You can use an Elastic High Performance Computing (E-HPC) client to submit jobs, query jobs, and manage job files. This way, you can manage jobs with enhanced efficiency and user experience. This topic describes the features, scenarios, and limits of an E-HPC client.
E-HPC provides two types of users: cluster administrator and job submitter.
A cluster administrator deploys and manages E-HPC clusters. For example, a cluster administrator can create clusters, configure clusters, scale out clusters, and manage job submitters.
A job submitter submits jobs to E-HPC clusters for high-performance computing.
An E-HPC client is designed for job submitters to submit jobs and manage job files. The following table describes the features of an E-HPC client.
You can submit a job by using the wizard of an E-HPC client, a script, or a job template.
You can query jobs based on filter conditions to schedule jobs and tasks.
You can use the file browser to visualize file management and transfer files between a local computer and a cluster.
You can configure software applications such as Fluent and Abaqus. You can submit jobs and run applications by using the wizard.
You can remotely log on to a cluster by using the terminal of an E-HPC client and run commands to manage the cluster. You can also use Virtual Network Computing (VNC) to connect to a cloud desktop and manage clusters in a visualized manner.
When you use an E-HPC client to manage jobs, you can use Alibaba Cloud Elastic Desktop Service (EDS) and cloud storage to build one-stop cloud HPC solutions that deliver data pre-processing, data post-processing, parallel computing, and data management capabilities.
Run an E-HPC client on a local computer
This solution is suitable for scenarios in which no pre-processing or post-processing is required for jobs. You can use an E-HPC client to manage jobs. For example, you can submit jobs, query jobs, upload data, and download data.
Run an E-HPC client in a cloud desktop
This solution is suitable for scenarios in which pre-processing and post-processing are required for jobs. Offline workstations for simulation and cloud desktops are deployed in a similar way. Before you use an E-HPC client to manage jobs, you need to install software for data pre-processing and post-processing in a cloud desktop. This optimizes user experience without compromising modeling or analysis results.
You can connect a cloud desktop to an Active Directory (AD) domain of an enterprise. This way, AD users can seamlessly access a cloud computing environment.
The scheduler of the cluster that you connect to an E-HPC client must be PBS.
A maximum of 15 users can be connected to a single cluster by using an E-HPC client at the same time.