CN101888414B - Mobile multitask parallel processing system - Google Patents
Mobile multitask parallel processing system Download PDFInfo
- Publication number
- CN101888414B CN101888414B CN 201010221473 CN201010221473A CN101888414B CN 101888414 B CN101888414 B CN 101888414B CN 201010221473 CN201010221473 CN 201010221473 CN 201010221473 A CN201010221473 A CN 201010221473A CN 101888414 B CN101888414 B CN 101888414B
- Authority
- CN
- China
- Prior art keywords
- submodule
- network
- node
- parallel processing
- computing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Images
Landscapes
- Multi Processors (AREA)
Abstract
The invention relates to a mobile multitask parallel processing system, which integrates six computation nodes with independent operating environments in one machine case, takes the factors of ventilation, shock resistance and the like into consideration, ensures the mobility and the portability of a computer integrated system under an emergency condition to adapt to immediate processing of remote sensing data under the emergency condition, and improves data guarantee for an emergency. The hardware of the mobile multitask parallel processing system mainly comprises a data processing node of a five-node multi-core CPU structure, a shared data storage unit and a gigabit ethernet exchanger. The shared data storage unit comprises a data management node having independent processing capacity and four blocks of 1TB hard disks to form a disk array through an RAID disk array card, and the computation nodes and the shared data storage unit are used for establishing a parallel processing platform with a plurality of computation nodes by using a local area network. The mobile multitask parallel processing system is mainly used for carrying large volume data and performing remote sensing data processing on large task parallel computation, and is particularly suitable for emergency responses, call geographic information data services and the like.
Description
Technical field
The present invention relates to the computer cluster technology field, relate in particular to a kind of mobile multitask parallel processing system.
Background technology
It is a kind of computer system that computer cluster is called for short cluster, and it couples together highly closely to cooperate by one group of loose integrated computer software and/or hardware finishes evaluation work; In some sense, they can be counted as a computer, and the single computer in the group system is called node, usually connect by local area network (LAN), but other possible connected mode is also arranged.Cluster computer is commonly used to improve computational speed and/or the reliability of single computer; Cluster computer is than single computer generally speaking, and is much higher such as work station or the supercomputer ratio of performance to price.
Cluster is divided into two kinds of isomorphism and isomeries, and their difference is: whether the architecture that forms between the computer of group system is identical.Cluster computer can be divided into following a few class by function and structure: 1) high availability cluster High-availability (HA) clusters; 2) load balancing cluster Load balancing clusters; 3) HPCC High-performance (HPC) clusters; 4) grid computing Grid computing.
Clustering is a kind of parallel or distributed processing system(DPS), is comprised of the stand-alone computer that much links together, and collaborative work as an independent integrated computer resource is finished same work by many computers.Computer node can be a uniprocessor PC, and the system of work station or multiprocessor (sharing the multiprocessor of storage) has internal memory, IO the device and operating system.Cluster generally refers to two or more computers (node) of linking together.Node can be together, also can physically disperse and links together by LAN.The computer cluster on one link together (on the LAN basis) is single system of picture for user and application program, that such system can provide a kind of reasonable price and can obtain desired properties and the fast and reliable solution of service, and can only reach by more expensive special-purpose shared drive system in the past.
And the design of present cluster computing system is all very huge, the single node design is complicated, be difficult for mobile, harsh to conditional requests such as ambient temperatures, be not easy to carry, the present invention calculates solution for a kind of high portability that the volume that solves the cluster computer existence, portability and high-performance propose, high integration and high performance cluster.
Summary of the invention
The purpose of this invention is to provide a kind of mobile multitask parallel processing system, to realize high portability, high integration, high performance cluster computing system.
For achieving the above object, the present invention adopts following technical scheme:
Drawing and pulling type close-coupled cabinet comprises that electrical control part, signal output, temperature control, shockproof processing and various interface form; Cabinet inside compact in design, bottom are power control part, are the whole system power supply, the middle part is computing node, and arrangement mode is erect-type, has the computing unit of six settings, top is four disks that keep flat, and linking to each other with the control computing unit by RAID makes up the 4TB disk array;
Be integrated in six computing nodes with independent operating environment in the above-mentioned cabinet, each computing node comprises mainboard, multi-core CPU, internal memory, hard disk, pattern displaying unit and network interface;
Shared memory cell has the disk array group that the node control of the computing of Linux environment and management function forms by 4 1TB hard disks, array control unit and one;
Gigabit ethernet switch is 16 mouthfuls gigabit ethernet switch, and the computing node of every platform independent operation is coupled by the binary channels network interface, takies respectively two gigabit exchange mouths.
