Cloud pricing models represent the financial framework through which organizations consume and pay for compute, storage, and services from providers like AWS, Azure, and GCP. Commitment-based pricing—reserved instances, savings plans, and committed use discounts—offers up to 72% savings over on-demand rates in exchange for 1-year or 3-year usage commitments. Understanding the tradeoffs between flexibility, savings, and risk is critical: on-demand pricing maximizes elasticity with no long-term lock-in, while commitments provide substantial discounts but require accurate capacity forecasting. The key mental model is coverage vs. utilization—commitments should cover your baseline steady-state workload, leaving variable spikes to on-demand or spot instances, ensuring you maximize discount application without overcommitting to unused capacity.
What This Cheat Sheet Covers
This topic spans 20 focused tables and 105 indexed concepts, 88 flashcards. Below is a complete table-by-table outline of this topic, spanning foundational concepts through advanced details.
A jump-to index of every table row in this cheat sheet.
An interactive map of every table and concept in this topic.
Table 1: Core Pricing Models
Every cloud bill is built from a small set of fundamental pricing constructs. Knowing the tradeoffs between each — flexibility, savings depth, and interruption risk — is the foundation of any cost strategy.
| Model | Example | Description | |
|---|---|---|---|
AWS EC2: $0.096/hour for t3.medium No commitment required | • Pay-as-you-go with no long-term contracts or upfront costs • charged by the hour or second • highest per-unit cost but maximum flexibility for unpredictable workloads | ||
AWS: 1-year Standard RI m5.large, us-east-1 Up to 72% discount vs on-demand | • Commit to a specific instance type, region, and term (1 or 3 years) for significant discounts • locks in configuration but offers highest savings for predictable workloads | ||
AWS Compute Savings Plan $5/hour commitment, 3-year Up to 66% discount; covers EC2, Fargate, Lambda | • Commit to a consistent dollar-per-hour spend on compute • more flexible than RIs—applies across instance families, sizes, regions, and services • slightly lower maximum discount than RIs but much greater adaptability | ||
AWS Spot: $0.0096/hour (~90% off) 2-minute interruption notice | • Purchase unused cloud capacity at steep discounts (up to 90% off) • instances can be reclaimed with short notice • ideal for fault-tolerant, stateless, or batch workloads | ||
GCP: 3-year resource CUD Commit 10 vCPUs in us-central1 Up to 57% discount | • GCP's commitment model—resource-based CUDs lock specific vCPU/memory/GPU in a region; spend-based (Flex) CUDs offer flexible dollar-per-hour commitments across services | ||
Azure: 3-year VM Reservation Standard_D4s_v3, East US Up to 72% savings (Linux) | • Azure's RI equivalent—commit to VM family, size, and region for 1 or 3 years • supports exchange and trade-in to savings plans; cancellations subject to $50,000 rolling refund cap | ||
GCP Compute Engine Run VM 25%+ of month → automatic discount Up to 30% off; no commitment required | • GCP automatically applies incremental discounts as VM runtime increases through the month • no upfront commitment; stacks with CUDs for compounded savings |