Supercomputing at your fingertips
ClusterTech HPC Environment Software Stack, CHESS is high-performance cluster software which creates an integrated system and application environment from a collection of servers. CHESS manages the arrangement, management, monitoring and scheduling of cluster resources, to dramatically increase computing capacity and accelerate application processing.
Developed entirely inhouse, CHESS combines enterprise-graded performance, outstanding stability and security, flexibly extendable modular-type function architecture and visual and a friendly graphic operating interface. It is applicable to any computationally intensive application domain including aerospace, automobile, electronics, education, scientific research, petroleum, meteorology and life science.
CHESS Software Architecture
CHESS functional modules:
With its modular design, CHESS can freely select combinations of modules according to user demand. Modules include cluster arrangement, cluster management, cluster monitoring, operation scheduling, reporting and WEB gateway.
CHESS Cluster Arrangement Module
In a large-scale cluster, the arrangement of operating system and software has always been troublesome for system administrators as it involves tedious and repetitive work. The CHESS Cluster Arrangement Module can help system administrators quickly and simply complete the arrangement of operating system and software for cluster nodes. The major functions are:
- Automatic quick arrangement of complete cluster system;
- System backup and recovery;
- Quick recovery of node operating system to the initial state;
- Able to distribute relevant system image for different nodes simultaneously and customize software package;
- Supporting no-disk mode.
CHESS Cluster Management Module
- Support role-based node management, remote batch On/Off, parallel command and file sharing;
- Automatic synchronic configuration of offline nodes, upon being online, to maintain the consistent configuration of all the nodes.
CHESS Cluster Monitoring Module
- Monitoring interface: support the physical allocation display of devices and display the information indicators of node status, including load, online/offline and CPU Temperature;
- Monitoring range: monitory the use status of system resources for the unit and each node (CPU, memory, hard disk and network use rate) and hardware status (CPU, memory, temperature, voltage, fan and network).
CHESS Operation Scheduling Module
Integrate computer resources (including blade server, workstation and microcomputer) for unified scheduling and unified management, and provide users with 7x24hr efficient computing capacity. The operation scheduling module of CHESS has the following basic functions:
- Operation submission: support various submission modes such as command line, WEB interface, application software integration interface, operation text and EXE file;
- Operation management: includes operation query, operation deletion, operation hang-up and operation release and able to sort according to operation ID, operation name, user name, status and queue;
- Operation system configuration: set up the scheduling parameter, scheduling policy, queue property and node property through WEB interface;
- Scheduling policy: includes advance reservation of resources, Backfill algorithm, dynamic priority, fair sharing, quota management, system diagnosis, system monitoring and statistics; support QoS and policy-based scheduling; support preemptive policy; use cluster resources preferentially for important operation;
- Support cluster partition scheduling.
CHESS WEB Gateway
- Support WEB-based high-performance computing. System administrators, end users and management can use CHESS via the WEB;
- System administrator: user management, network management, file system management, system monitoring and management;
- End user: submission and management of batch processing operations, submit X-window-based interactive front-desk operations, upload and download data;
- Users can complete almost all tasks without the need to login via a command line.
CHESS Reporting Module
Provide statistics, analysis and report of resources used, including CPU utilization, machine online status and number of operations.
CHESS Billing Module
Generate billing, printing and query functions based on the relevant charge rate.
Super-computing capacity for 5000 nodes and more.
Recover from any single-point faults and ensure that the cluster continues running in such events.
Supporting the management and scheduling of GPU clusters.
Login control support
Provide the login control function to ensure the unified management and scheduling of computing resources and avoid any disorderly competition.
No-disk cluster support
Start the operating system via network, without installing the local operating system at the computing node.
Breakpoint computing support
CHESS supports computing breakpoint resume based on application program level or BLCR level.
WEB-based integration platform support
The WEB integration platform supports control and operation of WEB-based batch processing and interactive e operations.
High-speed parallel file system
Memory file system support, provides class-leading I/O performance and increases throughput.