Giter Site home page Giter Site logo

letubert / reference-architectures Goto Github PK

View Code? Open in Web Editor NEW

This project forked from cloudposse-archives/reference-architectures

0.0 3.0 0.0 51 KB

Get up and running quickly with one of our reference architecture using our fully automated cold-start process.

License: Apache License 2.0

Makefile 7.10% HCL 79.95% Shell 12.94%

reference-architectures's Introduction

reference-architectures

Get up and running quickly with one of our reference architectures using our fully automated cold-start process.

NOTE: This project is under active development and subject to change. Please file issues for all bugs encountered.

Table of Contents

Known Limitations

  • AWS does not support programmatic deletion of accounts. This means that if you use this project to create the account structure, terraform is not able to completely destroy it.
  • AWS by default only permits one subaccount. This limit can be easily increased for your organization, but can take up to several days.
  • AWS will rate limit account creation. This might mean you'll need to restart the provisioning (just re-run make root).

High Level Overview

You can provision the basic reference architecture in 3 "easy" steps. =)

All accounts will leverage our terraform-root-modules service catalog to get you started. Later, we recommend you fork this and start your own service catalog suitable for your organization.

This process involves using terraform to generate the code (Dockerfile, Makefile, terraform.tfvar, etc) that you will use to manage your infrastructure.

This is a "bootstrap" process that gets you running from a cold-start. You do it once and then you literally throw this (e.g. reference-architectures/) repo away. This project generates all the pre-configured boilerplate scaffolding you need in the repos/ directory.

When you're done, in the repos/ directory you'll have one Git repo for each AWS account. These repos are what you'll want to push up to GitHub. Each repo contains everything necessary to administer that account. We practice a strict "share nothing" approach, which is why each account gets it's own repo, terraform state backend, and DNS zone. This maximally reduces the blast radius of any human errors in one account affecting any other account. Also, because each account has it's own repo, it's ideally suited for larger enterprise or corporate environments where various stakeholders will be responsible for running services in their account.

See the Next Steps section for where to go after this process completes.

Architecture

Our "reference architecture" is an opinionated approach to architecting accounts for AWS.

This process provisions (7) accounts which have different designations.

Here is what it includes. Enable the accounts you want.

Account Description
root The "root" (parent, billing) account creates all child accounts and is where users login.
prod The "production" is account where you run your most mission critical applications
staging The "staging" account is where you run all of your QA/UAT/Testing
dev The "dev" sandbox account is where you let your developers have fun and break things
audit The "audit" account is where all logs end up
corp The "corp" account is where you run the shared platform services for the company
data The "data" account is where the quants live =)
testing The "testing" account is where to run automated tests of untrusted infrastructure code
security The "security" account is where to run automated security scanning software
identity The "identity" account is where to add users and delegate access to the other accounts

Each account has its own terraform state backend, along with a dedicated DNS zone for service discovery.

The root account owns the top-level DNS zone and then delegates NS authority to each child account.

Assumptions

  1. We are starting with a clean AWS environment and a new "root" (top-level) AWS account. This means you need the "root" credentials, since a fresh AWS account doesn't even have any AWS roles that can be assumed.
  2. You have administrator access to this account.
  3. You have docker installed on your workstation.
  4. You have terraform installed on your workstation.

Checklist

Before we get started, make sure you have the following

  • Before you can create new AWS accounts under your organization, you must verify your email address.
  • Open a support ticket to request the limit of AWS accounts be increased for your organization (the default is 1).
  • Clone this repo on your workstation.
  • Create a temporary pair of Access Keys. These should be deleted afterwards.
  • Export your AWS "root" account credentials as AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY (this is temporary for bootstrapping).
  • An available domain we can use for DNS-base service discovery (E.g. ourcompany.co). This domain must not be in use elsewhere as the root account will need to be the authoritative name server (SOA).
  • Ensure that any users who will be added during this bootstrap process have setup their keybase profile. You'll need this if setting them up in the users section of the config/root.tfvars. For example you should be able to verify their public key on keybase.io by running curl https://keybase.io/$username/key.asc.

Get Started

1. Provision Root Account

The "root" account is the top-most AWS account from which all other AWS accounts are programmatically created.

WARNING: Terraform cannot remove an AWS account from an organization. Terraform cannot close the account. The child account must be prepared to be a standalone account beforehand. To do this, issue a password reset using the child account's email address. Login and accept the prompts. Then you should be good to go. See the AWS Organizations documentation for more information.

WARNING: Do not chain the make targets together (e.g. make root children finalize) as it is not currently supported.

This account is provisioned slightly different from the other subaccounts.

Update the configuration for this account by editing the configs/root.tfvar file.

Then to get started, run:

make root

NOTE: We need to know each account's AWS_ACCOUNT_ID for Step 2. NOTE: Sometimes provisioning of the account module fails due to rate limiting by AWS on creating subaccounts. If this happens, just run make root/provision to retry. If that works, just continue on to step 2, once it completes.

Here's what that roughly looks like (but entirely automated).
  1. Create a new account git repo.
  2. Render templates into the repo (including Dockerfile).
  3. Build a docker image.
  4. Run the docker image and start provisioning resources including the Terraform state backend and child accounts.
  5. Create the IAM groups to permit access to child accounts.
  6. Write a list of child account IDs so we can use them in the next phase.

2. Provision Subaccounts

Subaccounts are created by the root account, but are ultimately provisioned using the subaccount containers.

Update the configuration for all the child accounts by editing the configs/$stage.tfvar file (replace $stage with the name of the account).

To get started, run:

make children
Here's what that roughly looks like (but entirely automated).

For each child account:

  1. Create a new account git repo.
  2. Render the templates for a child account into the repo directory (include Dockerfile). Obtain the account ID from the previous phase.
  3. Build a docker image.
  4. Run the docker image and start provisioning the child account's Terraform state bucket, DNS zone, cloudtrail logs, etc.

3. Delegate DNS

Now that each subaccount has been provisioned, we can delegate each DNS zone to those accounts.

To finish up, run:

make finalize
Here's what that roughly looks like (but entirely automated).
  1. Re-use the docker images from phase (1) and phase (2).
  2. Update DNS so that root account delegates DNS zones to the child accounts.
  3. Enable cloudtrail log forwarding to audit account.

Next Steps

At this point, you have everything you need to start terraforming your way to success.

All of your account configurations are currently in repos/

  • Commit the changes in repos/. Open Pull Requests.
  • Ensure that the name servers for the service discovery domain (e.g. ourcompany.co) have been configured with your domain registrar (e.g. GoDaddy).
  • Delete your root account credentials. They are no longer needed and should not be used. Instead use the created IAM users.
  • Request limits for EC2 instances to be raised in each account corresponding to the region you will be operating in.
  • Set the child account's credentials. To do this, issue a password reset using the child account's email address. Login and accept the prompts. Setup MFA.
  • Ensure you have MFA setup on your root account.
  • Consider adding some other capabilities from our service catalog.
  • Create your own terraform-root-modules service catalog for your organization.

NOTE: This repo can be deleted once you're all done and pushed your changes in the repos/ directory to GitHub. The rest of your development should happen inside your infrastructure repos.

Getting Help

Did you get stuck? Find us on slack in the #geodesic channel.

reference-architectures's People

Contributors

osterman avatar daveyu avatar keen99 avatar

Watchers

James Cloos avatar Leon avatar  avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.