Described computing node mainboard adopts the miniaturization mainboard, is of a size of 171.45mm * 171.45mm, and power acquisition MicroATX power supply, volume are 125mm * 100mm * 63.51mm;
The system disk of node computer adopts 2.5 cun notebook hard disks, and is installed in the reverse side of unit, processor adopting Intel two nuclear or four core processors.
The long 460mm of described cabinet, wide 240mm, high 380mm, the both sides of described cabinet are designed with the wind hole, and in cabinet one side four ventilating fans have been installed.
Described multi-task parallel is processed the mobile cluster system and is also comprised:
The infrastructure service module is used for realizing the basic communication of intercomputer;
The basic management module is used for realizing that the gridding computer to participating in calculating manages and dispatches;
The grid computing module is used for realizing exchanges data and the parameter transmission of grid computing.
Described infrastructure service module further comprises:
Start the submodule of monitoring network inquiry;
Judge whether to listen to the submodule of network inquiry, if so, then resolve data query, start corresponding handling procedure; If not, then return and continue to judge whether to listen to network inquiry.
Described basic management module further comprises:
Start the submodule of monitoring network connection;
Judge whether to listen to the submodule of network connection, if not, then return and continue to judge whether to listen to the networking connection; If so, then carry out following module;
Set up the submodule of network connection;
Receive the submodule of networking command;
Resolve command, and exectorial submodule;
Send command execution results, and return the submodule that judges whether to listen to network connection.
Described grid computing module further comprises service end module, client modules, wherein:
The service end module comprises:
Determine whether the submodule of service end, if not, then enter client modules; If so, then continue to carry out following submodule;
Send the submodule that available network connects inquiry;
Judge whether the submodule of query response, if not, then other available computing nodes are not found in report, and connecting this computer is unique enabled node; If so, then carry out following module;
Open the network interface card port and start watcher thread, send the submodule that starts the order of network calculations node;
Judge whether to listen to the submodule of connection, if so, then set up network connection, Distribution Calculation task, distribution subtask data;
Report each subtask executing state, until the submodule that finish each subtask;
The submodule that gathers task result;
The client mould is striden and is comprised:
Need to judge whether the submodule of starter node, if so, then enter the service end pattern, if not, then carry out following step;
Send network connection, the submodule of Connection Service end;
Accept the subtask of distribution and the submodule of data;
Subtasking Report Tasks executing state is to the submodule of service end;
Finish the submodule of subtask.
The present invention has the following advantages and good effect:
1) considers the factors such as ventilation, antidetonation, ensured that Computer Integrated System is removable and portable under the emergency condition, to adapt to the in emergency circumstances fast processing of remotely-sensed data, be the emergent Data safeguard that improves;
2) utilize local area network (LAN) to set up the parallel processing system (PPS) of many computing nodes, realize high portability, high integration, high-performance that cluster calculates.
Description of drawings
Fig. 1 is integrated cabinet apparent size figure of the present invention.
Fig. 2 is each hardware cell Rankine-Hugoniot relations figure of the present invention.
Fig. 3 is mainboard in the computing unit of the present invention, power supply and hard disk relation structure diagram.
Fig. 4 is the graph of a relation of shared data storage cell of the present invention.
Fig. 5 is each hardware data exchange of the present invention and control graph of a relation.
Fig. 6 is parallel processing infrastructure service program frame figure of the present invention.
Fig. 7 is parallel processing basic management program frame figure of the present invention.
Fig. 8 is parallel processing communication module service end schematic diagram of the present invention.
Fig. 9 is parallel processing communication module client schematic diagram of the present invention.
Figure 10 is three mixed-media network modules mixed-media graphs of a relation of parallel processing of the present invention.
Embodiment
In order to realize the miniaturization of cabinet, respectively to power supply, computing unit size, arrangement mode, and temperature control etc. carried out well-designed, referring to integrated cabinet apparent size figure shown in Figure 1, long 460mm wherein, wide 240mm, high 380mm, the cabinet front is provided with signal lamp, show respectively six unitary operation situations, other has six concealed types to restart switch, can control respectively six different unit, and cabinet and external interface are provided with display, keyboard and mouse and other USB interface etc.
