Leading technical interconnection equipment
Furnished with global cutting-edge interconnection technology – InfiniBand. InfiniBand is a high-speed interconnection technology of industrial standards featuring a super-bandwidth of 40Gbps-56Gbps and an ultralow delay of less than 1μs. Coupled with the high-speed PCI-E bus technology of Inspur high-performance server nodes, it ensures the smoothness of system data exchange and hence maximizes the performance of the processor.
The system is also compatible with the conventional 1GbE and 10GbE technologies and offers greater flexibility for the user.
Diversified and flexible node options
Server nodes can be configured flexibly for different types of applications. The Intel® Xeon® processor with a high performance-price ratio can be adopted as the node processor which, coupled with the high bandwidth and the ultralow delay, is expected to elevate peak computing ability and actual computing ability of the whole system, to rival the very top industry products.
TS10000 offers a well-established active storage solution to the client. FC SAN, SCSI, IP SAN and various options are provided according to user application features so as to satisfy the user’s demands, conveniently and flexibly, for high-performance, high-reliability and high-capacity data storage.
Upward scalability to adapt to the future
Users’ needs for upward scalability are fully considered in the design of Inspur TS10000. It is characterized by all-around design and planning in terms of expansion of data processing resources in node, network expansion of data sharing and transmission between nodes, performance and capacity expansion of data storage system, general system architecture and power supply. As such, its performance and scale expand in sync and the early-stage investment and later-stage expansion are coordinated. Relying on our long-term project experiences, we offer each user with expert services and professional proposals and hence help our users save every penny in the course of systems expansion.
Easy management and extensive applications
Inspur TS monitoring system ClusterEngine collectively and remotely monitors and manages the software and hardware resources of the whole system, user group and application processes via a singular Web interface and it further provides multiple pre-alarm modes to report troubles in operating in a timely manner. TS monitoring system supports multi-platform monitoring and management of nodes and displays information in a tabular and graphical manner. These features improve the manageability of the system and reduce the complexity of administration substantially for the system administrator.
Inspur TS 10000 not only has a well-established local clustering operation management and dispatching system, but also realizes the multi-cluster operation allocation across LANs and WANs via Inspur CROWN Virtual Cluster grid experimental environment and hence increases the utilization rate of resources.
Open and generic computing platform
Inspur TS10000 high-performance server is furnished with standard, open and generic software, hardware and infrastructures so that the user can use it conveniently and develop its applications on its platform. Therefore, this platform is integrated with an excellent development and test environment for higher application efficiency and better transplantation. It has been applied extensively to hydrodynamics, CAE simulation, finite element analysis and other scientific and engineering computing applications.
The software and hardware adopted for Inspur TS10000 have all undergone stringent tests for performance, stability and compatibility to ensure compliance with industrial standards and openness requirements, so as to provide a high-performance and high-reliability computing platform to the user.
|Functional module||Management node: user login/cluster management use.
Storage node: cluster data storage center.
|Computing module||Computing node: undertake computing tasks; cluster computing center.|
|High-speed interconnected network||InfiniBand high-speed network or 1GbE is adopted for communication between nodes.|
|Administration network||1Gb exchange network is adopted to fulfill the task of managing and monitoring the cluster system.|
|Infrastructure||Includes the cabinet, independent power supply module, radiating system, video control and switching system, I/O equipment, etc.|
|Software system||Operating system, parallel encoding environment, professional mathematic function library, operating dispatcher, monitoring software, etc.|
|System||2 ~ 1024+ nodes|
|Processor||4 ~ 2048+ Intel® Xeon® processors|
|Memory||2GB ~ 16TB|
|Operating system||Redhat Linux AS4.0 / Redhat Linux AS5.0 / WCCS2003|
|Storage||Can be extended to over 200TB|
|Switch unit||InfiniBand / GigabitEthernet high-speed switch equipment|
|Network performance||InfiniBand bandwidth 40Gbps~56Gbps, delay lower than 1μs|
|Network||High-speed computing network and management network|
|Control console||KVM switch, keyboard, mouse, monitor|
|System management||Inspur TS monitoring software EffiScale, Inspur CROWN Virtual Cluster grid experimental platform|
|Operation dispatching||Inspur ClusterEngine service platform|
|Programming environment||MPI、PVM、OpenMP、C/C ++、Java、Fortran77、Fortran90/95、Perl、MKL、ATLAS、Goto-blas…|
|System size||600 × 1050 × 2070mm (W, D, H, single cabinet)|
|Power requirement||200 ~ 240V 50Hz ± 1Hz|
|Working environment||5oC ~ 35oC humidity 20% ~ 80%|