Configure flow control rules

Prerequisites

Connect your application to AHAS Application Protection. For more information, see Application Access Method.

Background information

In networking, traffic control manages the rate of data transmission. Because the volume of incoming requests can be unpredictable and system capacity is finite, you must control traffic to match your system's processing capabilities.

Create a flow control rule

1. Log on to the AHAS console.

2. In the upper-left corner of the AHAS console, select the region where your application is deployed.

3. In the left-side navigation pane, choose Traffic Protection > Application Protection.

4. On the Application Protection page, click the target application card on the Application Card tab.

5. Use one of the following methods to open the New Flow Control Rule page:

   • In the left-side navigation pane, click API Details. On the API Details page, click the or icon in the upper-right corner of a resource card, and then in the Manage Rules dialog box, click the New Flow Control Rule tab.

   • In the left-side navigation pane, click Application Overview. Then, find the target API and click Flow Control in the Actions column.

   • In the left-side navigation pane, click Rule Settings. On the Flow Control Rules tab, click New Flow Control Rule in the upper-right corner.

6. In the New Flow Control Rule dialog box, configure the rule parameters.

   • Standalone mode

     

     Parameter	Description
     API name	The resource to apply flow control to.
     Source application	The application that originates the calls. The default value is default, which means the rule applies to calls from any source application. For a Dubbo service, enter the Dubbo applicationName of the caller. The calling application must also be connected to Sentinel. For a web service, see Extended APIs. If you are using annotations or custom instrumentation, see Common classes and their methods. ContextUtil only takes effect at the call chain entry, which is the first instrumented point. For other applications under your account that are already connected to Traffic Protection, entering default means no specific source is distinguished. Important To maintain optimal performance, we recommend keeping the number of source applications below 200 to ensure memory usage remains manageable.
     Statistic dimension	Select a statistic dimension to control traffic. Current API: Directly controls traffic from the specified source application. If Source application is set to default, the rule applies to traffic from all sources. This is typically used to ensure a smooth, steady flow of traffic. For more information, see Use cases. Associated API: Controls the traffic of a resource that is associated with the current resource. This is typically used to reserve resources for high-priority APIs during periods of resource contention. For more information, see Use cases. Invocation link entry: Controls the entry traffic of the call chain where the resource is located. After you select Invocation link entry, you must also configure the Entry resource, which is the context name of the invocation link entry. This is typically used in scenarios such as warm-up to avoid traffic surges. For more information, see Common scenarios.
     QPS threshold	The QPS threshold that triggers the flow control rule for the specified statistic dimension.
     Flow control effect	Select a flow control effect to handle throttled requests. Fail-fast: When the threshold is reached, requests are immediately rejected. The system returns a response based on the adapter module settings in your application configuration. Warm up: You must set a specific warm-up period. For more information, see Warm up. If your system has been idle for a long time, this behavior gradually increases the request processing rate when traffic suddenly increases. The rate slowly ramps up to the configured QPS threshold over the specified warm-up period. By default, the rate starts at one-third of the QPS threshold. Excess requests are rejected using the fail-fast behavior. Queue and wait: Requests are processed at a steady rate. Excess requests are queued and processed later. This is often used for peak shaving. You must set a timeout period. If a request remains in the queue longer than the timeout period, it is rejected. For more information, see Peak shaving.
     Enable	Turns the rule on or off. Changes take effect immediately.

   • Cluster flow control mode

     

     Parameter	Description
     Enable cluster flow control	Enables cluster flow control to limit the total number of calls to this resource across the entire cluster.
     Total cluster QPS for the API	The estimated maximum QPS for the resource across the cluster. This value serves as a reference for the Token Server's automatic token allocation. Requests that exceed this value will degrade to standalone flow control.
     Threshold mode	You can set either a cluster QPS threshold or a standalone QPS threshold. Cluster QPS threshold: The threshold applies to the entire cluster. Standalone QPS threshold: The threshold applies to a single machine. The Token Server calculates the total cluster threshold based on the number of connected machines.
     Failure degradation strategy	Specifies the behavior when the application cannot connect to the Token Server or when the Token Server is unavailable. Degrade to standalone flow control: In case of a connection failure, the system degrades to standalone flow control based on a local threshold. You must set the standalone degradation threshold, which acts as the fallback limit for a single machine. Pass directly: In case of a connection failure, the system ignores the failure and allows all requests to pass.

7. Click Create.

Use cases

The following examples show how to configure flow control rules for common use cases.

• Scenario 1: Ensure a smooth traffic flow

  Request traffic often has peaks and valleys. By delaying traffic from peak periods for later processing, this strategy, also known as peak shaving, helps maximize throughput while maintaining a good user experience. For more information, see Peak shaving.

  For example, set Statistic dimension to Current API and Flow control effect to Queue and wait. Set the QPS threshold to 5. The system will then process one request every 200 ms. Excess requests are queued. If you also set the Wait time to 5s, any request expected to wait longer than 5s will be rejected immediately. See the following figure.

  

• Scenario 2: Prioritize high-priority APIs during resource contention

  Assume you have two resources, read_db and write_db, that handle database read and write operations. To prevent data loss, the write_db API has a higher priority. When write operations become too frequent, read requests should be throttled. This strategy is known as associated flow control. For more information, see associated flow control.

  For example, with Statistic Dimension set to Associated API, Flow Control Effect set to Fail-fast, and the Concurrency Threshold set to 10, read_db is throttled if the QPS of the write_db resource exceeds 10. This reserves sufficient resources for write_db and prevents write_db data loss. As shown in the following figure.

  

• Scenario 3: Use warm up to prevent traffic surges

  The warm up behavior prevents a sudden traffic surge from overwhelming your system, especially after a period of inactivity. It gradually increases the allowed traffic from a lower initial threshold up to the configured maximum over a set warm-up period. This is ideal for handling bursty traffic. For more information, see Warm up.

  For example, set Statistic dimension to Invocation link entry and Flow control effect to Warm up. Set the QPS threshold to 60 and the Warm-up time to 2s. By default, the warm-up process starts at one-third of the QPS threshold. In this example, when the entry-point QPS exceeds 20 (60 / 3), the allowed rate gradually increases to 60 over the 2-second warm-up period. See the following figure.

  

Manage flow control rules

The Flow Control Rules page lets you enable, disable, edit, or delete rules.

• Enable or disable a single rule:

  On the Flow Control Rules page, find the target rule and click the switch in the Status column to enable or disable it.

• Enable or disable multiple rules in batch:

  On the Flow Control Rules page, select the checkboxes for the rules you want to manage, and then click Enable in Batch or Disable in Batch.

• Edit a rule:

  On the Flow Control Rules page, find the target rule and click Edit in the Actions column to modify its settings.

• Delete a rule:

  On the Flow Control Rules page, find the target rule and click Delete in the Actions column.