Docs

Cache for Redis

Network connection management

Modifying bandwidth

Cache for Redis instances of different specifications have varying bandwidths. When the instance default bandwidth cannot meet the peak service traffic requirements (such as time-limited flash sales), you can easily cope with traffic peaks by increasing the read and write bandwidth of the instance separately without changing the instance specifications.

Prerequisite

The instance status is Running. For more details, refer to Instance status.

Bandwidth modification range

Cache for Redis supports different node specifications. The default bandwidths of nodes with different specifications vary, but the ranges of read and write bandwidths that can be modified are the same. Specifically:

Maximum modifiable write bandwidth for Redis instances across all node specifications: 288 MB/s.

Maximum modifiable read bandwidth for Redis instances across all node specifications: 512 MB/s.

tip

The above bandwidth limits are the default limits displayed on the Redis console. If you need to increase the additional read/write bandwidth limits that can be modified, submit a ticket to contact technical support for an application.

For Redis instances with sharded clusters enabled, the above bandwidth limits are the read/write bandwidth limits that can be modified for each single shard.

The overall bandwidth of a Redis instance cannot exceed a maximum of 2.5 GB/s. Therefore, the upper limit of the overall bandwidth of an instance = min[(number of shards x upper limit of bandwidth for a single shard), 2.5 GB/s]. For more details, refer to Bandwidth description.

The Cache for Redis supports a bandwidth throttling mechanism. When the business traffic exceeds the Default bandwidth of the instance, bandwidth throttling will be triggered. If bandwidth throttling occurs, you can resolve it by manually adjusting the bandwidth, optimizing big keys or hot keys, setting up read/write splitting, or scaling up instances. For more details, refer to Suggestions for reasonable instance bandwidth.

Precautions

Billing information

Additional bandwidth is billed on a gradient basis, meaning that the unit price of the bandwidth will increase as the amount of additional bandwidth increases. For more details on additional bandwidth pricing, refer to Additional bandwidth billing rules.

Procedure

In the upper-left corner of the top menu bar, select the project and region to which the instance belongs.

On the Instances page, click the target instance name.

On the Summary tab, in the Configuration section, click the edit icon next to Bandwidth to navigate to the Modify bandwidth page. Configure the following parameters as required.

tip

Hover your mouse over the value after Bandwidth to view the instance's current Default bandwidth and Additional bandwidth information in the pop-up window.

Parameters	Required to config	Description
Instance ID/Name	No	Instance ID and name.
Instance specification	No	The current specifications of the instance, including node specifications and node and shard counts.
Default read bandwidth per shard	No	The default read bandwidth for each data node. Different node specifications have different default bandwidths. For more information, refer to Instance specifications.
Target read bandwidth	Yes	Target read bandwidth: Supports direct input or step adjustment of additional bandwidth per single shard. The right side simultaneously displays the current instance's default bandwidth (not modifiable) and additional bandwidth. The console automatically calculates and previews the total instance bandwidth based on the number of shards. During step adjustment, the single-shard total, total instance bandwidth, and cost are updated in real time, eliminating the need for manual conversion by shard count.
Read bandwidth estimation	No	Simultaneously display post-expansion bandwidth from two dimensions: single-shard read bandwidth (default + additional + total) and total instance read bandwidth. The console automatically calculates the total instance bandwidth based on the number of shards (and multiplies it by the number of readable nodes when read/write splitting is enabled), eliminating the need for manual conversion. tip The estimated read bandwidth displayed here refers to the estimated total read bandwidth of the instance in non-read/write splitting scenarios. The instance's read/write splitting policy and read local node only settings affect the number of readable nodes. Therefore if the instance is configured with read/write splitting policy or read local node only, the total write bandwidth of the instance remains unchanged, and the total read bandwidth is increased by a multiple of the number of readable nodes, so the actual total read bandwidth may be larger than the estimated bandwidth. Different read/write splitting policies result in different calculation methods for instance's total read bandwidth, specifically: Disable read/write splitting: Instance total read bandwidth = Target bandwidth x Number of shards. Enable read/write splitting: Instance total read bandwidth = Target bandwidth x Number of shards x Number of readable nodes.
Default write bandwidth per shard	No	The default write bandwidth for each data node. Different node specifications have different default bandwidths. For more information, refer to Instance specifications.
Target write bandwidth	Yes	Target write bandwidth: supports direct input per shard or incremental adjustment of additional bandwidth. The right panel displays the current instance's default bandwidth (non-editable) and additional bandwidth. The console automatically calculates and previews the total instance bandwidth based on the number of shards. When incrementally adjusting, the total per-shard, total instance bandwidth, and costs update in real-time, eliminating the need for manual conversion based on the number of shards.
Write bandwidth estimation	No	Simultaneously display post-expansion bandwidth from two dimensions: single-shard write bandwidth (default + additional + total) and total instance write bandwidth. The console automatically calculates the total instance bandwidth based on the number of shards, eliminating the need for manual conversion. tip The estimated write bandwidth displayed here refers to the estimated total write bandwidth of the instance in non-read/write splitting scenarios. Enabling or disabling the read/write splitting strategy does not affect how the instance's total write bandwidth is calculated, that is, total write bandwidth = target bandwidth * shard number

The page content is fully displayed on a single screen. Cost information is updated in real time with the configuration in a sticky footer. After confirming the configuration and costs in the sticky cost bar at the bottom, click OK to submit.

Subsequent operations

To reduce costs and save resources, you can decrease additional bandwidth after the peak in service traffic. You can monitor instance traffic usage using metrics such as Outbound bandwidth utilization, Inbound bandwidth utilization, Network peak input rate and Network peak output rate. For specific procedures, refer to Viewing monitoring data.

If business traffic exceeds the instance's default bandwidth, leading to issues like bandwidth throttling, refer to Suggestions for reasonable instance bandwidth for optimization.

Relevant APIs

API	Description
ModifyDBInstanceAdditionalBandwidthPerShard	Call the ModifyDBInstanceAdditionalBandwidthPerShard API to modify the single shard bandwidth for target Redis instance.
DescribeDBInstanceBandwidthPerShard	Call the DescribeDBInstanceBandwidthPerShard API to query the details of the bandwidth per shard of target Redis, including the default bandwidth and the additional bandwidth.

Last updated: 2026.07.31 11:59:30