All Products
Search
Document Center

Server Load Balancer:Pay-as-you-go

Last Updated:Feb 25, 2026

Classic Load Balancer (CLB) supports pay-as-you-go billing, which includes Pay-by-LCU and Pay-by-specification (end-of-sale). You can release a pay-as-you-go instance at any time.

Note

Billing rules

Features

Description

Billing rules

The pay-as-you-go billing method lets you use resources before you pay for them. Bills are generated and fees are deducted from your account after each billing cycle.

Scenarios

The pay-as-you-go billing method is ideal for the following scenarios:

  • Scenarios with fluctuating business usage.

  • Resource usage is temporary and bursty.

Billing cycle

Pay-as-you-go resources are billed hourly (UTC+8). After the fees for a billing cycle are settled, a new billing cycle begins. If you use a resource for less than one hour in a billing cycle, the usage duration is rounded up to one hour. For example, if you create a pay-as-you-go CLB instance at 09:30:00 and release it at 12:30:00, you are charged for four hours, from 09:00:00 to 13:00:00.

Note

If your Alibaba Cloud account balance, including cash coupons, is less than the amount of the bill to be settled, you will receive a text message or email notification about the insufficient balance.

Instance metering methods

Pay-as-you-go CLB instances support two metering methods: Pay-by-LCU and Pay-by-specification (discontinued) . The billable items for a CLB instance vary based on its metering method, network type, and Internet metering method.

Note

In the following tables, a hyphen (-) indicates the item is not charged, and a check mark (✔) indicates the item is charged.

Usage-Based Billing

Instance network type

Internet metering method

Instance performance limit

Instance fee

Public IP retention fee

LCU & Specification fee

Internet network fee

LCU fee

Specification fee

Data transfer fee

Bandwidth fee

Internet

Pay-by-data-transfer

Automatically scales based on usage. Upper limit is subject toInstance Performance Limits.

-

-

Private network

-

Automatically scales based on usage, with upper limits referenced frominstance performance limits.

-

-

-

Pay-by-specification (discontinued)

Instance network type

Internet metering method

Instance performance limit

Instance fee

Public IP retention fee

LCU & Specification fee

Internet network fee

LCU fee

Specification fee

Data transfer fee

Bandwidth fee

Internet

Pay-by-data-transfer

Determined by the purchased specifications. The maximum specification is s3.large.

-

-

Pay-by-bandwidth

Determined by the purchased specifications. The maximum specification is s3.large.

-

-

Private network

-

Determined by the purchased specifications. The maximum specification is s3.large.

-

-

-

Important
  • Internal-facing pay-as-you-go CLB instances do not incur public IP retention fees, data transfer fees, or bandwidth fees. If you attach an independently purchased EIP to an internal-facing CLB instance, the public network fees incurred by the EIP are billed to the EIP and are not included in the CLB bill.

  • Internal network traffic between CLB and backend ECS instances in the same region is free.

Instance fee

The instance fee is the base fee for a CLB instance. This fee is charged for as long as the instance exists (until it is released). All pay-as-you-go instances incur an instance fee.

The billing cycle for CLB instance fee is one hour. If you use an instance for less than one hour in a billing cycle, the usage duration is rounded up to one hour. The billing duration is the period from instance creation to release.

Instance fee = Instance unit price (USD/hour) × Billing duration (hours), where the instance unit price is USD 0.021/hour .

Note

CLB instances created before December 1, 2024, 00:00:00 (UTC+8) are temporarily exempt from instance fees and are free until November 30, 2026, 23:59:59 (UTC+8). For more information, see Extension of the instance fee waiver period for CLB.

Public IP retention fee

The public IP retention fee is the fee for the configuration of the public IP address of a CLB instance. This fee is charged for as long as the public IP address exists, regardless of its usage. Only Internet-facing pay-as-you-go instances incur a public IP retention fee.

The public IP retention fee is calculated based on the unit price of the public IP address and the billing duration.