Fig. 2 has provided the Rankine-Hugoniot relations figure of each hardware node and auxiliary equipment in the cabinet, the cabinet bottom is power supply and 16 mouthfuls of gigabit networking switches, the centre is the computing node of vertical arrangement, and each computing node comprises power supply, computer motherboard, CPU, internal memory and system disk; The cabinet top is 4 SATA 1TB hard disks that are arranged in parallel, and links to each other with the control computing node by the SATA interface line, and makes up disk array by the RAID mode.The below describes in detail the Rankine-Hugoniot relations of the hardware cell in the cabinet:
1. this cabinet inside comprises six computing nodes with independent operating environment, and be integrated in the cabinet, each computing node comprises mainboard, multi-core CPU, internal memory, hard disk, pattern displaying unit and network interface etc., can finish the computing node of the program of independent operating Windows environment or Linux environment, wherein five is computing node, and one is the data storage management node.As shown in Figure 3, in the computing unit 31 be the computing node mainboard, 32 for computing node power supply, 33 for the computing node system disk.
2. shared memory cell has the disk array group that the node control of the computing of Linux environment and management function forms by 4 1TB hard disks, array control unit and one, to realize shared storage and the exchanges data of mass data, be core data control and the administrative center of native system.4 SATA 1TB hard disks link to each other by the RAID RAID card of SATA interface line with the control computing node, make up disk array by the RAID mode, and manage by Linux, realize shared data storage, as shown in Figure 4.
3. gigabit ethernet switch is 16 mouthfuls gigabit ethernet switch, the computing unit of every platform independent operation is coupled by binary channels, take respectively two gigabits exchange mouthful, with the exchanges data of accelerating each computing unit and disk array group and the information exchange of parallel data processing.
4. integrated cabinet is drawing and pulling type close-coupled cabinet, comprises that electrical control part, signal output, temperature control, shockproof processing and various interface form; The cabinet inside compact in design, the bottom is power control part, be the whole system power supply, the middle part is computing node, and arrangement mode is erect-type, has the computing unit of six settings, layout and arrangement mode are as shown in Figure 2, top is four disks that keep flat, and linking to each other with the control computing unit by RAID makes up the 4TB disk array, and it makes up relation and annexation as described in 2..Computing node, control node and shared memory cell all link to each other with gigabit ethernet switch by the gigabit netting twine and make up LAN, as shown in Figure 5.
5. the file I/O of communications protocol NetBIOS Network Based and many disks RAID, the functional utilization network additive storage device of file I/O (RAID storage) shares with the network disk of operating system and the network disk mapping function realizes, can realize storage and the Concurrency Access of mass data.
6. the message transmission of communications protocol TCP/IP Network Based, by the combination of UDP and TCP, realized the information synchronization of system and computational process and result of calculation synchronously.
7. the task distribution is the core of parallel processing with load balancing, independent development extension of network service of the present invention realizes calculating parallel control, Processing tasks walked abreast decompose and the creation task tabulation, then connect computing node, the taking-up task is distributed from task list.If certain computing node is handled current allocating task, then from task list, take out next task and assign to it, reach load balancing with this.
The below describes the present invention in detail with specific embodiment:
(1) compact cabinet design
Housing temperature makes up the natural wind passage mainly by Duct design, by flowing of natural wind, makes the cabinet inside heat radiation.It is the major way of cabinet inside heat radiation that air inside and outside the cabinet flows in order, has therefore designed the wind hole in the both sides of cabinet, and in cabinet one side four ventilating fans has been installed, and with the Air Flow inside and outside the acceleration cabinet, reaches the temperature purpose of control cabinet.
(2) compact designed of computing unit
In order to reduce the shared space of computing unit, mainboard adopts the miniaturization mainboard, is of a size of 171.45mm * 171.45mm, power acquisition Micro ATX power supply, and volume is 125mm * 100mm * 63.51mm.Hard disk adopts 2.5 cun notebook hard disks, and is installed in the reverse side of unit, thereby makes computer unit for the compactest, processor adopting Intel two nuclear or four core processors.
Parallel processing environment of the present invention is described below:
(1) parallel processing hardware structure diagram
Disk array or TB level disk group provide storage and the Concurrency Access ability of mass data, and computing node is finished the calculation task of intensive, and the control node is finished decomposition scheduling and the human-computer interaction function of parallel task.Utilize the gigabit networking switch that each terminal equipment is interconnected, ensure the high-speed data communications demand between each terminal equipment, as shown in Figure 5.
(2) parallel processing implementation procedure
Native system uses Socket blocking model exploitation service routine (hereinafter referred to as the infrastructure service program) take ICP/IP protocol as the basis, and the infrastructure service program mainly realizes the basic communication of intercomputer, and the infrastructure service program frame as shown in Figure 6.
