All Products
Search
Document Center

Hologres:Overview: Scaling methods and glossary

Last Updated:Apr 10, 2026

As query workloads grow, you need a way to add compute capacity without disrupting running jobs. Hologres supports both vertical and horizontal elastic scaling for compute groups, so you can match resources to workload demand — scaling individual query performance or concurrent throughput as needed.

Scaling directions

A Hologres instance can be divided into multiple compute groups. Each compute group supports one of two scaling directions, configured independently.

Direction

Available since

How it scales

Best for

Vertical (scale-up/down)

Hologres V2.0

Increases or decreases the computing resources of a compute group

Medium-to-large query workloads; high-concurrency, small-task workloads

Horizontal (scale-out/in)

Hologres V4.0

Increases or decreases the number of clusters in a compute group

High-concurrency, small-task workloads requiring flexible throughput and isolation

When to scale vertically

Vertical scaling adds or removes computing resources from a single compute group. Use vertical scaling when individual queries are slow or resource-intensive — for example, complex analytical queries that need more computing resources to run faster.

When to scale horizontally

Horizontal scaling adds or removes clusters within a compute group. Use horizontal scaling when the number of concurrent queries is growing and you need more flexible throughput and isolation capabilities.

Scaling methods

Method

Direction

Description

Manual scaling

Vertical

Manually adjust compute group resources or cluster count. See Compute group management.

Time-based scaling (Beta)

Vertical

Schedule vertical scaling tasks to run at set times. See Time-based scaling (Beta).

Auto scaling (Beta)

Horizontal

Set a scaling limit and let Hologres scale out automatically based on load. See Multi-cluster and auto scaling (Beta).

Resource terminology

The following terms apply across all scaling features, grouped by the feature they belong to.

Time-based scaling

Term

Definition

Compute group reserved resources

The baseline resources allocated to a compute group from the instance's reserved resources. Set when creating the compute group and adjustable later.

Compute group elastic resources

Resources dynamically added on top of reserved resources through time-based scaling.

Total computing resources of a compute group

Reserved resources + elastic resources.

Multi-cluster and auto scaling

Term

Definition

Reserved cluster count

The number of clusters reserved for a compute group. Set when creating the compute group and adjustable later. Resources for reserved clusters come from instance reserved resources.

Specifications of a single cluster

The resource size of each reserved cluster. Set when creating the compute group and adjustable later.

Compute group reserved resources

Specifications of a single cluster x Reserved cluster count.

Compute group elastic resources

Specifications of a single cluster x (Current cluster count - Reserved cluster count).

Total computing resources of a compute group

Reserved resources + elastic resources.

Instance-level resources

Term

Definition

Instance reserved resources

The total computing resources reserved for an instance, available under both subscription and pay-as-you-go billing. Divided into allocated and unallocated portions.

Instance allocated resources

The portion of instance reserved resources assigned to compute groups as their reserved resources.

Instance unallocated resources

The portion of instance reserved resources not yet assigned to any compute group.

Instance elastic resources

The total resources scaled out across all compute groups via time-based scaling or auto scaling, beyond their reserved allocations.

Total computing resources of an instance

Instance reserved resources + instance elastic resources.

What's next