Public IP retention fee = Unit price of public IP address × Billing duration × Number of instances

You can click to view the regions and unit prices. The pricing information in the following table is for reference only. The actual price is subject to the pricing displayed on the purchase page.

Region

Instance fee (USD/instance/hour)

China (Hangzhou), China (Shanghai), China (Qingdao), China (Beijing), China (Zhangjiakou), China (Hohhot), China (Shenzhen), China (Heyuan), China (Guangzhou), China (Chengdu), China (Ulanqab)

0.003

China (Hong Kong), Japan (Tokyo), UAE (Dubai), South Korea (Seoul)

0.009

US (Silicon Valley), US (Virginia), Mexico

0.005

Philippines (Manila), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), UK (London), Germany (Frankfurt), Thailand (Bangkok)

0.006

SAU (Riyadh - Partner Region)

0.008

LCU fee and specification fee

LCU fee

A Load Balancer Capacity Unit (LCU) is the smallest unit for measuring the resources consumed by a load balancer. The LCU fee applies to the performance capacity units of a pay-by-LCU CLB instance and is billed hourly based on actual usage. Only pay-as-you-go instances billed by LCU incur an LCU fee.

Instance performance limits

Pay-by-LCU instances automatically scale based on usage, and you are billed an hourly LCU fee according to actual LCU consumption. The performance limits are listed in the following table:

Metric

Maximum connections

New connections per second (CPS)

Queries per second (QPS)

Instance performance limits

1,000,000

100,000

50,000

LCU usage definition

CLB LCU fees are charged hourly, with each billing cycle lasting one hour. If you use an instance for less than one hour within a billing cycle, the usage duration is rounded up to one hour.

LCU fee per hour = LCU unit price (USD per LCU) × Number of LCUs consumed per hour

Number of LCUs per hour = max {Number of LCUs consumed by new connections, Number of LCUs consumed by concurrent connections, Number of LCUs consumed by data transfer, Number of LCUs consumed by rule evaluations}

A performance capacity unit (LCU) is a performance metric used to measure the performance of CLB when processing traffic. The performance metrics included in a single LCU depend on the protocol of the CLB listener.

  • For TCP traffic

    Metric name

    Description

    Metering time

    LCU coefficient

    Calculation of LCUs per hour for each metric

    New connections

    The number of new TCP connections processed per second.

    Second

    800

    Within a billing cycle, the system records all new TCP connections per second. Then, the system divides the maximum new connections per second by the LCU coefficient to obtain the number of LCUs for new connections in that billing cycle. The formula is as follows:

    Number of LCUs = Maximum number of new connections ÷ LCU coefficient

    Concurrent connections

    The number of concurrent TCP connections per minute.

    Minute

    100,000

    Within a billing cycle, the system records the concurrent TCP connections per minute. Then, the system divides the maximum concurrent connections per minute by the LCU coefficient to obtain the number of LCUs for concurrent connections in that billing cycle. The formula is as follows:

    Number of LCUs = Maximum Concurrent Connections ÷ LCU coefficient

    Data processing

    The volume of data processed by CLB for TCP requests and responses, in GB.

    Hour

    1 GB

    Within a billing cycle, the system records the total volume of data processed for TCP requests and responses. Then, the system divides the total data volume by the LCU coefficient to obtain the number of LCUs for data processing in that billing cycle. The formula is as follows:

    Number of LCUs = Total processed data amount ÷ LCU coefficient

  • For UDP traffic

    Metric name

    Description

    Metering time

    LCU coefficient

    Calculation of LCUs per hour for each metric

    New connections

    The number of new UDP connections processed per second.

    Second

    400

    Within a billing cycle, the system records all new UDP connections per second. Then, the system divides the maximum new connections per second by the LCU coefficient to obtain the number of LCUs for new connections in that billing cycle. The formula is as follows:

    Number of LCUs = Maximum New Connections ÷ LCU coefficient

    Concurrent connections

    The number of concurrent UDP connections per minute.