Participate in the necessary infrastructure service program of installing on the node computer that calculates at each, the infrastructure service program is the program (Internet resources are monopolized formula, do not support many instance accesses) of a single example.After the infrastructure service program had been installed, other computers just can be found this computer, and set up network connection by Query Result and this computer as long as use broadcast mode to send inquiry at network.This infrastructure service program specifically may further comprise the steps, referring to shown in Figure 6:
Start and monitor network inquiry;
Judge whether to listen to network inquiry, if so, then resolve data query, start corresponding handling procedure; If not, then return and continue to judge whether to listen to network inquiry.
Network inquiry is detectd as fixed length and is detectd in native system, and size definition is 64 bytes, defines such as table 1:
Definition is detectd in table 1 inquiry
| Identification fields 0-15 | Data segment 16-59 | Port numbers section 60-61 | Reservation paragraph 62-63 |
In data segment, deposit content and the Query Result of inquiry, be used at present depositing the Computer IP address of sending inquiry and the IP address of responding computer.
In infrastructure service program basic unit, native system has been set up a basic management program, and the basic management program is used take ICP/IP protocol as the basis, uses the Socket blocking model, is the program of B/S pattern.The basic management program mainly realizes this gridding computer that participates in calculating is managed and dispatches, comprise computer mode each concrete program of calculating of report, acceptance and coordination, force to finish or stop the concrete program of calculating, the whole resource of report computer operating position, shut down computer, restart computer etc., the basic management program frame may further comprise the steps as shown in Figure 7:
Start and monitor network connection;
Judge whether to listen to network connection, if not, then return and continue to judge whether to listen to the networking connection; If so, then carry out following step;
Set up network connection;
Receive networking command;
Resolve command, and fill order;
Send command execution results, and return and judge whether to listen to network connection.
Participate in the necessary basic management program of installing on the node computer that calculates at each, the infrastructure service program also is the program of a single example.
Network communication is detectd as fixed length and is detectd in the basic management program, and size definition is 512 bytes, and network communication is detectd and defined such as table 2:
Definition is detectd in communication in the table 4-3-2 basic management
| | Command parameter | 1 252 bytes | Command parameter 2256 bytes |
Organization definition is:
struct CMD_HDR
{
UINT cmd；
char strPar1[252]；
char strPar2[256]；
}；
As shown in figure 10, except infrastructure service program, basic management program, the program itself of carrying out grid computing also has a Network communication service module (hereinafter referred to as the network communication module) to carry out exchanges data and parameter transmission in the grid computing.The network communication module is used take ICP/IP protocol as the basis, use the Socket blocking model, it is the program of C/S model, each computer node is as client (Client), and the computer of initiating to calculate is as service end (Server), operation logic such as Fig. 8 and shown in Figure 9, service end may further comprise the steps:
Determine whether service end, if not, then enter client mode; If so, then continue to carry out following step;
Send available network and connect inquiry;
Judged whether query response, if not, then other available computing nodes are not found in report, and connecting this computer is unique enabled node; If so, then carry out following step;
Open the network interface card port and start watcher thread, send the order of startup network calculations node, and carry out following step;
Judge whether to have listened to connection, if so, then set up network connection, Distribution Calculation task, distribution subtask data;
Report each subtask executing state, until finish each subtask;
Gather task result.
Client may further comprise the steps:
Need to judge whether starter node, if so, then enter the service end pattern, if not, then carry out following step;
Send network connection, the Connection Service end;
Accept subtask and the data of distribution;
Subtasking Report Tasks executing state is to service end;
Finish the subtask.
Network communication is detectd as variable length and is detectd in the network communication module, and the head that individual fixed length is arranged is detectd in each network communication, and network communication is detectd and defined such as table 3:
Definition is detectd in communication in the table 3 network communication module
| Detect 32 byte | Detect data (optional) |
Detecing head is defined as:
typedef struct tagNETDAT_HDR
{
UINT datType；
UINT wParam；
UINT lParam；
UINT datSize；
void*pDat；
UINT resv[3]；
}NETDAT_HDR；
The network communication module is supported the program of grid computing to quote by all, therefore this module will be embedded in each program, and for quoting conveniently, we make template library and dynamic base with this module, in application program, only need to comprise its defined file, can use.
The network communication module is to use in concrete application program, is to belong to application layer, and the relation of above three mixed-media network modules mixed-medias as shown in figure 10.
