All Products
Search

This Product

    Elastic GPU Service:Quick reference

    Document Center

    Elastic GPU Service:Quick reference

    Last Updated:May 07, 2024

    From a resource management standpoint, GPU-accelerated instances are considered as Elastic Compute Service (ECS) instances and are managed in the same way you manage other ECS instances. When you use ECS, you may need to perform various operations on different resources, such as connecting to instances, replacing operating systems, resizing disks, upgrading or downgrading instance configurations, and using snapshots or images. This topic describes common operations on ECS resources.

    Limits

    • For information about the usage notes of ECS instances, see Usage notes.

    • For information about the limits of ECS resources, see Limits and View and increase instance quotas.

    • Before you apply for Internet Content Provider (ICP) filings for websites that are deployed on your ECS instances, make sure that the instances meet ICP filing requirements. You can apply for only a limited number of ICP filing service identification numbers for each ECS instance. For more information, see Check the instance for ICP filing and access information.

    Create and manage ECS instances

    Manage the billing of ECS instances

    Improve cost-effectiveness

    • You can purchase preemptible instances to reduce costs and use preemptible instances in conjunction with Auto Provisioning for automated provisioning of instances. For more information, see Create an auto provisioning group and Create a preemptible instance.

    • You can purchase reserved instances to improve the flexibility of paying for instances and reduce costs. For more information, see Purchase reserved instances.

    • You can purchase storage capacity units to offset the billing of pay-as-you-go cloud disks in the same region. For more information, see Purchase an SCU.

    Create and manage disks

    You can perform the following steps to use a disk as a data disk:

    1. Create a disk.

    2. Attach a data disk.

    3. Initialize a data disk whose size does not exceed 2 TiB on a Linux instance or Initialize a data disk whose size does not exceed 2 TiB on a Windows instance.

    4. Create a snapshot to back up data.

      For more information, see Create a snapshot for a disk.

    5. If the storage capacity of an existing disk does not meet your requirements, resize the disk. For more information, see the following topics:

    6. If a data error occurs on a disk, use a snapshot created at a specific point in time to roll back the disk.

      For more information, see Roll back a disk by using a snapshot.

    7. If you want to restore a disk to its initial state, re-initialize the disk.

      For more information, see Re-initialize a data disk.

    8. Detach a data disk.

    9. Release a disk.

    Create and manage snapshots

    You can perform the following steps to use snapshots:

    1. Create a snapshot. You can use one of the following methods to manually or automatically create a snapshot:

    2. View the snapshot size.

    3. Delete snapshots that are no longer needed to save storage space.

      For more information, see Reduce snapshot fees.

    Snapshots are used in the following scenarios:

    Create and manage custom images

    You can manage custom images in the ECS console. You can use a custom image to deploy a business environment in a quick manner.

    Create and manage security groups

    You can clone a security group across regions and network types to simplify business deployment. For more information, see Clone a security group.

    Create and attach instance RAM roles

    You can perform the following steps to create and attach an instance RAM role:

    1. Optional. Grant a RAM user the permissions to manage an instance RAM role.

      For more information, see Grant a RAM user the permissions to manage an instance RAM role.

    2. Create and attach an instance RAM role.

      For more information, see Use instance RAM roles to control access to resources.

    3. Replace the instance RAM role at any time based on your requirements.

      For more information, see Replace an instance RAM role.

    Create and manage SSH key pairs

    Create and manage ENIs

    Configure an IPv6 address for an ECS instance

    For more information, see Configure an IPv6 address for an ECS instance.

    Use tags

    You can use tags to manage resources to enhance efficiency. You can perform the following steps to use tags:

    1. Create or add a tag.

    2. Search for resources by tag.

    3. Remove or delete a tag.

    Use launch templates

    Launch templates help you create ECS instances that have the identical configurations. You can perform the following steps to use launch templates:

    1. Create a launch template.

    2. Manage launch template versions.

    3. Delete a launch template.

    Use deployment sets

    Deployment sets help you implement high availability for underlying applications. You can perform the following steps to use deployment sets:

    1. Create a deployment set.

    2. Create an ECS instance in a deployment set.

    3. Change the deployment set of an instance.

    4. Delete a deployment set.

    Use Cloud Assistant

    Cloud Assistant allows you to send remote commands to ECS instances without the need to use jump servers. You can perform the following steps to use Cloud Assistant:

    1. Optional. Manually install and configure Cloud Assistant Agent on specific instances.

      For more information, see Install Cloud Assistant Agent.

    2. Create a command.

    3. Run a command.

    4. Check execution results and troubleshoot common issues.