Background

The electric power industry is the basic industry of the national economy, since the establishment of the new China, the electric power industry has always been attached with great importance by the state and government. In the development process of the entire national economy, the electric power industry plays an extremely important role. With the social development, to solve the power demand, construction size, power structure and sustainable development problems, our state has implemented large-scale research and centralized construction on the electric power industry.

Core demands

The core demand of Central China Grid was to promote development, improve the power supply and management capacity of the entire State grid. To solve the huge platform demand and system management specification, the final size of the system must include the basic platform and 4 application platforms: real-time monitoring alarm application platform, scheduling plans application platform, security review application platform and scheduling management application platform. The basic platform and real-time monitoring alarm application platform of the system shall be built in the functional modules of safe area I and II, thus to realize the service hot line, message line, system management, man-machine interface, database management, data acquisition, SCADA, AGC, AVC, state estimation, network analysis, online safety stability analysis alarm, auxiliary scheduling decision, stable margin assessment, DTS and other functions of the system. Therefore, the system has extremely high requirement for both the computing performance and the computing accuracy, system expandability, reliability and other performances.

Demand analysis

To promote development, improve the power supply and management capacity of the entire State grid, the Central China Grid wished to build an intelligent grid scheduling technical support system to construct the Central China Grid as one system platform integrating security, management and application.

Solutions

Based on the user’s demands, Sugon designed a complete high-performance computing solution, i.e. Sugon high-performance computer cluster system. The high-performance computer system applied in this solution is based on general cluster framework, its integrates the computing capacity of hundreds of nods through the high-speed network, large-scale concurrent missions may be supported. Through adopting dual pass-through modules and connection with switches, Sugon was able to achieve the full-speed interconnection between all blades.

Based on the application characteristics, Sugon selected its independently developed 3rd generation dual core/ 4-core blade server TC 2600 blade server system. This product surpassed the traditional rack-mounted server from may aspects, this new generation of server product enjoys high stability, high availability, high scalability, high density and high performance. In this solution, Sugon adopted the blade server as the computing node.

The purpose was to meet the performance and convenience requirements of the users in aspects of super high computing density, super large storage space, super low energy consumption, excellent management system, etc.

The 7U high 10-blade industrial standard blade server has independent intellectual property rights in China, built-in Connect X DDR two-layer exchange modules were applied, the built-in management Ethernet switch module is able to achieve 40CPU of single case at the most, 160 core super high computing density, meanwhile, OpenMP concurrent computing in a single node was achieved. These high density computing may greatly save the space costs of the user.

Users adopting GirdView may conveniently and effectively manage its high-performance computing center from the levels of hardware, system, application, energy consumption, user behaviors, etc.

Based on PowerConf, the user may implement real-time monitoring on the energy consumption of the overall high-performance computing cluster system, so that to really realize “green computing” and greatly reduce the following use costs of the user.

System structure