leejunkim

EC2

8 min read

EC2

  • Elastic Compute Cloud
  • Allows you to rent a remote computer
  • One of the most important/popular services!

What is EC2

  • Diagram
  • It is a IAAS
    • Your unit of consumption is the instance, which is an operating system configured in some way
    • You can decide which OS/hardware/software/etc to install
    • You can deeply configure them as if they would be computers standing in your room, which makes this service powerful
    • Therefore, you can run ANY kind of workload on top of cloud
  • A private service (runs in private AWS zone) - uses VPC networking by default
    • An EC2 instance is configured to launch in a single VPC subnet, which is in a specific AZ
    • This means that an EC2 instance is AZ-resilient - if an AZ fails, then the EC2 fails

  • On demand billing - per sec/hr depending on the instance

EC2 Instance families

Different combinations of CPU, memory, storage, and networking capacity

  • Allows you to choose the appropriate combination of capacity to meet your application’s unique requirements
  • docs
  • General Purpose
    • provides a balance of compute, memory, and networking resources, and can be used for a variety of diverse workloads
    • A1, T1, Mac, etc
    • T1 is most popular
  • Compute Optimized
    • ideal for compute bound applications that benefit from high performance processors
    • C5, C4, etc (starts with C)
  • Memory Optimized
    • designed to deliver fast performance for workloads that process large data sets in memory
      • in memory caches, in-memory databases, real time big data analytics
    • R4, R5, etc
  • Storage Optimized
    • designed for workloads that require high, sequential read and write access to very large data sets on local storage
    • I3, I3en, etc
  • Accelerated Optimized
    • hardware accelerators, or co-processors
    • P2, P3, etc

EC2 Instance types

An instance type is a particular instance size + instance family

  • size: nano, micro, small, etc
    • instance sizes generally double in price & key attributes
  • instance type: t2 nano, etc

EC2 Tenancy

  • diagram
  • Dedicated Host
    • Entire physical server dedicated to you (you manage instance placement).
    • Full visibility into CPU sockets, cores, NUMA configuration.
    • Supports BYOL - Bring your own license (e.g., Windows Server, SQL Server, VMware licenses).
    • Billed per host (even if no instances are running).
  • Dedicated instance
    • Runs on hardware dedicated to you, but AWS fully manages the host.
    • No visibility into the physical server (you can’t control sockets/cores).
    • Complies with regulatory requirements (isolated from other customers).
    • Billed per instance (not per host).
    • AWS guarantees no other customer’s instances will share the host

Renting

  • Diagram
  • It’s not actually (typically) not an entire machine
  • Physical Machine
    • refers to the actual, physical hardware (servers) that AWS owns and operates in their data centers
    • real machines with processors, memory (RAM), storage, and networking components
  • You rent a “slice” of the physical computers in the AWS data centers
    • You rent a “slice/virtual server/EC2 instance”
    • These “slices” are isolated, which its own dedicated hardware which can’t be used by other instances (for security/performance reasons)
    • virtualization technology
  • When you use this EC2 service, you start EC2 instances

Configuring and launching

Configuration

  • What is AMI - Amazon Machine Image
    • A pre-configured template that includes an operating system and pre-installed software/tools (depending on the AMI).
  • Choose AMI
    1. AWS-Provided Default AMIs – Standard images maintained by AWS (e.g., Amazon 🐧Linux, Ubuntu).
    2. AWS Marketplace & Community AMIs – Shared by third-party vendors or the AWS community.
    3. Custom AMIs (Self-Created) – You can create your own AMI using EC2 Image Builder
      • you can even build & manage pipelines where you can automate the process of building + updating EC2 images, custom AMIs
      • useful if u can’t alr find one (or u don’t wanna pay for it)
  • Choose instance type (hardware profile)
    • you can compare different instance types all optimized for different workloads (different memory/storage/number of CPUs/etc)
    • Instance types are also grouped into instance (type) families
      • Main groups of instance types
  • Choose key/pair
    • The key pair ensures only you (or authorized users) can access the instance.
    • You use the private key to connect to the instance using SSH (for Linux) or RDP (for Windows).
    • Create new one or an existing one (or choose to not use it, not recommended)
  • Network settings
    • You can choose a subnet (or its defaulted)
    • If it’s a public subnet, choose Auto-assign public IP
    • Edit - You can even choose a VPC (and configure other details)
    • You create/manage a security groups
  • Storage
    • You can also add storage