    Minute

    50,000

    Within a billing cycle, the system records the concurrent UDP connections per minute. Then, the system divides the maximum concurrent connections per minute by the LCU coefficient to obtain the number of LCUs for concurrent connections in that billing cycle. The formula is as follows:

    Number of LCUs = Maximum concurrent connections ÷ LCU coefficient

    Data Volume Processed

    The volume of data processed by CLB for UDP requests and responses, in GB.

    Hour

    1 GB

    Within a billing cycle, the system records the total volume of data processed for UDP requests and responses. Then, the system divides the total data volume by the LCU coefficient to obtain the number of LCUs for data processing in that billing cycle. The formula is as follows:

    Number of LCUs = Total amount of data transfer/LCU coefficient

  • For HTTP(S) traffic

    Metric name

    Description

    Metering time

    LCU coefficient

    Calculation of LCUs per hour for each metric

    New connections

    The number of new HTTP(S) connections processed per second.

    Second

    25

    Within a billing cycle, the system records all new HTTP(S) connections per second. Then, the system divides the maximum new connections per second by the LCU coefficient to obtain the number of LCUs for new connections in that billing cycle. The formula is as follows:

    Number of LCUs = (Maximum New Connections) ÷ (LCU coefficient)

    Concurrent connections

    The number of concurrent HTTP(S) connections per minute.

    Minute

    3,000

    Within a billing cycle, the system records the concurrent HTTP(S) connections per minute. Then, the system divides the maximum concurrent connections per minute by the LCU coefficient to obtain the number of LCUs for concurrent connections in that billing cycle. The formula is as follows:

    Number of LCUs = Maximum concurrent connections ÷ LCU coefficient

    Processed data volume

    The volume of data processed by CLB for HTTP(S) requests and responses, in GB.

    Hour

    1 GB

    Within a billing cycle, the system records the total volume of data processed for HTTP(S) requests and responses. Then, the system divides the total data volume by the LCU coefficient to obtain the number of LCUs for data processing in that billing cycle. The formula is as follows:

    Number of LCUs = total processed data volume / LCU coefficient

    Rule evaluations

    The product of the number of forwarding rules processed by CLB and the queries per second (QPS). The first 25 forwarding rules are free of charge.

    • If the number of forwarding rules is greater than 25, Rule evaluations = QPS × (Number of forwarding rules - 25).

    • If the number of forwarding rules is less than 25, Rule evaluations = QPS.

    1,000

    Within a billing cycle, the system records all forwarding rule counts and all QPS values, and uses the maximum QPS to calculate the number of rule evaluations. The number of rule evaluations is then divided by the LCU coefficient to obtain the number of LCUs for rule evaluations in that billing cycle. The formula is as follows:

    Number of LCUs = Number of rule evaluations / LCU coefficient

The number of LCUs consumed by each listener in one hour is calculated based on the preceding metrics. You are charged based on the metric that consumes the most LCUs. The total LCU fee for a single CLB instance is the sum of the LCU fees for all its listeners.

LCU unit price

The unit price of an LCU is USD 0.007 per LCU-hour. The actual price is subject to the price on the purchase page.

After the hourly metrics of a CLB instance are converted into LCUs, the hourly LCU consumption is calculated based on actual usage. LCU consumption is accurate to 0.000001 LCU. For example, if you consume 0.1 LCU in an hour, the LCU fee for that hour is 0.1 × 0.007 = USD 0.0007.

LCU billing example

You created a pay-by-LCU CLB instance in the China (Hangzhou) region at 08:10:00 on June 8, 2022, configured a TCP listener and an HTTP listener for it, and then released the instance at 08:50:00 on June 8, 2022. The following table shows sample data for the maximum new connections, maximum concurrent connections, data processing volume, and maximum rule evaluations for the CLB instance during the period from 08:10:00 to 08:50:00 (one billing cycle).

Metric name

TCP

HTTP

New connections (per second)

The maximum number of new TCP connections within the hour is 1,600.

