Alibaba Cloud billing consists of two parts: billing items (what you pay for) and billing methods (how you pay).
Total cost = resource usage per billing item x unit price. Choose the right billing method to control your cloud spending.
Billing methods
Alibaba Cloud offers two billing methods: pay-as-you-go and subscription. Subscription includes subscription instances, Savings Plans, and resource plans.
|
Billing method |
Definition |
Features |
|
|
Pay-as-you-go |
Use resources first, pay later. Billed hourly or by the second. |
Flexible. Release resources at any time. |
|
|
Subscription |
Subscription |
Prepaid. Get stable service for a fixed period. |
Exclusive resources, more cost-effective than pay-as-you-go. |
|
Savings Plan |
Commit to a fixed-term spend for deeper discounts than pay-as-you-go. |
Offsets costs across different instances under the same product. |
|
|
Resource plan |
Pre-purchase a fixed amount of usage (storage, traffic) to offset pay-as-you-go charges. |
Directly offsets resource usage. |
|
Pay-as-you-go
Pay-as-you-go is a post-paid billing method that charges based on resource runtime, data volume processed, or number of requests. Resources are available immediately with no upfront planning required, but unit prices are higher than other billing methods.
Pay-as-you-go resources are metered by the second. Billing details are generated hourly or daily, and the monthly bill is issued on the 3rd of the following month.
Key considerations:
Hourly billing data may be delayed.
Billing stops after you release the resources.
When releasing a primary instance, check whether associated resources such as Elastic IP Addresses (EIPs) and snapshots are also released to avoid unexpected charges. Enable spending alerts and use budget management tools to monitor expenses.
Subscription instances
Subscription instances let you prepay for a fixed duration to get lower unit prices. Longer terms yield deeper discounts. Resources are dedicated, immediately available after purchase, and provide predictable costs.
Key considerations:
Resources stop and data is deleted upon expiration. Enable auto-renewal to prevent service interruptions.
Refund amounts are calculated based on usage duration and applicable discounts.
Savings Plan
A Savings Plan is a discount benefit that lets you commit to a minimum hourly spend over a fixed term (1, 3, or 5 years) in exchange for discounted pay-as-you-go prices on eligible resources.
When combined with pay-as-you-go, a Savings Plan provides significant discounts while keeping resources flexible.
Key considerations:
Savings Plans do not support refunds. Confirm the offset scope before purchase.
You are charged the committed amount regardless of actual usage.
Usage beyond the committed amount is billed at standard pay-as-you-go rates.
Resource plans
A resource plan is a prepaid billing method that automatically offsets usage at prices lower than pay-as-you-go.
Two types are available:
Capacity-based: You define a total usage allowance at purchase. The allowance decreases as you consume resources and resets to zero upon expiration.
Usage-based: The allowance resets each period (hourly, daily, or monthly). Unused portions do not carry over to the next period.
Key considerations:
Allowances reset to zero upon expiration and cannot be carried over or extended.
Resource plans offset only specific products and billing items. Confirm the offset scope before purchase.
Some resource plans require specific conditions to take effect.
Refund support varies by product.
What is the difference between Savings Plans and resource plans?
A Savings Plan is spending-based: you commit to an hourly spend amount on compute resources such as ECS, ECI, and ApsaraDB RDS without locking into specific instance types. This provides flexibility across different configurations.
A resource plan is usage-based: you purchase a fixed quantity of storage or network resources such as OSS storage plans or CDN data transfer plans, measured in gigabytes or offset counts.
Billing items
A billing item is the smallest metering unit for Alibaba Cloud resources and services. Each billing item corresponds to a distinct type of resource consumption with its own metrics, billing cycle, and unit price.
Billing items fall into two categories:
Basic billing items: Charges based on compute, storage, and network usage. Examples include instance type fees (ECS), storage fees (disks, OSS), data transfer fees (outbound Internet traffic, CDN), request fees (API calls), and compute fees (CPU and memory duration).
Value-added billing items: Fees for advanced capabilities such as image processing, transfer acceleration, and software subscriptions.
