET 1000 4365ms 23942rpm 0.0%
MySQL Partition GET 1000 4369ms 23595rpm 0.0%
Custom Load Balancer GET 1000 4626ms 15035rpm 11.53%
Original GET 2000 8968ms 7702rpm 43.1%
Redis GET 2000 8064ms 14738rpm 37.7%
MySQL Partition GET 2000 8487ms 16717rpm 29.3%
Custom Load Balancer GET 2000 6924ms 14657rpm 36.6%
Original GET 5000 error error error
Redis GET 5000 error error error
MySQL Partition GET 5000 error error error
Custom Load Balancer GET 5000 error error error
Original GET 10000 error error error
Redis GET 10000 error error error
MySQL Partition GET 10000 error error error
Custom Load Balancer GET 10000 error error error

Steps to launch load balancer:

  1. Start new EC2 instances
  2. Update config.js file with the public DNS
  3. Update port numbers on each of the instances to be 3100, 3101, 3012, ...etc
  4. Comment out load balancer code
  5. Uncomment out redis code

I created my own load balancing code in the server by using the http request method. My code will loop through the available additional servers which will be listed as an array in the config file and continually loop through that array every time an api call is made. So if there are 3 elements in the array, the first API call will go to the first endpoint, the second will go to the second, the third will go to the third, and the fourth will start on the first again.

Obstacles

Error: Invalid protocol: ec2-13-56-149-165.us-west-1.compute.amazonaws.com Resolution My EC2 instances for the extra servers were using different ports and my security group wasn't allowing for those different ports. So I had to change that but in addition, in my config file, I forgot to include http:// at the beginning and the port number at the end.

3.7.5 NGINX Load Balancer

Steps to install NGINX yum install nginx

If you get an error message telling you to use amazon extras then use: sudo amazon-linux-extras install nginx1

Update config file sudo vim /etc/nginx/nginx.conf

# you must set worker processes based on your CPU cores, nginx does not benefit from setting more than that
worker_processes auto; #some last versions calculate it automatically

# number of file descriptors used for nginx
# the limit for the maximum FDs on the server is usually set by the OS.
# if you don't set FD's then OS settings will be used which is by default 2000
worker_rlimit_nofile 100000;

# only log critical errors
error_log /var/log/nginx/error.log crit;

# provides the configuration file context in which the directives that affect connection processing are specified.
events {
    # determines how much clients will be served per worker
    # max clients = worker_connections * worker_processes
    # max clients is also limited by the number of socket connections available on the system (~64k)
    worker_connections 4000;

    # optimized to serve many clients with each thread, essential for linux -- for testing environment
    use epoll;

    # accept as many connections as possible, may flood worker connections if set too low -- for testing environment
    multi_accept on;
}

http {
    # cache informations about FDs, frequently accessed files
    # can boost performance, but you need to test those values
    open_file_cache max=200000 inactive=20s;
    open_file_cache_valid 30s;
    open_file_cache_min_uses 2;
    open_file_cache_errors on;

    # to boost I/O on HDD we can disable access logs
    access_log off;

    # copies data between one FD and other from within the kernel
    # faster than read() + write()
    sendfile on;

    # send headers in one piece, it is better than sending them one by one
    tcp_nopush on;

    # don't buffer data sent, good for small data bursts in real time
    tcp_nodelay on;

    # reduce the data that needs to be sent over network -- for testing environment
    gzip on;
    # gzip_static on;
    gzip_min_length 10240;
    gzip_comp_level 1;
    gzip_vary on;
    gzip_disable msie6;
    gzip_proxied expired no-cache no-store private auth;
    gzip_types
        # text/html is always compressed by HttpGzipModule
        text/css
        text/javascript
        text/xml
        text/plain
        text/x-component
        application/javascript
        application/x-javascript
        application/json
        application/xml
        application/rss+xml
        application/atom+xml
        font/truetype
        font/opentype
        application/vnd.ms-fontobject
        image/svg+xml;

    # allow the server to close connection on non responding client, this will free up memory
    reset_timedout_connection on;