LCU conversion value:

1600÷800=2

The maximum number of new HTTP connections within the hour is 100.

LCU conversion value:

100÷25=4

Concurrent connections (per minute)

The maximum number of concurrent TCP connections within the hour is 480,000.

LCU conversion value:

480,000÷100,000=4.8

The maximum number of concurrent HTTP connections within the hour is 12,000.

LCU conversion value:

12,000÷3,000=4

Data processing (per hour)

The volume of traffic for TCP requests and responses processed in the hour is 4 GB.

LCU conversion value:

4÷1=4

The volume of traffic for HTTP requests and responses processed in the hour is 3.6 GB.

LCU conversion value:

3.6÷1=3.6

Rule evaluations (per second)

0

Assume that 40 forwarding rules are configured for the HTTP listener, and the maximum number of requests received in one second within the hour is 400.

In this example, the number of forwarding rules you configured exceeds the free quota. The maximum number of rule evaluations within the hour is:

(40-25)×400=6,000

LCU conversion value:

6,000÷1,000=6

In this example, the metric that consumed the most LCUs for the TCP listener during the hour is concurrent connections (4.8 LCUs). The metric that consumed the most LCUs for the HTTP listener is rule evaluations (6 LCUs).

LCU fee for the TCP listener = USD 0.007 per unit × 4.8 units = USD 0.0336
LCU fee for the HTTP listener = USD 0.007 per unit × 6 units = USD 0.042
Hourly LCU fee for the CLB instance = LCU fee for the TCP listener + LCU fee for the HTTP listener = USD 0.0336 + USD 0.042 = USD 0.0756
Based on the preceding consumption, the average monthly LCU fee for the CLB instance is estimated as follows: USD 0.0756 × 24 hour × 30 day = USD 54.432

Estimate LCU consumption

You can use the CLB LCU calculator to estimate LCU consumption.

Specification fee

The specification fee applies to pay-by-specification CLB instances and is a fixed charge based on the selected specifications. This fee does not vary with actual usage. Only pay-by-specification pay-as-you-go instances incur a specification fee.

Guaranteed-performance instances are defined by the following three key metrics. Different instance specifications provide different performance metrics. For more information, see Guaranteed-performance instance FAQ.

  • Maximum connections (Max Connection): The maximum number of concurrent connections that a CLB instance can handle. When this limit is reached, new connection requests are dropped.

  • Connections Per Second (CPS): Defines the rate at which new connections can be established. If the rate of new connections exceeds the CPS specified for the instance, additional connection requests are dropped.

  • Queries Per Second (QPS): A metric specific to Layer 7 listeners that represents the number of HTTP(S) requests that can be processed per second. If the request rate exceeds the QPS defined by the instance specification, new connection requests are dropped.

Click to view instance specifications, corresponding load balancing capabilities, and pricing details. The regions and prices in the following tables are for reference only. For actual pricing, see the buy page.

Specification

Maximum connections

New connections per second (CPS)

Queries per second (QPS)

Unit price in the Chinese mainland and China (Hong Kong) (USD/hour)

Unit price in regions outside China (USD/hour)

Specification 1: Small I (slb.s1.small)

5,000

3,000

1,000

0.01

0.012

Specification 2: Standard I (slb.s2.small)

50,000

5,000

5,000

0.05

0.06

Specification 3: slb.s2.medium

100,000

10,000

10,000

0.10

0.12

Specification 4: slb.s3.small

200,000

20,000

20,000

0.20

0.24

Specification 5: Large II (slb.s3.medium)

500,000

50,000

30,000

0.31

0.37

Specification 6: slb.s3.large

1,000,000

100,000

50,000

0.51

0.61

Example: At 10:00:00 on November 20, 2021, you created an Internet-facing instance with the slb.s2.small specification in the China (Hangzhou) region. You then released the instance at 12:34:00 on November 21, 2021. In this case:

Specification unit price for a Standard I (slb.s2.small) instance in the China (Hangzhou) region = 0.05 USD/hour
Usage duration = 27 hours
Specification fee = 0.05 USD/hour × 27 hours = 1.35 USD

Internet network fee

Data transfer fee

A data transfer fee is charged for the outbound Internet traffic (downstream traffic) of an Internet-facing CLB instance that uses the pay-by-data-transfer billing method. Inbound Internet traffic (upstream traffic) is not billed. This fee applies only to pay-as-you-go instances.

