Cloud VM Pricing Models: AWS vs Azure

AWS EC2 Pricing Models

• On-Demand Instances: Pay per second for compute with no long-term commitment.

• Savings Plans: Commit to a consistent USD/hour spend for 1 or 3 years; applies across instance families to reduce costs.

• Reserved Instances: Commit to specific instance attributes (type, region, tenancy, OS) for 1 or 3 years for discounted pricing.

• Spot Instances: Use unused capacity at steep discounts; capacity can be reclaimed by AWS with short notice.

Azure VM Pricing Models

• Pay-As-You-Go: On-demand pricing with no commitment.

• Savings Plan: Commit to a consistent spend for 1 or 3 years to lower prices across eligible VMs.

• Reservations: Pre-purchase specific VM resources for 1 or 3 years for discounts.

• Spot: Access unused capacity at reduced rates; eviction occurs when Azure needs capacity.

Key Pricing Components

• CPU: Hourly cost scales with vCPU count and generation.

• RAM: Hourly cost scales with GB of memory.

• Network Speed: Instance class and region affect included network bandwidth.

• Note: Costs vary by region and OS; review regional pricing before deployment.

Units of Importance for VM Selection

• CPU: vCPU count and clock speed.

• RAM: GB of memory.

• Hard Disk: Capacity (GB), IOPS, throughput (MB/s).

• GPU: Core count, memory capacity, and clock speed.

Selection Tips

• Match workload needs to instance family (compute-optimized, memory-optimized, storage-optimized, GPU).

• Use On-Demand for spiky or short-term workloads; Savings Plans/Reservations for steady-state.

• Use Spot for fault-tolerant, interruptible jobs.

• Validate network and storage performance against application requirements.

Additional Resources

• AWS Pricing Calculator: Estimate EC2 and related service costs.
  https://calculator.aws/

• azure pricing calculator : Estimate VM and related service costs azure-pricing-calculator

Create Image in Azure

• please follow class recording to create image in Azure