Scale a Cluster

Scaling changes a cluster's compute capacity and cache. Choose the scaling strategy that matches your workload pattern, then review the cost and cache impact before applying it.

Choose a scaling strategy

Use one scaling strategy at a time. Enabling AutoScaling disables Manual Scaling and Time-based Scaling.

Scaling method	Best for	How it works	What you configure
Manual scaling	Immediate, one-time resource changes, such as scaling out before a known workload, scaling in after traffic drops, or adding cache for a larger working set.	You manually choose the target compute or cache size.	Target compute size, or separate cache scale-out size.
Auto Scaling	Dynamic or hard-to-predict workload changes.	VeloDB Cloud evaluates recent resource usage and automatically chooses a suitable CPU tier.	AutoScaling, Min CPU Cores, and Max CPU Cores.
Time-based scaling	Predictable and recurring workload peaks and lows, such as daily business hours and off-peak periods.	The cluster scales at fixed daily times according to rules.	At least two scheduled rules with different target vCPU values.

Scale Compute to change the vCPU capacity available for queries, loading, and concurrent workloads. Compute scaling changes cache capacity proportionally. Scale Cache separately only when you need more room for hot data without additional vCPU capacity.

Note:

VeloDB Cloud does not support separate cache scale-in.

Warning:

• Queries can be delayed or interrupted while the cluster scales.
• When scaling in compute, cache space shrinks proportionally, and data outside the new cache capacity is evicted. Query latency may increase while the cache warms up again.

Open scaling settings

1. Log in to the VeloDB Cloud console.
2. In the upper-left corner, select the warehouse that contains the cluster.
3. In the left navigation pane, under Compute, click Clusters.
4. Click the target cluster card to open its details page.
5. In On-Demand Resources, click Scale Out/In to open Elastic Scaling.

Before you scale

• SaaS free-trial clusters do not support scaling.
• A cluster must be in a status that allows operations. Avoid changing capacity during critical traffic unless capacity is needed immediately.
• Compute and cache charges change with the selected capacity. Review the displayed cost before you confirm a change.
• A scheduled rule runs only while its cluster is running normally. See Time-based scaling for schedule requirements and retry behavior.