Amazon Elastic Compute Cloud (EC2) is the core compute service of AWS, providing resizable virtual machines in the cloud. It is the AWS equivalent of Azure Virtual Machines .
Term
Description
Instance
A running virtual machine
AMI
Amazon Machine Image โ the OS + software template for an instance
Instance Type
The hardware spec (CPU, RAM, GPU, network)
Key Pair
SSH public/private key for secure access
Security Group
Stateful firewall controlling inbound/outbound traffic
Elastic IP
Static public IP address that you own
User Data
Script that runs on first boot to configure the instance
Family
Optimized For
Example Types
t (General purpose, burstable)
Dev/test, low-cost
t3.micro, t4g.small
m (General purpose, balanced)
Web servers, small databases
m6i.large, m7g.xlarge
c (Compute optimized)
CPU-intensive apps, HPC
c6i.large, c7g.2xlarge
r (Memory optimized)
In-memory databases, caches
r6i.large, r7g.4xlarge
p / g / trn (Accelerated)
ML training, GPU workloads
p4d.24xlarge, g5.xlarge
i / d (Storage optimized)
High I/O, data warehouses
i4i.large, d3.xlarge
t3.micro is included in the AWS Free Tier (750 hours/month for 12 months).
Model
Description
Best For
On-Demand
Pay by the second, no commitment
Dev/test, unpredictable workloads
Reserved Instances
1 or 3-year commitment, up to 72% discount
Steady-state production workloads
Savings Plans
Flexible commitment on compute spend, up to 66% off
Flexible reserved-like pricing
Spot Instances
Spare capacity at up to 90% discount โ can be interrupted
Fault-tolerant batch, ML training
Dedicated Hosts
Physical server dedicated to you
Licensing compliance (Windows Server, SQL Server)
Dedicated Instances
Your instances run on dedicated hardware
Compliance, isolation
Storage
Description
Persistence
EBS (Elastic Block Store)
Network-attached block storage โ primary OS disk
Persists after stop/terminate (configurable)
Instance Store
Physically attached ephemeral storage
Lost when instance stops
EFS (Elastic File System)
Shared NFS storage across multiple instances
Persistent, shared
S3
Object storage accessed via API
Persistent, not a filesystem
Pending โ Running โ Stopping โ Stopped โ Terminated
Stop โ Saves state, no billing for compute (EBS still billed)
Terminate โ Permanently deleted (EBS may be deleted depending on config)
Reboot โ Restarts the OS, instance stays on the same host
Each EC2 instance belongs to a VPC and a Subnet
Security Groups control traffic (default: deny all inbound)
Public instances need an Internet Gateway and a public IP or Elastic IP
Private instances use a NAT Gateway for outbound internet access
ENI (Elastic Network Interface) โ Virtual network card that can be moved between instances
Auto Scaling Groups (ASG) automatically adjust the number of instances based on demand:
Define minimum, desired, and maximum instance counts
Scale out (add instances) on high CPU/memory/custom metric
Scale in (remove instances) when demand drops
Integrated with Elastic Load Balancer (ELB) for traffic distribution
Feature
AWS EC2
Azure VM
VM image format
AMI
Azure Managed Image / Marketplace
Disk type
EBS (gp3, io2)
Managed Disks (Standard SSD, Premium SSD)
Instance metadata
169.254.169.254 IMDS
169.254.169.254 IMDS
Reserved pricing
Reserved Instances / Savings Plans
Reserved VM Instances
Auto scaling
Auto Scaling Group (ASG)
Virtual Machine Scale Sets (VMSS)
Load balancer
ALB / NLB
Azure Load Balancer / Application Gateway
Static IP
Elastic IP
Static Public IP
--image-id ami-0abcdef1234567890 \
--instance-type t3.micro \
--security-group-ids sg-0abc123 \
--subnet-id subnet-0abc123
aws ec2 describe-instances \
--filters " Name=instance-state-name,Values=running " \
--query ' Reservations[*].Instances[*].[InstanceId,InstanceType,PublicIpAddress] ' \
aws ec2 stop-instances --instance-ids i-0abc1234567890
aws ec2 terminate-instances --instance-ids i-0abc1234567890