Important

For Internet-facing CLB instances that use the pay-by-data-transfer billing method, the peak bandwidth is not a guaranteed performance metric. It serves only as a reference value and an upper limit for bandwidth. The peak bandwidth may be restricted during resource contention. For more information, see Bandwidth Limits.

For Internet-facing CLB instances that use the pay-by-data-transfer billing method, the billing cycle and bill cycle are both 1 hour. If you use an instance for less than 1 hour within a billing cycle, the usage duration is rounded up to 1 hour.

Data transfer fee = Unit price of data transfer (USD/GB) × Data transferred (GB)

The following table lists the unit prices for data transfer. The listed regions and prices are for reference only. For the actual prices, see the buy page.

Region

Data transfer fee (USD/GB)

Hangzhou, Shanghai, Beijing, Zhangjiakou, Hohhot, Shenzhen, Heyuan, Guangzhou, and Ulanqab

0.125

China (Qingdao)

0.113

China (Hong Kong)

0.156

US (Silicon Valley), US (Virginia)

0.078

Malaysia (Kuala Lumpur)

0.112

Philippines (Manila), Singapore, Indonesia (Jakarta), Thailand (Bangkok)

0.117

Japan (Tokyo)

0.087

Germany (Frankfurt), UK (London)

0.070

Mexico

0.078

UAE (Dubai)

0.447

South Korea (Seoul)

0.123

SAU (Riyadh - Partner Region)

0.141

Example: Assume that you created a pay-by-data-transfer Internet-facing CLB instance in the China (Hangzhou) region at 10:00:00 on November 20, 2021 and released the instance at 12:34:00 on November 21, 2021. The total outbound Internet traffic is 5 GB. In this case:

Unit price of data transfer in the China (Hangzhou) region = USD 0.125 per GB
Outbound data transfer: 5 GB
Data transfer fee = USD 0.125 per GB × 5 GB = USD 0.625

Bandwidth fee

The bandwidth fee applies to pay-by-bandwidth Internet-facing CLB instances. This is a fixed fee based on the purchased bandwidth and does not depend on actual traffic usage. Only pay-by-bandwidth Internet-facing pay-as-you-go instances are charged a bandwidth fee.

Important

For Pay-by-Bandwidth Internet-facing CLB instances (Pay-by-Specification, discontinued), the peak bandwidth is a committed performance metric that is guaranteed even during periods of resource contention. For more information, see Bandwidth limits.

For Internet-facing CLB instances that use pay-by-bandwidth, both the billing cycle and the bill cycle are 1 day. If you use the instance for less than 1 hour in a billing cycle, the usage duration is rounded up to 1 hour. If the usage duration is less than one day, you are charged based on the actual number of hours you use that day. The usage duration refers to the period during which the instance uses bandwidth.

  • For Internet-facing CLB instances billed on a pay-by-bandwidth basis, the bandwidth unit price uses tiered pricing with a 5 Mbps threshold.

  • If you change bandwidth within a billing cycle, CLB calculates bandwidth fees based on the maximum bandwidth peak within a day.

Note

For pay-by-bandwidth instances, the inbound peak bandwidth is the same as the outbound peak bandwidth.

For a CLB instance whose maximum bandwidth is from 1 to 5 Mbit/s, the bandwidth fee is calculated using the following formula: Bandwidth fee = Maximum bandwidth × Bandwidth unit price × Usage duration.
For a CLB instance whose maximum bandwidth is higher than 5 Mbit/s, the bandwidth fee is calculated using the following formula: Bandwidth fee = [5 × Unit price of bandwidth from 1 to 5 Mbit/s + (Maximum bandwidth - 5) × Unit price of bandwidth higher than 5 Mbit/s] × Usage duration.

