Purchase Models
Emphasis: Providing big savings through different use cases
Focus on cost optimization through various purchasing options
Ec2 Pricing Options
Amazon EC2 Pricing Options
Section titled “Amazon EC2 Pricing Options”AWS Free Tier: Amazon EC2
Section titled “AWS Free Tier: Amazon EC2”Available to new AWS customers for 12 months following AWS sign-up:
- 750 hours per month of t4g.small instance (region dependent)
- 750 hours per month of Linux, RHEL, or SLES t2.micro or t3.micro instance (region dependent)
- 750 hours per month of Windows t2.micro or t3.micro instance (region dependent)
Amazon EC2 Pricing Models
Section titled “Amazon EC2 Pricing Models”Amazon EC2 provides purchasing strategies to help optimize costs based on needs:
Strategy Categories:
Section titled “Strategy Categories:”Capacity Reserved Models
Emphasis: Providing reserved instances to guarantee availability when needed
Focus on capacity assurance and planning
Dedicated Models
Emphasis: Providing dedicated hardware for compliance and regulation requirements
Focus on isolation and regulatory compliance
Amazon EC2 Purchase Models
Section titled “Amazon EC2 Purchase Models”Pay for compute capacity by the second or by the hour with no long-term commitments
Recommended use cases:
- Spiky workloads
- Experimentation workloads
Benefits:
- Most flexibility with no long-term contract and low rates
- Good for short-term, spiky, or unpredictable workloads
- Developing and testing applications on Amazon EC2 for the first time
Make a 1-year or 3-year commitment and receive significant discount off on-demand prices
Recommended use cases:
- Committed workloads
- Steady-state workloads
Benefits:
- Good choice for predictable or steady-state compute needs
- Substantial savings compared to On-Demand for consistent, heavy use
- Discounted usage price is fixed for as long as you own the Reserved Instance
Same discounts as Reserved Instances with more flexibility in exchange for a $/hour commitment
Recommended use cases:
- All Amazon EC2 workloads
- Amazon EC2 workloads that might need flexibility with committed usage
Types:
- Compute Savings Plans: Most flexibility, up to 66% savings, apply regardless of instance family, size, AZ, Region, OS, or tenancy
- EC2 Instance Savings Plans: Less flexible, up to 72% savings, apply to specific instance family within specific Region
Spare Amazon EC2 capacity at substantial savings off On-Demand Instance prices
Recommended use cases:
- Fault-tolerant workloads
- Flexible workloads
- Stateless workloads
Characteristics:
- Can be interrupted with 2-minute notification
- Workload must have ability to be stopped and restarted
- Runs when capacity is available and maximum price exceeds Spot price
- Spot blocks: Run continuously for finite duration of 1-6 hours without interruption
Amazon EC2 Capacity Reservations
Section titled “Amazon EC2 Capacity Reservations”On-Demand Capacity Reservations
Purpose: Guarantees access to EC2 capacity when needed, for as long as needed
Recommended use cases:
- Workloads that need to meet regulatory requirements for high availability
- Workloads that require capacity assurance
- Business-critical events requiring capacity assurance
- Disaster recovery strategies requiring reserved capacity in different AZ or Region
Amazon EC2 Capacity Blocks for ML
Purpose: Reserve GPU instances for future date to run machine learning (ML) workloads
Recommended use cases:
- Training and fine-tuning ML models
- Running experiments and building prototypes
- Planning for future surges in demand for ML applications
Features: Pay only for compute time needed, no long-term commitment, reserves P5 instances
Amazon EC2 Dedicated Options
Section titled “Amazon EC2 Dedicated Options”Amazon EC2 dedicated options provide EC2 instance capacity on physical servers dedicated for your use (single-tenant hardware):
Characteristics:
- Per-instance billing
- Automatic instance placement
- Isolates the hosts that run your instances
Billing: Two components - hourly per-instance usage fee and dedicated per-Region fee paid once per hour regardless of number of Dedicated Instances running
Characteristics:
- Per-host billing
- Visibility of sockets, cores, and host ID
- Affinity between a host and an instance
- Targeted instance placement
- Add capacity using allocation request
- Use server-bound software licenses and address compliance requirements
Recommended for:
- Users looking to save money on licensing costs
- Workloads that need to run on dedicated physical servers
- Users looking to offload host maintenance onto AWS while controlling maintenance schedules
Billing: Three pricing models - On-Demand, Reservation Pricing, and Savings Plans
Isolation Both options isolate hosts running your instances from hosts running instances for other accounts and provide same performance, security, and physical features.
Amazon EC2 Cost Optimization Guidelines
Section titled “Amazon EC2 Cost Optimization Guidelines”To optimize the cost of Amazon EC2 instances, combine the available purchase options:
Recommended Approach:
Section titled “Recommended Approach:”- Identify steady-state workloads → Use Reserved Instances or Savings Plans
- Stateful spiky workloads → Use On-Demand Instances
- Fault-tolerant, flexible, stateless workloads → Use Spot Instances
Amazon EC2 pricing models include On-Demand Instances, Reserved Instances, Savings Plans, Spot Instances, and Dedicated Hosts. Per-second billing is available only for On-Demand Instances, Reserved Instances, and Spot Instances that run Amazon Linux or Ubuntu. Use a combination of Reserved Instances, Savings Plans, On-Demand Instances, and Spot Instances to optimize Amazon EC2 compute costs.