Hyper Computing Cluster takes high-performance CVMs as nodes and interconnects with RDMA (Remote Direct Memory Access), providing high bandwidth and ultra-low latency network services, significantly improving network performance, and meeting the parallel computing requirements of large-scale high-performance computing, artificial intelligence, big data recommendation, and other applications.
Ensure you understand the region where the selected Hyper computing instance is located. For information on available regions, please see Available Regions.
Purchase Steps
This article takes the GPU HCCPNV5v as an example to guide you in quickly purchasing a Hyper Computing Cluster instance.
Step 1: Create a Hyper Computing Cluster.
Note:
If you have not purchased a Hyper Computing Cluster, please see this step to create a Hyper Computing Cluster. If you already have Hyper Computing instances, please choose whether to create as needed.
Instances within the same cluster are interconnected with the RDMA network, while instances across clusters are isolated.
1. Log in to the CVM console, and choose HPC Cluster > Cluster Resources in the left sidebar.
2. On the HPC Cluster page, select the target region at the top and click Create. This article takes Shanghai as an example. For available regions of GPU type HCCPNV5v instances, see Available Regions.
3. In the Create Cluster window, configure AZ, Cluster Name, Cluster Description, and Tags as needed, as shown in the following figure:
4. Click OK to create the cluster.
Step 2: Go to the purchase page.
You can enter the purchase page to start purchasing instances through the following two ways:
Log in to the CVM console, select Instances in the left sidebar, and click Create Instance.
On the Hyper Computing Cluster page, click scale-out in the operation column of the target cluster, as shown in the following figure:
Step 3: Select the billing mode, network, region, and model.
After going to the purchase page, select the billing mode, network, region, and model. This article uses Shanghai Zone 5 and GPU HCCPNV5v as configuration examples. The actual information on the purchase page shall prevail:
Billing mode: You can choose as needed. For details, see Billing Overview.
Regions and availability zones: The available availability zones are subject to the instance purchase page.
Instance: Select Hyper Computing Cluster for Architecture. This article uses GPU HCCPNV5v as an example. You can choose as needed.
Step 4: Select the image and Hyper Computing Cluster.
1. Select the instance image as needed. Hyper Computing instances support three image types: public image, custom image, and shared image.
2. Select the desired Hyper Computing Cluster cluster to join, as shown below:
Note:
GPU Hyper Computing instances must have the corresponding GPU driver to run normally. You can install the relevant driver in the following two ways:
If you select a public image, some instances support selecting Install GPU driver automatically to pre-install the corresponding version of the driver. It is recommended that you choose this method. This method is only supported for some Linux public images.
1. Select the storage for the Hyper Computing instance. As shown in the figure below:
Main parameters are described as follows:
System disk: Used for installing the operating system. Both the type and size can be flexibly selected (storage capacity adjustment is not supported for local system disk instances).
Data disk: Used to expand the storage capacity of CVMs, providing efficient and reliable storage devices.
Note:
Local system disk instances do not support storage capacity adjustment.
Local storage poses a risk of data loss and is only suitable for scenarios with high storage I/O performance requirements and a high-availability architecture. It is not applicable to usage scenarios where the application layer lacks a data redundancy architecture.
2. Click Next: Set Network and Host after setting is completed.
Step 6: Set the network, security group and host.
1. Select the network and bandwidth of Hyper Computing instance. As shown in the figure below:
Network: Select an existing VPC or create a new VPC.
Public IP address: If your instance requires public network access, select Assign a standalone public IP address. A public IP address will be assigned to the instance upon creation.
Bandwidth cap: The public network bandwidth cap of the instance. Configure as needed.
2. Create or select an existing security group to control the port access range. As shown in the figure below:
3. Set the login password or key of Hyper Computing Cluster instances
4. Set other custom configurations as needed.
5. Click Next: Confirm Configuration.
Step 7: Confirm the configuration information.
1. Please verify the instance information in the Confirm Configuration step.
Confirm whether the configuration items such as instance specification, image selection, storage, public network configuration, and security groups meet expectations.
You can select or verify the quantity and duration of purchase.