In the native system, the original mass data of using in the computational process, the network mapping that we adopt operating system to provide realizes the mutual exchange of data, the parameter that needs in the parameter that computational process is used and the computational process to exchange is all finished by the network communication module.Native system is considered in hardware designs and namely will be used original mass data also will carry out the exchange of intermediate data, has adopted two network interface cards to realize the load balancing of exchanges data, and original mass data is used separately a network interface card, and intermediate data uses another network interface card.
Above-mentioned example is used for the present invention that explains, rather than limits the invention, and in the protection range of spirit of the present invention and claim, the present invention is made any modification and change, all falls into protection scope of the present invention.
Claims (6)
1. a mobile multitask parallel processing system is characterized in that, comprising:
The infrastructure service module is used for realizing the basic communication of intercomputer;
The basic management module is used for realizing that the gridding computer to participating in calculating manages and dispatches;
The grid computing module is used for realizing exchanges data and the parameter transmission of grid computing;
Drawing and pulling type close-coupled cabinet comprises that electrical control part, signal output, temperature control, shockproof processing and various interface form; The cabinet bottom is power control part, is the whole system power supply, and the middle part is computing node, and arrangement mode is erect-type, has the computing unit of six settings, and top is four disks that keep flat, and linking to each other with the control computing unit by RAID makes up the 4TB disk array;
Be integrated in six computing nodes with independent operating environment in the above-mentioned cabinet, each computing node comprises mainboard, multi-core CPU, internal memory, hard disk, pattern displaying unit and network interface;
Shared memory cell has the disk array group that the node control of the computing of Linux environment and management function forms by 4 1TB hard disks, array control unit and one;
Gigabit ethernet switch is 16 mouthfuls gigabit ethernet switch, and the computing unit of every platform independent operation is coupled by binary channels, takies respectively two gigabit exchange mouths.
2. mobile multitask parallel processing system according to claim 1 is characterized in that:
Described mainboard adopts the miniaturization mainboard, is of a size of 171.45mm * 171.45mm, and power acquisition MicroATX power supply, volume are 125mm * 100mm * 63.51mm;
Hard disk adopts 2.5 cun notebook hard disks, and is installed in the reverse side of unit, processor adopting Intel two nuclear or four core processors.
3. mobile multitask parallel processing system according to claim 1 and 2 is characterized in that:
The long 460mm of described cabinet, wide 240mm, high 380mm, the both sides of described cabinet are designed with the wind hole, and in cabinet one side four ventilating fans have been installed.
4. mobile multitask parallel processing system according to claim 1 is characterized in that:
Described infrastructure service module further comprises:
Start the submodule of monitoring network inquiry;
Judge whether to listen to the submodule of network inquiry, if so, then resolve data query, start corresponding handling procedure; If not, then return and continue to judge whether to listen to network inquiry.
5. mobile multitask parallel processing system according to claim 4 is characterized in that:
Described basic management module further comprises:
Start the submodule of monitoring network connection;
Judge whether to listen to the submodule of network connection, if not, then return and continue to judge whether to listen to network connection; If so, then carry out following module;
Set up the submodule of network connection;
Receive the submodule of networking command;
Resolve command, and exectorial submodule;
Send command execution results, and return the submodule that judges whether to listen to network connection.
