This topic describes the resource scaling modes (vertical scaling and horizontal scaling) that Hologres compute group instances support, helping you choose the best method for your business needs.
Scaling modes
Hologres lets you divide an instance into multiple compute groups. You can scale the resources of each group on demand using one of the following methods:
Vertical scaling (scale up/down): Available since Hologres V2.0. This method increases or decreases the computing resources of a compute group and is ideal for handling both medium-to-large and high-concurrency small tasks.
Horizontal scaling (scale out/in): Available since Hologres V4.0. This method increases or decreases the number of clusters in a compute group. It offers more flexible throughput and workload isolation, making it ideal for high-concurrency, small-task scenarios.
Resource scaling methods
You can scale resources in the following ways:
Manually scale resources vertically or adjust the number of clusters horizontally. For more information, see Manage compute groups.
Time-based scaling (vertical): Configure a scheduled task to scale computing resources vertically at specific times. For more information, see Time-based scaling (Beta).
Auto scaling (horizontal): Set a scaling limit, and the system automatically scales out based on the current workload. For more information, see Multi-cluster and auto scaling (Beta).
Key concepts
The following terms apply to time-based scaling:
Reserved resources for a compute group: The computing resources you define when creating a compute group. This amount can be modified later. These resources are allocated from the instance's reserved resources.
Elastic resources for a compute group: Additional computing resources provisioned on top of reserved resources through time-based scaling.
Total computing resources for a compute group: The total computing resources used by a compute group: the sum of its reserved and elastic resources.
The following terms apply to multi-cluster and auto scaling:
Reserved cluster count: The number of clusters you define for a compute group when you create it. This value can be modified later. The computing resources for these reserved clusters are allocated from the instance's reserved resources.
Per-cluster specification: The portion of an instance's reserved computing resources allocated to each reserved cluster. You define this specification when creating a compute group, and it can be modified later.
Reserved resources for a compute group:
Per-cluster specificationmultiplied byReserved cluster count.Elastic resources for a compute group:
Additional computing resources provisioned through auto scaling beyond the reserved resources.
Per-cluster specificationmultiplied by (Current cluster count-Reserved cluster count).
Total computing resources for a compute group: The total computing resources used by a compute group: the sum of its reserved and elastic resources.
The following computing resources are defined at the instance level:
Reserved resources for an instance: The total computing resources reserved for an instance. These resources are available with both subscription and pay-as-you-go billing methods. These resources are categorized as either allocated or unallocated:
Allocated instance resources: The portion of an instance's reserved resources allocated as reserved resources for all compute groups.
Unallocated instance resources: The portion of an instance's reserved resources not yet allocated to any compute group.
Elastic resources for an instance: The total additional computing resources provisioned by all compute groups in an instance through time-based scaling or auto scaling.
Total computing resources for an instance: The total computing resources used by an instance: the sum of its reserved and elastic resources.