You can click to view bandwidth unit prices. The regions and prices in the following table are for reference only. The prices on the buy page are final.

Region

Bandwidth fee for 1-5 Mbps (USD/hour)

Bandwidth fee for 1-5 Mbps (USD/day)

Bandwidth fee for over 5 Mbps (USD/hour)

Bandwidth fee for over 5 Mbps (USD/day)

China (Hangzhou), China (Shanghai), China (Beijing), China (Zhangjiakou), China (Hohhot), China (Shenzhen), China (Heyuan), China (Chengdu)

0.006

0.144

0.02

0.48

China (Qingdao)

0.005

0.12

0.016

0.384

China (Hong Kong)

0.006

0.144

0.02

0.48

Singapore, Indonesia (Jakarta), Malaysia (Kuala Lumpur), Germany (Frankfurt), UK (London), Thailand (Bangkok), South Korea (Seoul)

0.006

0.144

0.02

0.48

Japan (Tokyo)

0.007

0.168

0.023

0.552

UAE (Dubai)

0.048

11.52

0.118

2.832

US (Silicon Valley), US (Virginia), Mexico

0.006

0.144

0.02

0.48

Example: At 10:00:00 on November 20, 2021, you created an Internet-facing CLB instance with a bandwidth of 2 Mbps in the China (Hangzhou) region. At 08:00:00 on November 21, 2021, you changed the bandwidth to 20 Mbps. You then released the instance at 12:34:00 on November 21, 2021. In this case:

Maximum bandwidth on the day when you purchased the CLB instance: 2 Mbit/s
Maximum bandwidth on the day when you modified the maximum bandwidth of the CLB instance: 20 Mbit/s
Bandwidth fee for the first day: (2 × 0.006) × 14 = USD 0.168
Bandwidth fee for the next day: [5 × 0.006 + (20 - 5) × 0.02] × 13 = USD 4.29
Total fee = 0.168 + 4.29 = USD 4.458

Upgrade/Downgrade Billing Method

The billing method and the procedures to upgrade or downgrade pay-as-you-go CLB instances are as follows:

image
  • Pay-as-you-go instance billing methods include billing by usage and billing by specification (discontinued). CLB supports converting only from billing by specification to billing by usage, not from billing by usage to billing by specification. For more information, see Upgrade/Downgrade Pay-as-you-go Instances.

  • You can upgrade or downgrade the instance type, public network billing method, and bandwidth value for pay-by-specification instances.

  • If you do not change the public network billing method, the change takes effect immediately. If you change the public network billing method, all instance upgrades or downgrades take effect at 00:00:00 the next day, and you cannot perform other upgrade or downgrade operations before they take effect.

Cost optimization suggestions

Strategies to reduce costs

Fee type

Optimization strategy

Internet data transfer fee

Purchase a Data Transfer Plan to cover data transfer cost.

Specification fee

For pay-by-specification instances, adjust the instance type as needed to avoid resource waste.

Idle resources

Promptly release CLB instances and public IP addresses that are no longer in use to avoid unnecessary fees. EIPs associated with private-network CLB instances must be released separately.

Methods to estimate costs

  • You can use the CLB LCU Estimator to estimate LCU consumption.

  • You can estimate future costs based on historical bill data and business growth trends.

How to troubleshoot sudden cost increases

  1. You can use the Cloud Monitor console to view change trends for metrics such as traffic and the number of requests for CLB instances, and identify when costs increased.

  2. You can check the CLB access logs—which must be enabled in advance—and backend service logs to analyze whether unusual traffic or malicious attacks occurred.

  3. In the Expenses and Costs console, compare historical bills to identify the specific billing items that caused the cost increase.

  4. You can set billing alerts to receive automatic notifications when your costs exceed a threshold, which helps you promptly detect abnormal activities.