Tier 1 AMIs - base AMIs with Linux and cloned MDI server repos
A Tier 1 base AMI is defined by its Linux operating system. It is created infrequently to serve as the building block for Tier 2.
Summary of the base AMI:
- source AMI = Ubuntu 22.04 standard image, X86_64
- Linux user = ubuntu, the AWS standard
- region = Ohio, us-east-2
- instance type = t3.medium (2 vCPU, 4 GB RAM)
- storage = 8 GB EBS SSD
- Docker = installed and ready to build MDI images
- MDI repositories = two are cloned into the base AMI itself:
Linux operating system
The MDI can run on any operating system, but we use Ubuntu Linux by default, with version 22.04 LTS being current as of this writing.
AWS region
AWS AMIs are region specific, i.e., they are only available to be used for launching instances in the same region as the AMI itself. We build all supported AMIs in the Ohio, us-east-2, AWS region closest to Ann Arbor, MI.
Instance type
An AMI is not tied to a specific instance type, but we create our AMIs with the same resources as is recommended for eventually running MDI public server instances, i.e., t3 medium.
Storage
Storage volume size can be expanded when a new EC2 instance is launched, so instances used to generate MDI AMIs only need enough storage to handle the required installations, which is initially modest but increases when creating Docker images in AMI Tier 2.
MDI server repositories
As throughout the MDI project, we compartmentalize MDI public server functions into two distinct repositories:
MiDataInt/mdi-aws-ami prepares the base AMI for subsequent installation of the MDI. It installs things into the server OS itself, like Docker. The configuration script in this repository could be run on any computer, but the intent is that it be run on a fresh Ubuntu AMI on AWS.
MiDataInt/mdi-web-server carries code that configures, assembles, and launches Docker images and containers carrying an MDI server installation. Please note that the AMI itself does not run the MDI, which is installed into and run in Docker containers by scripts found in mdi-web-server.git.
TIER 1 INSTRUCTIONS - create the base image
We only need to create a base image once per Ubuntu version. Accordingly, the steps below are usually only performed by MDI Project administrators. They clone the mdi-aws-ami repo into a new EC2 instance and run the server configuration script to prepare for installing the MDI in later tiers/AMIs.
Launch an AWS instance
Launch an EC2 instance with the specifications listed above (or, choose a different base OS or AWS region, if desired).
Log in to the new instance using an SSH terminal
Details for how to log in to an AWS instance are amply documented by Amazon. Among many choices, we typically use Visual Studio Code with a remote connection established via SSH.
- https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/AccessingInstances.html
- https://code.visualstudio.com/docs/remote/remote-overview
Clone this repository
From within an SSH command shell on the new instance (note that git is pre-installed with Ubuntu):
cd ~
git clone https://github.com/MiDataInt/mdi-aws-ami.git
Run the server setup script
cd mdi-aws-ami
bash ./initialize-mdi-instance.sh
It will take a few minutes for all of the server components to be installed.
Security considerations for private base AMI
Base images are generally kept private since general users will start from a Tier 2, i.e., empty, AMI. Accordingly, no further action is required to secure a Tier 1 AMI (unlike Tier 2, below).
Save the base AMI
From within the AWS Management Console, select the running EC2 instance and execute:
Actions –> Images and templates –> Create image
The base image should be named and described according to the following conventions. The timestamp can be used to infer the version of the various MDI repos installed into a given server instance.
name
mdi-base_ubuntu-22.04_yyyy-mm-dddescription
Michigan Data Interface, base image, Ubuntu 22.04, yyyy-mm-dd