EC2: VM as a service:

• autoscaling (mem/cpu metrics based instance start/stop)
• EBS (persistent storage)
• Images (how to create your own software stack)
• ELB (managed Load Balancer)

Virtual Networks:

• subnets
• reserved IP
• firewall rules

Management:

• CloudTrail (audit of all users AWS usage)
• CloudWatch (metrics/log collection)
• CloudFormation (aws specific “terraform”)
• LightSail: preconfigured stacks: LAMP, Nginx, MEAN and Node.js. Easier to manage than pure ec2 instances. – [optional]

Databases:

• DynamoDB (amazon’s proprietary schemaless NoSql) written in JAVA, can be self hosted, billed by per request, fully managed. -[optional]
• RDS (hosted relational db)
• postgres
• mariadb (mysql)
• oracle
• sql server
• DocumentDB (hosted mongodb) – [optional]
• ElastiCache (hosted redis) – [optional]
• Storage options, availability/speed/price:
• EBS (disk as a service)
• snapshots
• S3 (object storage, rest api, industry standard)
• EFS (nfs as a service)
• Pricing: AWS makes it really complicated:
• reserved instances (prepay for 1-3 years)
• spot price (can be 70% cheaper, with some risk)
• IAM
• Creating Roles, Policies, binding them
• authentication best practices (no root account, rotate keys, 2fa)
• Organizing teams
• Even create sub accounts
• HTTPS: learn how to use Letsencrypt. Not AWS specific, but important to
  know about a free automated was of creating certificates.

Optional:

• Lambda (Serverless): no need to pay for a VM if an app is used only rarely:
• Wide range of language can be used: javascript/python/java/go …
• Lot of Infrastructure automation possibility, almost all service can trigger a lambda function
• CI/CD hook triggers
• chatops (bot listens on slack, and deploys a VM…)
• billed by 100ms
• API Gateway: rest endpoint implemented as Lambda

Application Integration:

• SQS: managed queue. Easy glue between apps / services
• SNS: manages topic/subscription service:
• system to system communication: triggers: lambda/sqs/http endpoint
• user notification: email, sms, push notification
• Elastic Beanstalk: dev uploads a bundle (jar/zip) beanstalk will create VM instance/network/ … resources automatically.
• Route53: managed DNS

Containers:

• ECS (AWS opinionated docker solution)
• EKS (managed kubernetes)

Webserver – AWS Console

Goal: create a Webserver via the AWS provided browser based GUI.

• setup shortcuts, bookmarks, resource group
• Create Keypairs (ssh)
• start an Ubuntu/Debian instance

Webserver – cli

Goal: Same as previous exercise, but everything from the terminal

• aws sdk installation
• shell scripting basics (functions/variables/loops)

Webserver – Spot Instance

Goal: Learn how to save up to 70% of the cost for dev/qa environments.

Webserver – EBS – snapshot

Goal: learn how to use persistent disks, instead of volatile instance storage.

Create an Image

Goal: Instead of having a word docs describing what to install/configure, create a reusable blueprint.

• start with an “artisan” image (quick and dirty)
• automate it with packer

Instance Templates

Goal: Instead writing word docs about how to start an instance (which region, which instance type, how big the disk, networks, subnetworks …) his is a
prerequisite for the next:

Autoscaling and LoadBalancing

Goal: how to survive Black Friday? Automation which checks cpu/memory load metrics, and scales up/down.

Database management

Goal: Understand the pro/cons (hint: price is the only issue) of maintaining the DB or let AWS manage it.

• WebApplication the AWS way
• Static webhosting on S3
• Serverless Web app: Lambda + S3 + managed DB
• Beanstalk
• LightSail

Serverless

Goal: Understand how Lambda fits into both infrastructure glue code/ automation, and also complete applications (rest API)