An Elastic High Performance Computing (E-HPC) cluster is a group of Elastic Compute Service (ECS) instances that deliver high-performance computing capabilities. Compared with typical ECS instances, E-HPC clusters are more powerful, reliable, and also provide advantages in scalability and availability. This topic describes the cluster types supported by E-HPC and provides references for you to create and manage clusters.
Supported cluster types
The following table describes the cluster types supported by E-HPC. E-HPC clusters are classified based on the deployment mode, architecture, and scheduler.
Deployment mode | Edition | Description | References |
Alibaba Cloud public cloud | Standard Edition | An on-cloud HPC cluster that consists of ECS instances. After a cluster is created, E-HPC installs the scheduler and service components. You need to maintain the service availability of the cluster. The following scheduler types are available:
| |
Managed Edition | An on-cloud HPC cluster that consists of ECS instances. The management nodes are maintained by E-HPC. The scheduler is Slurm and the compute nodes are ECS instances. |
How do I select a cluster type?
You can select a cluster type based on your business requirements.
If you want to build an HPC cluster from scratch, create a standard cluster.
If you want E-HPC to manage the nodes, create a managed cluster.
Manage clusters
Automated deployment of cluster environments:
Manage cluster resources:
Manage cluster users: Manage users
Job scheduling