Manually Scale Clusters and Frontend Groups

Manual scaling lets you resize a cluster immediately when workload or cache requirements change. Use it for a one-time capacity change, such as preparing for a known workload or reducing resources after traffic drops.

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.

Choose what to scale, compute or cache

Choose to scale compute or cache according to the workload bottleneck:

You need...	You can...
More query concurrency or faster execution	Scale out compute. Cache capacity increases proportionally.
Lower compute cost after traffic drops	Scale in compute. Cache capacity decreases proportionally.
Lower cache cost when the working set is smaller	Scale in compute. VeloDB Cloud does not support separate cache scale-in.
More room for hot data or larger working sets without more vCPU	Scale out cache separately. Compute capacity does not change.

How cache scales with compute

By default, target cache space scales in proportion to compute resources (vCPU). When you scale in compute, the target cache space uses the lower limit for the target vCPU. Separate cache scale-out follows a different rule and does not change compute capacity.

For example:

• Scale out: The cluster has 4 vCPU and 100 GB of cache space. If you scale compute out to 64 vCPU, the cache space scales out to 1,600 GB: 100 GB * (64 vCPU / 4 vCPU).
• Scale in: The cluster has 64 vCPU and 400 GB of cache space. If you scale compute in to 8 vCPU, the cache space scales in to 100 GB: max(100 GB, 400 GB * (8 vCPU / 64 vCPU)).

If you scale out cache separately before you scale out compute, compute scaling uses the separately scaled cache space as the base. For example, a cluster starts with 4 vCPU and 200 GB of cache space. If you scale cache out separately to 400 GB, then scale compute out to 8 vCPU, the cache space scales out to 800 GB: 400 GB * (8 vCPU / 4 vCPU).

Before you scale

Check the current workload before changing resources. VeloDB Cloud performs scaling as smoothly as possible to minimize business impact, but it is still a good practice to avoid frequent resource changes during critical traffic peaks unless the extra capacity is needed immediately.

Note the following limitations and requirements:

• SaaS free-trial clusters do not support scaling.
• VeloDB Cloud only supports separate cache scale-out, and does not support separate cache scale-in.
• After you scale out cache separately, wait at least 6 hours before you scale out cache separately again.
• If AutoScaling is enabled, manual scaling is unavailable. Disable AutoScaling before you perform manual scaling.
• Manual scaling is available from the cluster Details page when the cluster is in a status that allows operations.
• Compute and cache charges change with the selected capacity. Review the displayed cost before you confirm the change.

Manually scale a cluster

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.
6. Under Scaling Strategy, select Manual Scaling.
7. Under Scaling Configuration, select Compute or Cache as needed.
8. In Target Configuration, select the target capacity.
9. Review the displayed cost, then click Confirm.

The cluster status changes to Scaling. When the operation finishes, it returns to Running.

Scale Frontend Groups

Frontend (FE) nodes handle query parsing, planning, and metadata management for a warehouse. Use Frontend Groups to view the warehouse's FE resource group and scale it when high shard counts or high concurrent query volume create an FE bottleneck.

Note:

• Frontend Groups is currently in Private Preview. It expands FE resources only. It does not change a cluster's compute or cache capacity.
• To enable FE group expansion for a warehouse, contact VeloDB Cloud Support. When enabled, Frontend Groups appears under Compute in the VeloDB Cloud console.

Before you begin

• Every warehouse includes a free Small FE group. Scaling to a larger specification adds a billable Fe Compute item.
• Use the free baseline unless the warehouse has a high number of data shards or a high concurrent query volume that requires additional FE capacity.

FE resource specifications

Each warehouse is provisioned with a default Small FE group at no charge. You can scale up to larger specifications when needed. Scaling up is charged per vCPU at the Cluster Compute rate.

Specification	Compute	Pricing
Small	4 vCPU, 16 GB RAM	Free
Medium	8 vCPU, 32 GB RAM	Billed by vCPU-hour
Large	16 vCPU, 64 GB RAM	Billed by vCPU-hour
XLarge	32 vCPU, 128 GB RAM	Billed by vCPU-hour

View a Frontend Group

1. Log in to the VeloDB Cloud console.
2. In the upper-left corner, select the warehouse.
3. In the left navigation pane, under Compute, click Frontend Groups.
4. Click the default meta group to open its details page.

The group card shows its compute specification, node count, and creation time. The details page shows Frontend Group ID, Frontend Group Name, CPU Architecture, Created At, Started At, Running Time, and On-Demand Resources.

Scale a Frontend Group

Scale a Frontend Group when query planning, metadata operations, or high query concurrency are constrained by FE capacity. Scaling above Small starts Fe Compute billing. Scaling back to Small returns the group to the free baseline and removes the Fe Compute billing item.

1. On the Frontend Groups page, click the meta group to open its details page.
2. Click Scale Out/In.
3. In Scaling Configuration, review Current Configuration.
4. Under Target Configuration, select the target Compute specification.
5. Review On-Demand New Cost, then click Confirm.

After scaling completes, the warehouse uses the selected FE specification.

Billing

After you scale above Small, use Billing Statements to review Fe Compute usage and charges.

1. In the left navigation pane, click Billing.
2. Click Billing Statements.
3. On the Summary tab, select Monthly, Daily, or Hourly granularity and set the start and end month. The summary shows the billing status, pretax discounted amount, deduction amounts, and outstanding balance.
4. On the Details tab, select Monthly, Daily, or Hourly granularity and set the start and end date. Find rows where Component is Fe Compute. Click Export on either tab to download the statement.

The details table uses UTC (UTC + 0) for Start Time and End Time and includes the following fields:

Column	Description
Warehouse ID	The warehouse that incurred the charge.
Cluster ID	The cluster that incurred the charge, when applicable.
Cluster Name	The cluster name, when applicable.
Component	The billed resource, such as Fe Compute, Cluster Compute, Cluster Cache, or Storage.
Billing Method	The billing method, such as On-Demand.
Usage	The measured usage, such as vCPU-Hour or GB-Hour.
Pretax Gross Amount	The amount before discounts and deductions.
Discount Amount	The discount applied to the gross amount.
Pretax Discounted Amount	The amount after discounts and before deductions.
Outstanding Balance	The remaining amount due after deductions.

See also

• Review scaling strategies to choose the right approach for the workload.
• Auto Scaling for dynamic workload changes.
• Time-Based Scaling when workload peaks follow a fixed schedule.