Billing items combine with billing methods to calculate usage and generate bills. Final cost = usage of each billing item x unit price.
Choose a billing method
The key to choosing a billing method is understanding your workload patterns.
Stable workloads suit prepaid methods to lock in costs. Fluctuating workloads suit pay-as-you-go for elasticity. Combine methods to balance cost and flexibility.
Stable workloads: lock in costs with subscription instances
For systems with fixed configurations that run continuously, such as enterprise websites and core databases. Multi-year purchases receive higher discounts and ensure high availability.
Elastic workloads: reduce costs flexibly with Savings Plans
For scenarios that require frequent configuration changes, such as fast-iterating Internet applications and growing SaaS services. Balances discounts with upgrade and downgrade flexibility, avoiding refund losses from subscription changes.
Fluctuating workloads: elastic response with layered cost reduction
For workloads with periodic load fluctuations, such as daytime peaks and nighttime lows, or busy weekdays and idle weekends:
Base load: Use a Savings Plan to lock in costs.
Fluctuating portion: Use pay-as-you-go for on-demand scaling.
This approach balances cost optimization and resource efficiency, preventing overpayment for peak loads or idle resources during off-peak periods.
Burst workloads: start and stop on demand with zero idle costs
For unpredictable traffic surges such as sales promotions, marketing campaigns, and trending events. Use pay-as-you-go to scale out before events and release resources immediately afterward. No advance planning required, no idle costs incurred.
Exploratory workloads: experiment flexibly and pay on demand
For early-stage projects with uncertain usage, such as MVP validation, technology evaluation, and POC testing:
Use pay-as-you-go. Start resources as needed, release when finished, and pay only for actual usage.
After validation, migrate to subscription instances or a Savings Plan for lower costs.
Storage and data transfer: batch purchase with automatic offset
For stable, predictable usage such as log storage, data backup, and CDN data transfer. Estimate average monthly usage from historical consumption and purchase a matching resource plan. Offsets apply automatically with no manual intervention, offering better value than pay-as-you-go.
Combination strategies
Most production environments combine multiple billing methods. A typical strategy:
Core stable load: Use subscription instances or a Savings Plan to lock in base costs.
Elastic scaling portion: Use pay-as-you-go to handle peak demands.
Storage and data transfer: Use resource plans for bulk purchases to get discounts.
Stable workloads: lock in costs
For production systems with fixed configurations that run continuously, such as enterprise websites and core databases, subscription instances are the recommended choice. Multi-year purchases typically receive greater discounts. These workloads require high resource availability and have a low frequency of configuration changes.
For long-running workloads that require frequent configuration changes, such as fast-iterating Internet applications and growing SaaS services, a Savings Plan combined with pay-as-you-go is more suitable. This combination retains upgrade and downgrade flexibility while providing discounted pricing, avoiding the refund losses associated with subscription instance changes.
Fluctuating workloads: elastic response
When workload patterns show clear periodic fluctuations, such as daytime peaks and nighttime lows or busy weekdays and idle weekends, adopt a layered strategy: purchase a Savings Plan for base load costs and use pay-as-you-go for the fluctuating portion. This avoids overpaying for peak loads and wasting resources during off-peak periods.
For unpredictable traffic surges such as e-commerce promotions, marketing campaigns, and trending events, pay-as-you-go is the best choice. Scale out before the event to handle traffic spikes and release resources immediately afterward. No advance planning required, no idle resource costs.
Exploratory workloads: experiment flexibly
When usage is uncertain during early project stages, such as new product MVP validation, technology evaluation, or POC test environments, pay-as-you-go minimizes experimentation costs.
Create test environments on demand, release them after testing, and pay only for actual usage. After validating the business model and clarifying resource requirements, switch to a prepaid plan for better cost efficiency.
Storage and data transfer: batch purchase
For stable, predictable storage and data transfer usage such as log storage, data backup, and network acceleration traffic, resource plans offer the best value.
Estimate average monthly consumption from historical data and purchase a matching resource plan. Offsets apply automatically with no additional steps required.