Billing, Pricing, and Support
Compute and Storage Pricing Options
CoreLearn how commitment, interruption tolerance, tenancy, capacity assurance, data transfer, and storage access patterns change the way AWS resources are priced.
Aligned to the current CLF-C02 exam guide, verified August 15, 2026.
Why this matters
CLF-C02 expects you to choose a purchasing or storage option from business clues without calculating a detailed bill or memorizing prices.
Must Know
- On-Demand Instances require no long-term commitment and suit short, irregular, or uncertain workloads.
- Spot Instances use spare EC2 capacity at a steep discount but can be interrupted, so the workload must tolerate interruption.
- Reserved Instances provide an EC2 billing benefit under offering terms; Savings Plans discount eligible usage in exchange for a consistent spend commitment.
- Dedicated Hosts provide a physical server allocated to one customer and support host-level licensing needs. Dedicated Instances use single-tenant hardware without the same host visibility.
- Capacity Reservations reserve EC2 capacity in a specific Availability Zone. They do not provide a discount by themselves.
- Data-transfer cost depends on direction and location. Data into AWS is often free, while data out or between Regions commonly incurs charges; verify the service pricing page for a real design.
- Storage pricing reflects stored amount plus factors such as access frequency, requests, retrieval, minimum storage duration, resilience, and data transfer.
Compare and Distinguish
- On-Demand vs commitment discounts: On-Demand preserves flexibility; Reserved Instances and Savings Plans trade commitment for eligible discounts.
- Spot vs On-Demand: Spot is lower cost but interruptible; On-Demand is not reclaimed because EC2 needs the capacity back.
- Dedicated Host vs Dedicated Instance: a Host exposes the physical host allocation for licensing or compliance needs; a Dedicated Instance provides single tenancy without host placement control.
- Capacity Reservation vs Reserved Instance: one assures capacity; the other is principally a billing benefit. They can be used together when their attributes align.
- S3 Standard vs infrequent-access/archive classes: lower storage rates can bring retrieval charges, access constraints, or different resilience characteristics.
Scenario examples
- Scenario: A batch job can restart if EC2 reclaims capacity. Think: Spot Instances exchange interruption tolerance for a lower price.
- Scenario: A stable workload can commit to eligible compute spend while retaining flexibility across supported compute usage. Think: Savings Plans.
- Scenario: A licensed product requires visibility into the physical server assigned to the company. Think: Dedicated Hosts, not merely Dedicated Instances.
- Scenario: Monthly archive objects are rarely retrieved and can wait hours when needed. Think: Compare S3 Glacier storage classes by retrieval and access requirements.
Exam traps
- A Capacity Reservation does not automatically lower the hourly compute price.
- A purchasing commitment does not create high availability or reserve capacity in every location.
- Spot is inappropriate when an interruption would violate the stated requirement.
- Data-transfer pricing is not simply “all inbound free and all outbound charged”; paths and services matter.
- The lowest storage rate is not automatically the lowest total cost when retrieval and request patterns are considered.
Key takeaways
- Flexible: On-Demand. Interruptible: Spot. Committed discount: RI or Savings Plans.
- Physical-host need: Dedicated Host. Single tenancy: Dedicated Instance. Capacity assurance: Capacity Reservation.
- Storage economics follow the access pattern, not only the number of gigabytes.
How it works
- AWS meters service consumption and applies the rate or commitment benefit that matches eligible usage.
- Spot capacity can be interrupted when AWS needs it back; applications must checkpoint, retry, or otherwise tolerate that outcome at the architecture level.
- S3 lifecycle rules evaluate objects and transition or expire them according to configured criteria.
- Consolidated billing can apply eligible pricing benefits across linked accounts without combining their identities or resources.
When to use it
- Use On-Demand while demand is uncertain or the workload is short lived.
- Use a commitment discount after stable eligible usage is understood.
- Use Spot for fault-tolerant, flexible work.
- Use dedicated tenancy or capacity assurance only when the requirement names that property.
- Choose a storage class after classifying frequency, retrieval time, resilience, and retention.
Security and governance implications
- Dedicated tenancy can support licensing or isolation requirements, but it does not replace IAM, encryption, or network controls.
- Cost optimization must preserve stated availability, security, compliance, and retention requirements.
Troubleshooting signals
- For an unexpected bill, identify service, Region, usage type, data-transfer path, commitment coverage, and storage retrieval or request activity before assuming a price changed.
- For an unused commitment, compare eligible usage attributes and sharing settings; do not infer that every purchase applies to every compute service.
More detail
- Reserved Instance flexibility depends on the offering. Regional RIs can provide instance-size flexibility within supported families and platforms, while zonal RIs can include capacity reservation behavior. Keep the CLF focus on billing and sharing concepts rather than detailed modification rules.
- Eligible Reserved Instance and Savings Plans benefits can be shared across accounts in an AWS Organization under consolidated billing settings. The accounts remain separate security boundaries.
- Savings Plans apply discounts to eligible usage based on a spend commitment. Current plan types and coverage evolve, so verify product documentation rather than memorizing a static catalog.
- Storage classes serve different access and resilience needs. Lifecycle policies can automate object transitions, but the policy is a management mechanism rather than a compute purchasing contract.
Ready for the quiz?
- Why is a Capacity Reservation different from a Reserved Instance?
- Which workload property makes Spot appropriate?
- When would a Dedicated Host be more relevant than a Dedicated Instance?
- Which storage-access clues matter besides the storage price per GB?
Related objectives
- D4.1