What is E-HPC? - Elastic High Performance Computing - Alibaba Cloud Documentation Center

Elastic High Performance Computing (E-HPC) is a computing service provided by Alibaba Cloud. E-HPC delivers excellent and stable computing performance that is achieved by auto scaling. E-HPC aggregates computing capabilities and uses parallel computing to address major challenges in science, engineering, and commerce. E-HPC is used in a variety of fields, including scientific research, petroleum exploration, finance, weather forecast, pharmaceuticals, gene sequencing, and graphics processing.

Why E-HPC?

Compared with other HPC clusters, E-HPC clusters provide deployment flexibility, resource elasticity, data security, high service availability, and visualization.

Benefits of E-HPC clusters

Flexible deployment
You can create E-HPC clusters in the E-HPC console in a few steps. You can deploy a high-performance computing environment and application software in a few clicks. This provides high processing power for various scenarios. After you create an E-HPC cluster, you can directly use it to submit jobs for computing.
Elastic resources
E-HPC monitors the loads of nodes in each E-HPC cluster based on job queue usage and then dynamically adjusts computing resources accordingly. If the job queue is empty, the E-HPC cluster scales down to the specified minimum number of nodes. When one or more jobs are submitted to the job queue, a suitable number of nodes are created to make full use of the computing resources.
Secure data
- E-HPC clusters are isolated from each other by using virtual private clouds (VPCs). Security groups are used to implement triple access control among the nodes of the E-HPC clusters in a VPC. This feature ensures the network security for E-HPC clusters.
- The data of E-HPC clusters is stored in Apsara File Storage NAS. The encrypted transmission and encrypted storage features of NAS can protect data from theft or tampering. The data is replicated into multiple copies and stored at the backend of NAS. This feature can be used to significantly increase data security.
High availability
E-HPC cluster nodes are deployed based on Elastic Compute Service (ECS), Super Computing Cluster (SCC), and Elastic GPU Service resources. This significantly improves the availability of the E-HPC clusters.
Visualized results
E-HPC provides the Virtual Network Computing (VNC) service, which allows you to visualize computing results as graphics. For example, you can view the rendered effects of animations or the cinemagraphs of weather changes.
Quick cloud migration
E-HPC supports mainstream HPC schedulers and tools to ensure pain-free workload migration to the cloud. Users can use their original job scripts and commands.

Benefits of E-HPC over traditional computing clusters

Item	E-HPC cluster	Traditional computing cluster
Deployment	You can use an E-HPC cluster immediately after you purchase ECS instances for the E-HPC cluster. You can run high-performance computing jobs on the E-HPC cluster and obtain computing results in real time. E-HPC clusters are available in multiple specifications. You can select specifications based on your scenarios.	If multiple users use a traditional computing cluster at the same time, the users must wait in a queue due to the limited resources. If you want to scale up a traditional computing cluster, you need to spend a long time purchasing and deploying on-premises resources.
Security	Data is stored on the cloud for high reliability and security.	Data is stored on clients, which may have high security risks due to uncontrollable user behavior.
Cost	Resources are dynamically and automatically scaled up and down to reduce costs without concerns about service reliability and availability. When you use E-HPC, you do not need to upgrade the hardware. E-HPC provides optimized software licenses and a scheduling mechanism to reduce costs.	Resources are purchased based on peak-hour needs. Most of the resources sit idle during off-peak hours. You must make a one-time purchase of the required infrastructure. You need to upgrade the hardware. Hardware upgrades increase the cost of computing clusters.
O&M	E-HPC provides automatic O&M. This provides the auto scaling feature. It can be used to replace unhealthy nodes. E-HPC provides a variety of software applications for high-performance computing. The software applications are upgraded at the same time as E-HPC.	The resource procurement and delivery are time-consuming. This causes difficulties in upgrades and configuration changes. Professional O&M engineers are required.

Common scenarios

You can configure different types of resources for different scenarios.

Weather forecast

You can use E-HPC in combination with numerical forecast models to calculate and analyze meteorological and environmental data. This way, you can forecast weather conditions and environmental changes.

Energy exploration

E-HPC can be used in the exploration industry to analyze data and simulate the geological structure of a surveyed area. Then, you can use the data to accurately locate energy resources.

Life sciences

Bioinformatics: You can use E-HPC to sequence a large number of biological genomes and perform related tasks. Then, you can obtain the information of genomes and the result of data analysis to address the challenges in biology and medicine.
Dynamics simulation: You can use E-HPC to simulate large-scale molecular dynamics. The simulation can be used to analyze and predict the interactions between protein and lipids and to track the changes of the two substances.
Drug development: E-HPC can be used by drug developers to perform concurrent high-throughput screening on a large number of small molecule libraries.

Scientific research and education

E-HPC can provide supercomputing services for governments, universities, and supercomputing centers. These can be used to support numerical simulation and simulation verification during scientific researches. This way, education experts and scientific researchers no longer need to invest a large amount of time and resources in acquiring skills and knowledge in processors or supercomputers. Instead, they can focus on research.

Manufacturing simulation

You can use E-HPC clusters to analyze complex engineering structures and mechanical structures. You can also simulate and optimize product structures and performances based on a large amount of data. E-HPC applies to multiple industries, such as smart car manufacturing, aerospace, machinery manufacturing, and construction.

Animation rendering

E-HPC supports concurrent, large-scale computing based on multiple clusters in animation rendering of movies or TV programs.

How to use E-HPC?

Operation	Description	References
Create users	A user is an identity that is used to submit jobs in an E-HPC cluster. E-HPC clusters allow you to create users that have different permissions. This way, administrators and ordinary users can be distinguished.	Users
Configure computing resources	E-HPC allows you to configure and manage computing resources such as nodes, scheduler queues, and auto scaling settings.	Nodes and queues
Submit jobs	You can submit jobs to a cluster by using the E-HPC console. If you want to submit a small job, you can directly edit the script in the console. If you want to submit a large job, you can upload the job script to an Object Storage Service (OSS) bucket and then import the script to the E-HPC console. You can also use an SSH client to log on to the cluster and submit jobs.	Submit jobs Submit jobs by using a submitter
Analyze performance	The E-HPC Portal is a platform that you can use to monitor cluster performance, view cluster performance history, and profile the performance of a specific process.	Feature overview
Visualize job running	You can use a visualization tool, such as VNCViewer, to view the running of a job and intuitively manage the job.	Use VNCViewer to intuitively access a cluster