> ## Documentation Index
> Fetch the complete documentation index at: https://docs.scitix.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Resources Overview

> Learn about the relationships among instance resources, storage resources, resource pools, quotas, and orders in Siverse.

Resources is Siverse's entry point for managing compute and storage. Development environments, training jobs, and inference services consume the instance and storage resources managed here during runtime.

Before starting a task, confirm that sufficient compute capacity, storage volumes, resource pools, and quotas are available. If resources are insufficient, the task may fail to be created, be queued, or be rejected during quota validation.

## Feature Entry

After you access **Resources**, the commonly used entry points are as follows:

| Entry | Primary Use |
| - | - |
| **Instance** | View resource pools, instance specifications, personal quotas, and supported workload types. |
| **Volume** | View volume capacity, authorized users, and directory-level permissions. |
| **Resource Dashboard** | View resource utilization, user usage rankings, and historical trends. |

When comparing resource, quota, or usage data, use the same filter criteria, such as **Region**, **Cluster**, **Resource Type**, **Resource Pool**, and time range.

## Resource Objects

| Object | Description |
| - | - |
| Instance | GPU / CPU computing resources used to run development environments, training tasks, and inference services. |
| Storage resources | Persistent storage volumes (Volume) used to store code, data, model files, training outputs, and service dependency files. |
| Resource pool | A computing resource pool organized by region, cluster, instance specification, and usage scope, including shared, dedicated, and exclusive resource pools. |
| Quota | The amount of resources available to you. Tasks consume quota while running; quota is returned when tasks are stopped or released. |
| Orders | Records of resource purchases and renewals. Orders determine the resource quota amount, validity period, and billing status. |

## Resource Pool Type

Instance resources are organized by resource pool. Different resource pools are suitable for different isolation, cost, and management requirements.

| Resource Pool Type | Applicable Scenario | Description |
| - | - | - |
| shared resource pool | A team jointly uses a set of general-purpose resources. | Quotas are shared and consumed by users within the tenant. Suitable for general development, debugging, and elastic tasks. Authorized users and user-level configurations can be managed in the resource pool details. |
| dedicated resource pool | Resources are reserved for specified users or teams. | Can be used for fixed quotas, user authorization, and workload type restrictions. |
| Exclusive resource pool | Tasks requiring exclusive access to entire nodes or stronger isolation. | Organized by whole-machine resources, with support for node health status and defragmentation. |

## Resource Quotas and Scheduling Boundaries

* Resources are organized by region and cluster. Before creating a task, ensure that the resources, storage volumes, images, datasets, and models are in regions and clusters that you can access.
* Instance resources and storage resources are purchased and consumed as quotas. After purchasing instance resources, ensure that the quota has been added to the available resource pool. After purchasing storage resources, you must also create Volumes within the storage quota.
* A resource pool can be configured with user-level policies, such as authorized users, individual quotas, priorities, and resource guarantees. It can also restrict the workload types that can be submitted. If the current user or task type is not allowed, the submission will be rejected by quota validation.
* Dedicated resource pools require whole-machine validation. GPU Instances must be allocated in increments of a full 8-GPU machine, while CPU Instances must be allocated in increments of a full 120-CU machine.
* After an order expires, the associated quotas become invalid. Before expiration, confirm whether renewal is required.

## Resource Usage Dir

Resources are typically prepared and used as follows:

1. Purchase instance resources or storage resources to generate resource orders. When purchasing instance resources, if the target dedicated or exclusive resource pool already exists, you can directly Specify the quota destination. If no destination is specified, the quota is allocated to the shared resource pool.
2. Based on usage scope and isolation requirements, create or adjust dedicated and exclusive resource pools, and adjust the resource pool allocation of existing instance quotas as needed.
3. Create or adjust storage volumes and authorize the users who need to use them.
4. In Develop, Training, or Inference, select the resource pool, instance specification, and storage volume.
5. In Resource Dashboard, view resource usage, scheduling rate, and user usage; in Task Dashboard, view task resource consumption.
6. Before an order expires, renew it in Order to prevent quota expiration from affecting task execution.

## Related Documentation

* [Prepare instance resources](/siverse/resources/prepare-compute-resources.md)
* [View resource usage](/siverse/resources/view-resource-usage.md)
* [Troubleshoot insufficient task resources or long queue times](/siverse/resources/troubleshoot-insufficient-resources-or-queued-tasks.md)
* [Create and use storage volumes](/siverse/resources/create-and-use-storage-volumes.md)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.