Traffic spikes from live streaming events, flash sales, or game launches can saturate your instance's fixed bandwidth within seconds, causing slow responses or connection drops. Burstable bandwidth automatically adds extra bandwidth when traffic exceeds your fixed allocation and releases it when traffic subsides — so you get the performance headroom you need without over-provisioning.
Key concepts
| Term | Definition |
|---|---|
| Fixed bandwidth | The total non-burstable bandwidth available to an instance. Includes the base bandwidth from the instance specification and any additionally purchased fixed bandwidth. |
| Burstable bandwidth | Extra bandwidth automatically added when traffic exceeds fixed bandwidth. Billed separately based on actual usage. |
| Per-node bandwidth cap | The maximum total bandwidth per single data node is 288 MB. This cap applies to fixed bandwidth and burstable bandwidth combined. |
Supported instances
Burstable bandwidth is available only for:
-
Cloud-native Redis Open-Source Edition
-
Tair (Enterprise Edition) memory-optimized instances
-
Tair (Enterprise Edition) persistent memory-optimized instances
Choose between burstable bandwidth and Bandwidth Auto Scaling
These two features are mutually exclusive — enable only one per instance.
| Feature | Best for | Scaling speed | Cost | Scope |
|---|---|---|---|---|
| Burstable bandwidth (this topic) | Unpredictable, instantaneous spikes lasting seconds or minutes. No threshold configuration required — the system scales based on real-time usage. | Seconds | 2.5x the price of Bandwidth Auto Scaling | Only available in cloud-native deployment mode. |
| [Bandwidth Auto Scaling](https://www.alibabacloud.com/help/en/redis/user-guide/enable-bandwidth-auto-scaling) | Predictable, periodic load changes lasting hours or longer. Requires bandwidth usage thresholds for scaling up and down. | Minutes | Lower. See Billing items. | Cloud native read/write splitting scales all nodes uniformly. Cluster and classic read/write splitting scale each node independently. |
How it works
The system monitors inbound and outbound traffic in real time. When either direction exceeds the instance's fixed bandwidth, the system automatically increases bandwidth within seconds, up to the 288 MB per-node cap.
When the traffic spike ends and bandwidth usage drops below the fixed bandwidth level, the system releases the extra burstable bandwidth automatically.
For cluster or read/write splitting instances, the Bandwidth (MB/s) value on the Basic Information page shows the total bandwidth across all data nodes.
Billing
| Item | Details |
|---|---|
| Billable item | The burstable portion — bandwidth that exceeds the base bandwidth and fixed bandwidth. |
| Billing method | Billed hourly based on actual burst usage. |
| Unit price | 2.5x the price of a manual bandwidth upgrade. Prices vary by region. See Billing items. |
| Per-second cost formula | ⌊Accumulated burstable bandwidth⌋ × Hourly unit price ÷ 3600 |
| Settlement | At the end of each hour, the system totals per-second costs to generate the bill. |
Secondary nodes and read-only nodes receive the same bandwidth extension as the primary node and are not currently charged.
Enable burstable bandwidth
-
Log on to the console and go to the Instances page. In the top navigation bar, select the region where your instance resides, then click the instance ID.
-
In the Basic Information section, click Modify next to Bandwidth (MB/s).
-
In the panel that appears, in the Elastic Burst section, select Enable.
The change takes effect in about one minute. Monitoring metrics may take 1–5 minutes to reflect the update. Enabling or disabling this feature does not interrupt instance access.
Monitor traffic
After enabling burstable bandwidth, monitor your instance's traffic usage to evaluate bandwidth requirements and cost-effectiveness. See View performance monitoring.