Lec 04-3: Managing Amazon EC2

Lanching options

• AWS Management Console
• AWS Command Line Interface (CLI)
• AWS Software Development Kit (SDK)
• Other AWS services: AWS CloudFormation, AWS OpsWorks, AWS Elastic Beanstalk

---

Amazon EC2 Lifecycle States

Pending - Not billed

• preparing to enter the running state
• when it launching for the first time or when it starting after being in the stopped state.

Running - Billed

• running and ready for use

Rebooting - Billed

• rebooting its operating system

Shuttle-down - Not billed

• preparing to be terminated

Terminated - Not billed

• permanently deleted and cannot be started

Stopping - Not billed if preparing to stop. Billed if preparing to hibernate

• preparing to be stopped or stop-hibernated

Stopped - Not billed

• shut down and cannot be used. can be started at any time

Stop-hibernated - Not biled

• shut down and cannot be used. can be started at any time.

Stopped, Stop-hibernated, Stopping -> Only instances backed by Amazon EBS

---

Scaling options

Vertical Scaling

• upgrading or downgrading server capacity of an EC2 instance
• 3 Steps
  1. Stop the instance
  2. Make the desired change to the instance size or type
  3. Start the instance
• Disadvantages of Vertical scaling
  • Manual process
  • Challenging when you have multiple instances to manage
  • Limitd on the instance scalability
  • Requires the instance to be unavailable while making changes

Use Case

• The instance size is too weak to process basic workload tasks
• Horizontal Scaling would not work in this scenario, because adding more instances would not address the issue of instances that cannot handle basic processing needs.

Horizontal Scaling

• adding or removing Amazon EC2 instances
• It can be automated by using the Amazon EC2 Auto Scaling service
• Can add as many instances as you need and terminate them when they are no longer needed.
• Benefits
  • Improves fault tolerance
  • Increases application availability
  • Lowers costs
  • Requires no downtime to implement

Use Case

• More requests are coming in than your instance can keep up with
• Vertical scaling would not be the best option because your application would need to be offline while you scaled up and down.

Auto Scaling

• Amazon EC2 Auto Scaling can help you automate horizontal scaling according to conditions that you define
• Can launch instances across Availability Zones
• Replaces instances
• Helps to save money

Amazon EC2 Auto Scaling Options

• Scheduled scaling
  • known load changes based on a schedule
• Dynamic Scaling
  • follow the demand curve, which reduces the need to manually provision EC2 capacity in advance
  • Can use target tracking scaling policies to select a metric such as CPU utilization
• Predictive scaling
  • Use machine learning to schedule
  • removes the need for manual adjustment to EC2 Auto Scaling parameters

---

Traffic management

Elastic Load Balancing (ELB)

• ELB is scalable AWS service that automatically distributes incoming application traffic accross multiple EC2 target.

ELB Options

• Application Load Balancer
  • ideal for load balancing HTTP and HTTPS traffic.
  • operates at the request level(layer 7)
• Network Load Balancer
  • ideal for load balancing TCP and UDP traffic
  • when you need ultra-high performance and static IP addresses for your application
  • operates at the connection level (layer 4)
• Gateway Load Balancer
  • helps you to manage your third-party virtual appliances
  • operates at the network level (layer 3) and provides a gateway for distributing traffic accross multiple virtual appliances

Additional Options for Connecting to EC2

• EC2 Instance Connect
  • using SSH
  • connect with a valid username without a key pair
• Session Manager
  • fully-managed AWS Systems Manager capability
  • manage instance through an interactive one-click, browser-based shell or through the AWS CLI
• SSH Client
  • connect with a key pair
• EC2 serial console
  • provides access to your intance's serial port
  • you can enter commands to an instances

---

Cost optimization

• Using the right size instance
• Increasing elasticity
  • reducing the amount of server capacity that is idle by implementing deployments that are elastic
  • Automatic scaling
• Choosing optimal pricing models
• Optimizing storage choices