Project 1: The Data Science Toolkit

The goal of this first project is to set up a jupyter notebook.
A jupyter notebook is a web application where you can run python / r code but also save notebooks for later.
The notebook is going to be running in a docker container hosted on an EC2 instance.

Table of Contents

1. Create ssh keys
2. Create an EC2 instance
3. Install Docker on you EC2 instance
4. Install Jupiter to run on Docker
5. Launch Jupyter
6. Access you Jupiter notebook from a browser

Create ssh keys

In order to connect to the EC2 instance we will be building at step 2 we first need to create a pair of private / public ssh keys which we will load into the list of autorized autorized keys on the EC2 instance

In your laptop terminal : ssh-keygen
You don't need to provide a file to add it and a password

Create an EC2 instance

Go to Amazon AWS and create an account

1. Go to EC2
2. Configure security groups:
   • Open ports 22 (SSH) 80, 8888,
   • The group should look like this screenshot :
3. Configure your instance:
   • Choose Ubuntu
   • Choose t2;micro
   • The rest of the configuration is straight forward untill you launch the instance
   • Add you ssh public key :
     • In your terminal : cd ./ssh and cat id_rsa.pub
     • copy the output
4. Launch and retrieve to public ip adress

Install Docker on you EC2 instance

Connect to your EC2 instance : ssh ubuntu@myEC2instanceIP

The prompt should now like this : ubuntu@ip-172-31-17-108:~$

Install using : curl -sSL https://get.docker.com | sh

Install Jupiter to run on Docker

As said in the introduction, we are not running Jupyter from our EC2 instance but from a Docker container hosted on our EC2 instance. This is going to make Jupyter's installation much easier :
We just need to pull Jupyter's docker image : docker pull jupyter/datascience-notebook

## Launch Jupyter Check that you have correctly pulled the jupyter image : docker images
Then run : docker run -v /home/ubuntu:/home/jovyan -p80:8888 -d jupyter/datascience-notebook

Access you Jupiter notebook from a browser

Open you browser and go to ipAdressOfYourEC2Instance:80
You will be asked a token / password which you can find by running :
docker ps to list all docker process running : find yours and copy it's id
Then run : `docker exec IdOfYourProcess jupyter notebook list

How much does it cost ?

The price of the EC2 instance depends its configuration which includes :

• The type of machine : general, compute oriented, GPU, memory oriented
• Number of vCPU - This indicates your ability to split your computation into pieces running simultaneously (paralel computing)
• Memory size : the size of data you can perform computation on
• Storage size You pay based on the time your instance is up - you can shut it down when you don't use it and will not be paying for this time.

Type	vCPU	Memory	Price per Hour ($)	Price per Month ($)
general	1	2	0.023	6.6
general	8	32	0.4	115
compute oriented	4	61	0.9	259
compute oriented	32	488	7.8	2074