All Products
Search
Document Center

Tair (Redis® OSS-Compatible):Enable bandwidth auto scaling

Last Updated:Mar 28, 2026

When your instance reaches its default bandwidth limit, traffic is throttled and service performance degrades. Bandwidth auto scaling monitors your instance's bandwidth usage in real time and adjusts bandwidth up or down automatically — no manual intervention or connection interruption required.

Prerequisites

Before you begin, make sure that:

How it works

The system keeps actual bandwidth usage between the increase threshold and the decrease threshold you set — similar to a thermostat that maintains temperature within a target range. When usage crosses a threshold, the system calculates a new target bandwidth and scales accordingly.

Scaling formula:

Target bandwidth (MB/s) = Actual bandwidth usage (MB/s) / ((Increase threshold + Decrease threshold) / 2)

Example: An instance has a default bandwidth of 96 MB/s, an increase threshold of 70%, a decrease threshold of 30%, and an observation window of 15 minutes.

  • When Average Bandwidth Usage >= 70%, the system scales up: target bandwidth = (96 × 70%) / ((70% + 30%) / 2) = 135 MB/s.

  • If Average Bandwidth Usage then drops to <= 30%, the system scales back down to the default (96 MB/s minimum).

Scaling limits:

  • Maximum increase: 6x the instance's default bandwidth, up to a maximum increment of 192 MB/s

  • Minimum after decrease: the default bandwidth of the instance type

Behavior by architecture

ArchitectureScaling behavior
Cloud Native read/write splittingScales bandwidth of all nodes based on the node with the highest bandwidth usage
Cluster or Classic read/write splittingMonitors and scales at the data shard or read replica level; each node scales independently

Use cases

ScenarioDescription
Handle traffic spikesTemporarily increase bandwidth during flash sales or planned high-traffic events, then scale back down to reduce costs.
Mitigate big key impactsWhen many read/write operations on big keys create short-term bandwidth pressure, auto scaling buys time while you address the underlying issue.
Handle request skew in distributed architecturesIn cluster or read/write splitting architectures, some data shards or read replicas can hit their bandwidth limits while others remain idle. Auto scaling increases bandwidth only for the hot nodes, avoiding the need to upgrade the entire instance.

Billing

Billing is hourly, based on the additional bandwidth you use beyond the default. The default bandwidth included with your instance type is not charged. Pricing varies by region. For details, see Billable items.

Limitations

  • After an automatic scale-up, the system waits at least 1 hour before triggering an automatic scale-down.

  • There is a 1-minute cooldown between two consecutive automatic scale-up operations.

  • Bandwidth can be increased by up to 6x the default bandwidth of the instance type, with a maximum increment of 192 MB/s. Examples: If you need higher bandwidth than these limits allow, consider upgrading to Tair (Enterprise Edition), where each instance type supports a maximum bandwidth of at least 96 MB/s. You can also upgrade the instance type or switch to a cluster architecture. See Instance types for details.

    • A 2 GB standard Tair memory-optimized instance has a default bandwidth of 96 MB/s. Its maximum bandwidth is 96 + 192 = 288 MB/s.

    • A 256 MB standard Redis Community Edition instance has a default bandwidth of 10 MB/s. The maximum additional bandwidth is 60 MB/s, so the maximum is 70 MB/s.

  • Do not manually adjust instance bandwidth while bandwidth auto scaling is enabled. The two features interact as follows: Examples:

    • Default bandwidth: 10 MB/s, manually set to 70 MB/s — scale-up is never triggered (already at the 6x limit).

    • Default bandwidth: 10 MB/s, manually set to 40 MB/s — scale-up can increase to 70 MB/s; scale-down reduces to 10 MB/s.

  • If the instance has unexpired bandwidth plans, you cannot enable bandwidth auto scaling. First unsubscribe from those plans. See Unsubscription management.

  • The following operations disable bandwidth auto scaling. Re-enable the feature afterward if needed:

    OperationException
    Major version upgradeNone
    Change the configurations of an instanceFor standard architecture instances, bandwidth settings remain valid after a spec change.
    Change the zone of an instanceFor standard architecture instances, bandwidth settings remain valid after a zone migration.

Enable bandwidth auto scaling

  1. Log on to the console and go to the Instances page. In the top navigation bar, select the region where your instance resides, find the instance, and click the instance ID.

  2. In the Configuration Information section, click Modify next to Bandwidth.

  3. Turn on the Auto Bandwidth Scaling switch.

    The first time you access the Database Autonomy Service (DAS) console, follow the on-screen instructions to grant the required permissions. The system automatically grants the AliyunServiceRoleForDAS service-linked role to DAS.
  4. In the DAS console dialog box, configure the scaling policy:

    Average Bandwidth Usage is the larger of the inbound and outbound traffic average usage. The maximum bandwidth increment is 6x the default bandwidth, up to 192 MB/s. The current dialog box also shows the calculated maximum for your instance type.
    CategoryParameterDescription
    Automatic Bandwidth UpgradeAutomatic Bandwidth UpgradeSelect the check box to enable automatic scale-up.
    Average Bandwidth Usage Not Less ThanThe increase threshold. When the larger of inbound and outbound traffic average usage meets or exceeds this value, scale-up is triggered. Valid values: 50%–95%.
    Observation WindowThe time window over which average usage is measured before scaling is triggered. A shorter window reacts faster to spikes; a longer window avoids unnecessary scaling for brief bursts.
    Automatic Bandwidth DowngradeAutomatic Bandwidth DowngradeSelect the check box to enable automatic scale-down. Requires Automatic Bandwidth Upgrade to be enabled first.
    Average Bandwidth Usage Not Greater ThanThe decrease threshold. When average usage drops to or below this value, scale-down is triggered. Valid values: 10%–70%. Must be at least 10% lower than the increase threshold.
  5. Click OK. In the Tair console, the Bandwidth Auto Scaling switch turns on, confirming that the feature is active.

  6. (Optional) Configure alerts to receive notifications when auto scaling events occur. Follow the system prompts, or configure alerts manually:

    1. Select the system-recommended alert template. The system adds the Auto Scaling Event monitoring item.

    2. Select the Alert Contact Group to receive notifications.

    3. Click Submit Configuration and confirm in the dialog box.

    If you have already configured an alert template for the instance, the alert prompt is not shown. To configure alert templates and rules manually, see Configure an alert template and Configure alert rules.

What's next

To schedule a bandwidth upgrade for a specific time window instead of using continuous auto scaling, see Schedule temporary bandwidth upgrades for a Tair instance.