Launch

  • Once you launch it, you can access it in Instances or EC2 Dashboard
  • If you click, you can get more info about it
    • public IPv4 address (typically added automatically) which you could use for sending an HTTP request to that instance, in case HTTP traffic is allowed
  • Connect to instance
    • more on it below
  • If connected, you get a console where you can run commands on that instance
    • you can download/install/start software, download code/application, etc etc
    • whatever u want
  • If you’re finished
    • “Instance state” → Stop Instance (or Terminate/Delete)
    • If you don’t use it just terminate it

Connecting to instance

  • EC2 Instance Connect
    • popular
    • X need key pair ONLY when using Amazon Linux 2 AMIs
    • make sure SSH traffic anywhere in the world is allowed on the instance
    • accessed in the Management Console
  • SSH client
    • you can connect to a running EC2 instance via any SSH client as long as you have a valid key pair
  • Session Manager
    • advanced option
    • manages fleets of EC2 Instances

Running a web service

  • When configuring (before launch)
    • allow HTTP traffic from the internet
    • you don’t need to allow SSH since the example code will do that automatically at start
    • In advanced > User data
      • allows you to run commands or scripts automatically when an instance first starts
      • helps automate instance setup without needing to manually connect via SSH
      • example code - as soon as the EC2 instance launches, it will install and start a web server without you needing to log in
      • AWS gives u more automation options
  • After launch
    • Use the public IPv4 address or the Public IPv4 DNS

EC2 Pricing Models

  • diagram
  • On demand - least commitment
    • Pay by hour, NO discounts
    • Price depends on the config (instance type and the region)
      • look at the pricing page for the details
      • charged by a second or the hr
      • no up-front payment & no long-term commitment
    • when you launch an EC2, this is the default
    • good for workloads that are short-term, spiky, or unpredictable
    • you pay per second/hour even tho the instance is idle
  • Spot instances - greatest savings
    • Configuration > Advanced Details > Check Request Spot Instances
    • You get spare instances with a discounted price (up to 90%!), but can be reclaimed any time
    • Useful for short utility tasks, not good for stuff like web servers
    • good for interruptible tasks!
  • Savings Plan
    • Buy them separately in the Cost Center in the Billing dashboard of your account
    • Pricing depends on the amount of compute usage ($/hour) rather than a specific instance type.
      • You commit to spending a fixed amount (e.g., $10/hour) on compute usage for 1 or 3 years, but you can use any instance type, region, or even AWS services like Fargate or Lambda.
    • Discounted price
  • Reserved Instances - best long term
    • Designed for apps that has steady state, predictable usage, or requires reserved capacity
    • committing to a term with upfront payment yields the largest discount
    • Pricing depends on the specific instance type, region, and commitment term.
      • You commit to a specific EC2 instance type (e.g., t3.large) in a specific region for 1 or 3 years in exchange for a discount
    • If you start an instance which meets the certain instance config requirements, you won’t be charged for it
  • Dedicated (not here) - most expensive
    • dedicated servers
    • can be on-demand or reserved or spot
    • when you need guarantee of isolated hardware

Example

A mid-sized financial services company called SecureFinance uses EC2 to run their core banking application. This application was developed 10 years ago and isn’t containerized.

  • Multiple EC2 instances running their Java-based banking application
  • These EC2 instances are distributed across multiple Availability Zones for high availability
  • They use specific EC2 instance types optimized for memory (r5 family) because their application requires large in-memory data processing
  • Their database tier runs on dedicated EC2 instances with provisioned IOPS storage volumes
  • They’ve created custom AMIs (Amazon Machine Images) with their specific security configurations and compliance tools pre-installed

Graph View

Backlinks