Analytical workloads on read-only In-Memory Column Index (IMCI) nodes often have uneven demand — high during business hours, low overnight. Without auto-scaling, you either over-provision to handle peaks or accept degraded performance during spikes. PolarDB Serverless scales compute resources on read-only IMCI nodes automatically, reducing costs during off-peak hours and absorbing traffic spikes without manual intervention.
Cost impact example: In a scenario that requires 32 cores and 64 GB of memory for 12 peak hours and 16 cores and 32 GB for 12 off-peak hours, serverless reduces compute node costs by 25% compared to always running at peak capacity.
Prerequisites
Before you begin, make sure you have:
A PolarDB for MySQL 8.0 Enterprise Edition cluster
At least one read-only IMCI node already added to the cluster
Choose a method
| Method | How it works | Best for |
|---|---|---|
| Specify a scaling range | Add elastic limits to an existing fixed-spec node. The node scales within your defined minimum and maximum during demand changes. | Workloads with occasional spikes where you have already provisioned a baseline spec |
| Create a serverless node | Add a new node with no fixed spec. Resources scale from 1 to 16 PolarDB Capacity Units (PCUs) automatically. | Cost-sensitive workloads with significant peak-to-off-peak variation |
How IMCI nodes optimize serverless scaling
Read-only IMCI nodes extend the standard serverless architecture with four optimizations:
| Optimization | What it does | Version requirement |
|---|---|---|
| Resource utilization | Key system parameters — including cache, thread pool, and execution memory — are tuned automatically during scale-up and scale-down events. | All supported versions |
| SQL query acceleration | When complex queries create resource contention and slow down queued queries, the node automatically allocates additional resources to relieve the bottleneck. | 8.0.1 revision 8.0.1.1.39 or later; 8.0.2 revision 8.0.2.2.20 or later |
| Faster IMCI loading after restart | After a cluster restart, the node uses elastic resources to accelerate index rebuilding, reducing the time before IMCI becomes available. | 8.0.1 revision 8.0.1.1.36 or later; 8.0.2 revision 8.0.2.2.20 or later |
| Accelerated log replay | When SQL processing consumes most resources and delays IMCI log replay, the node automatically allocates additional resources for replay, preventing query routing delays. | 8.0.1 revision 8.0.1.1.45 or later |
| Faster DDL processing | The node adaptively increases thread resources to speed up DDL statement execution on column indexes. | 8.0.1 revision 8.0.1.1.44 or later; 8.0.2 revision 8.0.2.2.25 or later |
Specify a scaling range for an existing node
Use this method if the cluster already has a read-only IMCI node with a defined spec and you want to add elastic scaling on top of it.
Log on to the PolarDB console.
In the upper-left corner, select the region where your cluster is deployed.
Find the cluster and click its ID.
On the Basic Information page, scroll to the Database Nodes section and click Enable Serverless.
In the Enable Serverless dialog box, set Maximum Resources for Single Node and Minimum Resources for Single Node. For valid values, see Enable the serverless feature for a cluster with defined specifications.
Click OK.
The serverless feature is now enabled on the read-only IMCI node.
Create a serverless read-only IMCI node
Use this method to add a new node without a fixed spec. PolarDB creates the node and manages all scaling automatically, with resources ranging from 1 PCU to 16 PCUs.
In the left-side navigation pane, click Clusters.
In the upper-left corner, select the region where your cluster is deployed.
Find the cluster and click its ID.
On the Basic Information page, scroll to the Database Nodes section and click Enable Serverless.
In the Enable Serverless dialog box, configure the Read-only IMCI Nodes count (for example, set it to 1) and click OK.
PolarDB automatically provisions the new read-only IMCI node with resources ranging from 1 PCU to 16 PCUs.