6. mobile multitask parallel processing system according to claim 1 is characterized in that:
Described grid computing module further comprises service end module, client modules, wherein:
The service end module comprises:
Determine whether the submodule of service end, if not, then enter client modules; If so, then continue to carry out following submodule;
Send the submodule that available network connects inquiry;
Judge whether the submodule of query response, if not, then other available computing nodes are not found in report, and connecting this computer is unique enabled node; If so, then carry out following module;
Open the network interface card port and start watcher thread, send the submodule that starts the order of network calculations node;
Judge whether to listen to the submodule of connection, if so, then set up network connection, Distribution Calculation task, distribution subtask data;
Report each subtask executing state, until the submodule that finish each subtask;
The submodule that gathers task result;
Client modules comprises:
Need to judge whether the submodule of starter node, if so, then enter the service end pattern, if not, then carry out following step;
Send network connection, the submodule of Connection Service end;
Accept the subtask of distribution and the submodule of data;
Subtasking Report Tasks executing state is to the submodule of service end;
Finish the submodule of subtask.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201010221473 CN101888414B (en) | 2010-07-02 | 2010-07-02 | Mobile multitask parallel processing system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201010221473 CN101888414B (en) | 2010-07-02 | 2010-07-02 | Mobile multitask parallel processing system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101888414A CN101888414A (en) | 2010-11-17 |
| CN101888414B true CN101888414B (en) | 2013-02-13 |
Family
ID=43074130
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201010221473 Expired - Fee Related CN101888414B (en) | 2010-07-02 | 2010-07-02 | Mobile multitask parallel processing system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101888414B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103118103A (en) * | 2013-01-29 | 2013-05-22 | 浪潮电子信息产业股份有限公司 | Cloud server framework capable of achieving multi-node interconnection and management |
| CN103309624A (en) * | 2013-05-27 | 2013-09-18 | 山东卡尔电气股份有限公司 | Large-capacity storage system with parallel multiple channels |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1410904A (en) * | 2002-09-24 | 2003-04-16 | 武汉拓锐计算机系统有限责任公司 | Load balance modulator possessing TCP connection fault tolerant function and its modulating method |
| CN1645396A (en) * | 2005-02-22 | 2005-07-27 | 中国科学院计算技术研究所 | Applying method for linux group system in securities line |
| WO2007029253A2 (en) * | 2005-09-06 | 2007-03-15 | Beyond Blades Ltd. | 3-dimensional multi-layered modular computer architecture |
-
2010
- 2010-07-02 CN CN 201010221473 patent/CN101888414B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1410904A (en) * | 2002-09-24 | 2003-04-16 | 武汉拓锐计算机系统有限责任公司 | Load balance modulator possessing TCP connection fault tolerant function and its modulating method |
| CN1645396A (en) * | 2005-02-22 | 2005-07-27 | 中国科学院计算技术研究所 | Applying method for linux group system in securities line |
| WO2007029253A2 (en) * | 2005-09-06 | 2007-03-15 | Beyond Blades Ltd. | 3-dimensional multi-layered modular computer architecture |
Non-Patent Citations (2)
| Title |
|---|
| Jianqing Zhang,Tao Ke,Mingwei Sun.The Parallel Computing Based on Cluster Computer in the Processing of Mass Aerial Digital Images.《2008 International Symposiums on Information Processing》.2008, * |
| 熊盛武,王鲁,杨婕.构建高性能集群计算机系统的关键技术.《微计算机信息》.2006,第22卷(第3期), * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101888414A (en) | 2010-11-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Jin et al. | A review of power consumption models of servers in data centers | |
| Kristiani et al. | The implementation of a cloud-edge computing architecture using OpenStack and Kubernetes for air quality monitoring application | |
| US20200241999A1 (en) | Performance monitoring for short-lived functions | |
| US10346156B2 (en) | Single microcontroller based management of multiple compute nodes | |
| Durand et al. | Euroserver: Energy efficient node for european micro-servers | |
| CN104335137B (en) | Manage the power consumption and performance of computing system | |
| Pierson | Large-scale distributed systems and energy efficiency: a holistic view | |
| Arthi et al. | Energy aware cloud service provisioning approach for green computing environment | |
| Bambrik | A survey on cloud computing simulation and modeling | |
| CN106302628B (en) | Unified management scheduling method for computing resources in ARM architecture network cluster | |
| US20230164088A1 (en) | Low Latency Queuing System | |
| Ismail | Energy-driven cloud simulation: existing surveys, simulation supports, impacts and challenges | |
| Almeida et al. | Energy monitoring as an essential building block towards sustainable ultrascale systems | |
| CN103984394A (en) | High-density and energy-saving blade server system | |
| CN101888414B (en) | Mobile multitask parallel processing system | |
| Naveenkumar et al. | Performance Impact Analysis of Application Implemented on Active Storage Framework | |
| Hlavacs et al. | Distributed energy efficiency in future home environments | |
| Niu et al. | User-aware partitioning algorithm for mobile cloud computing based on maximum graph cuts | |
| Zamani et al. | A feasibility analysis of power-awareness and energy minimization in modern interconnects for high-performance computing | |
| Liu et al. | Improving the energy efficiency of data-intensive applications running on clusters | |
| Fang et al. | A real-time and reliable dynamic migration model for concurrent taskflow in a GPU cluster | |
| Pereira et al. | A microscope for the data centre | |
| Zhang et al. | Optimising data access latencies of virtual machine placement based on greedy algorithm in datacentre | |
| CN113254151B (en) | Method and device for virtualizing supercomputer system | |
| CN213024363U (en) | Big data mining distributed storage device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20130213 Termination date: 20180702 |