1 AWS Pricing Fundamentals
Learn how AWS charges for measurable resource usage, how common compute, storage, and data-transfer costs vary, and how to estimate a workload’s bill.
How AWS pricing is measured
AWS pricing is generally : charges reflect the resources and activity used rather than the purchase and maintenance of equivalent infrastructure in advance. The actual charge depends on the AWS service, Region, configuration, and pricing option. Rates and service terms can change, so check the current service pricing page when estimating costs.
A service’s bill may include one or more measurable pricing dimensions:
Time: instance-seconds or hours, or how long a service runs.
Capacity: provisioned compute, database capacity, or other reserved resources.
Storage: the amount of data stored over time, often measured in , together with the selected .
Requests and operations: counts of reads, writes, API calls, or other actions.
Data transfer: the amount of data moved, its direction, and where it travels.
Configuration: factors such as Region, instance type, operating system, license, or redundancy level.
Some services use tiers, so the unit rate can change at different usage levels. A solution’s total cost may combine charges from several services and pricing dimensions.
Compute costs and purchase options
Compute charges commonly depend on the resource type and how long it runs. For example, an Amazon EC2 instance’s rate varies with its instance type, Region, and operating system or software license. Stopping or terminating resources when they are not needed can reduce usage-based compute charges, although attached resources such as storage may continue to incur costs.
EC2 purchase options suit different workload needs:
supports flexible usage without a long-term commitment.
and can lower eligible costs in exchange for a usage or configuration commitment.
use spare capacity at a discount, but may be interrupted. They suit workloads that can tolerate interruptions.
Compare options against workload needs rather than assuming a discount will always reduce total cost.
Storage costs and access patterns
Storage is not always priced as a single fee per GB. Amazon S3 charges can depend on the volume and duration of storage, , requests, retrievals, and certain data-management or replication features. Less frequently accessed storage classes may have different retrieval charges or minimum-storage-duration rules.
Choose a based on access patterns and the full set of applicable charges, not just the storage rate. For example, an archival class may reduce storage costs for a bucket holding archived data, while retrieving data or moving objects between classes can add charges. A frequently accessed dataset may be better suited to a class with different retrieval economics.
Data-transfer costs
depends on the service and route. In general, transfer into AWS from the internet is not charged, but this is not a universal rule for every service or transfer path. Transfer out to the internet may be charged by volume. Transfers between Availability Zones or Regions may also incur charges; rates and exceptions vary, so consult the relevant service pricing page. For some transfers, charges are associated with the source side.
A workload that repeatedly sends large files from AWS to users can incur data-transfer-out costs even if its compute and storage usage stays constant. Keeping data close to the services that use it, reducing unnecessary transfers, or using a delivery service may affect total cost; compare applicable prices for the design.
Estimating a workload
Break the design into services and estimate each service’s billable units: compute runtime and configuration, stored data and , request volume, and expected transfer paths and volumes. Apply current rates for the intended Region and purchase option, and account for any tiers or additional features.
can help create an estimate, but actual charges depend on real usage and applicable terms. Verify current service-specific prices before deployment.