Cloud Pricing Models

How you pay

Spot capacity can be taken back with short notice (two minutes on AWS, about 30 seconds on Azure and Google Cloud), so only use it for work that can be interrupted and retried.

What you're charged for

• Compute per second or hour, storage per GB-month, requests for some services.
• Data transfer: traffic out to the internet, between regions and often between zones. Inbound is usually free. This surprises many teams.
• Managed-service units (load balancer capacity units, database I/O, function invocations and GB-seconds).

HOURS = 730                                    # hours in an average month
on_demand_rate = 0.096                         # example: a 2 vCPU / 8 GB instance, USD per hour
plans = {"on-demand": 1.0, "3-year savings plan": 0.45, "spot (when available)": 0.30}

baseline, peak_extra, peak_hours = 6, 10, 120   # 6 instances always on, 10 more during peaks
for name, factor in plans.items():
    cost = baseline * HOURS * on_demand_rate * factor
    print(f"baseline on {name:22}: ${cost:8,.2f} / month")
mix = baseline * HOURS * on_demand_rate * 0.45 + peak_extra * peak_hours * on_demand_rate
print(f"smart mix (plan for baseline, on-demand for peaks): ${mix:,.2f} / month")

baseline on on-demand             : $  420.48 / month
baseline on 3-year savings plan   : $  189.22 / month
baseline on spot (when available) : $  126.14 / month
smart mix (plan for baseline, on-demand for peaks): $304.42 / month

Use the providers' calculators (AWS Pricing Calculator, Azure Pricing Calculator, Google Cloud Pricing Calculator) before building, and budgets with alerts after.