AWS EC2 Launch Types and Tenancy
What you'll learn: How different EC2 purchasing models and tenancy options affect cost, availability, and isolation of your virtual machines.
Understanding Launch Types
AWS offers four main ways to launch EC2 instances, each balancing cost against flexibility:
On-Demand Instances are like renting a hotel room by the hour—you pay for what you use with no commitment. Perfect for unpredictable workloads or testing.
Reserved Instances work like signing a lease: commit to 1 or 3 years and save up to 75% compared to On-Demand. Best for steady, predictable workloads like production databases.
Spot Instances let you bid on unused AWS capacity at up to 90% discount, but AWS can reclaim them with a 2-minute warning. Ideal for flexible, fault-tolerant jobs like batch processing or data analysis.
Dedicated Instances and Dedicated Hosts provide physical server isolation, important for regulatory compliance or licensing requirements.
Tenancy Models
Shared tenancy (default) means your VMs run on physical hardware shared with other AWS customers—but always isolated at the hypervisor level. Think of it as a secure apartment building where you never see your neighbors.
Dedicated tenancy gives you physical hardware that only runs your instances. Like renting an entire building instead of an apartment. It costs significantly more but meets strict compliance requirements.
When to Use Each
Choose On-Demand for flexibility, Reserved for cost savings on predictable loads, Spot for interruptible workloads, and dedicated tenancy only when compliance demands it. Most workloads run perfectly fine on shared tenancy.
Key Takeaway: Launch types control your payment model (On-Demand, Reserved, Spot), while tenancy determines physical isolation (shared vs dedicated)—select based on your workload's cost sensitivity, predictability, and compliance needs.