    # request timed out -- default 60
    client_body_timeout 10;

    # if client stop responding, free up memory -- default 60
    send_timeout 2;

    # server will close connection after this time -- default 75
    keepalive_timeout 30;

    # number of requests client can make over keep-alive -- for testing environment
    keepalive_requests 100000;
}

After making changes to servers, make sure to run the following two commands in your NGINX instance sudo service nginx reload sudo service nginx restart

To test that your config is set up correctly nginx -t

Benchmark after NGINX

Strategy Route RPS Latency Throughput Error Rate
Original GET 1000 5062ms 7630rpm 22.9%
Redis GET 1000 4365ms 23942rpm 0.0%
MySQL Partition GET 1000 4369ms 23595rpm 0.0%
Custom Load Balancer GET 1000 4626ms 15035rpm 11.53%
NGINX (4 servers) GET 1000 63ms 60000rpm 0.0%
Original GET 2000 8968ms 7702rpm 43.1%
Redis GET 2000 8064ms 14738rpm 37.7%
MySQL Partition GET 2000 8487ms 16717rpm 29.3%
Custom Load Balancer GET 2000 6924ms 14657rpm 36.6%
NGINX (4 servers) GET 2000 243ms 119879rpm 0.0%
Original GET 5000 error error error
Redis GET 5000 error error error
MySQL Partition GET 5000 error error error
Custom Load Balancer GET 5000 error error error
NGINX (4 servers) GET 5000 2042ms 74671rpm 50.6%
NGINX (6 servers) GET 5000 2676ms 145311rpm 15.7%%
Original GET 10000 error error error
Redis GET 10000 error error error
MySQL Partition GET 10000 error error error
Custom Load Balancer GET 10000 error error error
NGINX GET 10000 error error error

After changing MySQL's max connections limit from 151 to 200, the error rate dropped from the above 15.7% to 7.5%

Strategy Route RPS Latency Throughput Error Rate
NGINX (6 servers) GET 5000 2370ms 177271rpm 7.5%%

After changing MySQL's max connections limit from 200 to 100, the error rate dropped down to 0.0%

Strategy Route RPS Latency Throughput Error Rate
NGINX (6 servers) GET 5000 3382ms 145223rpm 0%

After creating the multiple servers I installed an npm module called "mysql-pool-booster" and it decreased my latency by about 1000ms

Strategy Route RPS Latency Throughput Error Rate
NGINX (6 servers) GET 5000 1985ms 217612rpm 0.2%

After adding 2 more servers

Strategy Route RPS Latency Throughput Error Rate
NGINX (6 servers) GET 5000 1985ms 217612rpm 0.2%
NGINX (8 servers) GET 5000 418ms 293478rpm 0.0%
NGINX (8 servers) GET 6000 2254ms 213137rpm 6.5%
NGINX (8 servers with Medium instance) GET 65000 1565ms 309611rpm 0.9%
NGINX (8 servers with Medium instance - 200 max connections) GET 10000 2530ms 289788rpm 20.2%
NGINX (8 servers with Medium instance - 100 max connections) GET 10000 2437ms 292481rpm 22.2%

4. Pro Tips

Create images on AWS

  • Select your instance on EC2 dashboard
  • Click Actions -> Image -> Create Image

Then when you create a new instance later, load up your AMI.

By creating an image, you'll save everything that was created and installed on that current instance so that you don't have to rebuild everything whenever you create a new instance.

Keep your SSH session from freezing after inactivity

Update config at ssh_config sudo vim /etc/ssh/ssh_config Host * ServerAliveInterval 100

Update config at sshd_config

sudo vim /etc/ssh/sshd_config
ClientAliveInterval 60
TCPKeepAlive yes
ClientAliveCountMax 10000

ClientAliveInterval - server will wait 60 seconds before sending null packet to client to keep the connection alive TCPKeepAlive - makes sure certain firewalls don't drop idle connections ClientAliveCountMax - Server sends messages to client even though it hasn't received any message back from client

Restart ssh server sudo systemctl restart sshd

MySQL Query Profiling (check query speed)

SET PROFILING = 1;

Run queries that you want to profile SELECT * FROM games WHERE game_id = 9500000; SELECT * FROM games PARTITION (p9) WHERE game_id = 9500000;

Show profiles to look at query speeds SHOW PROFILES;

About

No description, website, or topics provided.

Resources

Readme
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

No packages published

Contributors 2

  •  
  •  

Languages