CN101226491A - Interactive method for gridding middleware based on component - Google Patents

Interactive method for gridding middleware based on component Download PDF

Info

Publication number
CN101226491A
CN101226491A CNA2008100574104A CN200810057410A CN101226491A CN 101226491 A CN101226491 A CN 101226491A CN A2008100574104 A CNA2008100574104 A CN A2008100574104A CN 200810057410 A CN200810057410 A CN 200810057410A CN 101226491 A CN101226491 A CN 101226491A
Authority
CN
China
Prior art keywords
grid
module
data
resource
component
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.)
Granted
Application number
CNA2008100574104A
Other languages
Chinese (zh)
Other versions
CN100538650C (en
Inventor
郑纬民
武永卫
黄炜元
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tsinghua University
Original Assignee
Tsinghua University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tsinghua University filed Critical Tsinghua University
Priority to CNB2008100574104A priority Critical patent/CN100538650C/en
Publication of CN101226491A publication Critical patent/CN101226491A/en
Application granted granted Critical
Publication of CN100538650C publication Critical patent/CN100538650C/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Stored Programmes (AREA)

Abstract

The invention relates to a mesh intermediate interact method based on components, belonging to high-performance mesh computer technical field. The invention is characterized in that the method comprises dividing mesh into a resource layer, a core module layer and a mesh manage layer, dividing the mesh manage layer into an execution, a data, an information and a safe module, using a mesh component module GCM to modularize the four modules into sub components, combining the sub components to form a composite component, therefore, the sub components can be call the execution, data, information and safe sub components relative to different mesh intermediates, to realize the interaction between different mesh intermediates. The invention has strong versatility.

Description

Based on the mutual exchange method of the grid middleware of assembly
Technical field
Belong to distributed computing technology and system field based on the mutual exchange method of the grid middleware of assembly, relate in particular to high-performance grid computing technology field.
Background technology
The field that people find the solution problem in the modern scientific research is in continuous expansion, and the complexity of problem, scale be all in continuous expansion, especially high-end sciemtifec and technical sphere, and as high-energy physics, bioinformatics etc.Addressing these problems needed computing power is also increasing substantially.Independent PC or workstation can not satisfy computation requirement.Along with the fast development of computer and network technologies, a lot of tissues and R﹠D institution all have the very strong supercomputer of computing power, as cluster, parallel computer etc.But these machines are often because just isolated not playing one's part to the full for our unit serves is in idle condition in the most of the time.Therefore, break the restriction of region, the collaborative all kinds of resources that extensively distribute on network of using have become new requirement.In order to satisfy above-mentioned this requirement, its target is to realize the shared and collaborative work of heterogeneous resource on the grid virtual environment, eliminates information island and resource isolated island just for the proposition of gridding technique and development.
In order to make things convenient for transparent visit gridding resource, different academic research tissue and companies have developed many grid middlewares.Middleware is the software layer between network layer and application layer, and it can provide services such as identification, affirmation, mandate, catalogue and safety.Middleware is not only realized interconnection, the interoperability between also will realizing using.It is based on the software of distributed treatment, and the most outstanding characteristics are its network communicating functions.Grid middleware also should have following characteristics: satisfy the needs of widely applying; Run on multiple hardwares and operating system platform; Support Distributed Calculation, the mutual of the application of the transparency of across a network, hardware and operating system platform or service is provided; The agreement of support standard; The interface of support standard.
Yet in the grid field of today, different application also lacks a unified standard according to the grid middleware of the needs exploitation oneself of oneself between some universal middlewares.Can't be mutual mutually between the different middlewares, the disunity of grid middleware has caused bigger information island to a certain extent.This in order to eliminate by the information island that grid middleware brought, become more and more important mutually alternately between the different grid middlewares.Comprising two layers of meaning mutually alternately: phraseological mutual mutually---different middlewares are intercoursed information according to a unified standard; Semantically mutually mutual---different middlewares are resolved the information of exchange, understand and these information of being suitable for.
Reach the mutual many difficult points and the problem of existing alternately of syntax and semantics:
Lack a unified mesh standard.Up to the present, the unified standard of grid circle neither one.Even proposed some standards, as OGSA, JSDL, but these standards are not accepted by all grid groups.Current grid middleware is not supported these standards fully.
The diversity that grid is realized.Even the standard such as OGSA and JSDL has been arranged, the unified standard of realization neither one of these standards.Different grid middlewares is different to the realization of same standard.
Lack a general solution.The mutually mutual solution of present grid all is to propose at specific grid middleware.Though this method has solved the mutual mutually of grid middleware to a certain extent, is not a method in common.Such as, UNICORE cooperates with Globus, allow the grid that makes up with UNICORE and Globus can intercommunication, but, need to redesign original system so and make it can the new middleware of compatibility if new grid middleware needs and original system interaction.
Owing to exist above-mentioned difficult point, industry member and academia all to spend the mutual interaction problems of very big manpower and materials research grid.The UNICORE of European Union project and U.S. Globus project cooperation, design one and can make UNICORE and the Globus mutual mutual interactive frame of grid mutually, famous grid middleware Condor has also realized mutually mutual with OGSA also having such as ALiEn-EDG projects such as WorldGrid.But the mutually mutual project of these grids all is that specific two or more grid middlewares are realized mutually alternately.The way of all these projects all is to add the driver that needs other mutual mutually grid middlewares in the grid middleware of oneself at present, realizes mutual mutually in this way.There is this conspicuous shortcoming in this mutual implementation mutually.When new grid middleware occurred, all existing grid middlewares all needed to realize original system and the new mutual driver module of middleware, also need to realize to having the interactive module of middleware now with stylish middleware.This is not an a general solution mutual method mutually.
Grid Component Model (GCM) is the component model that is easy to develop grid application of a lightweight.GCM with the Fractal component model as the basis.GCM separates interface with realization, realized enhanced scalability.The GCM component model of a standard comprises:
Content: the assembly content is an abstract entity by assembly controller control.The content of an assembly is made of the binding of sub-component and sub-component.He is the structure of a recurrence.
Controller: controller has comprised the various control function of assembly.The controller of assembly can be carried out any control operation to assembly.
Service end interface: the component interface that receives operational order.
Client-side interface: the component interface that sends operational order.
Functional interface: the functional interface of assembly is realized the concrete function of assembly.
Control interface: any interface of NOT-function aspect is referred to as control interface.
The component model standard of GCM two class components: sub-component and composite component.Sub-component is an independent functional module, and specific function is provided, and provides normalized interface, for other module invokes.Composite component itself is not realized any function, and composite component combines with one or more sub-components, and a complete assembly operating flow process is provided.
Fig. 1 has provided the example of a GCM assembly.GCM is a model based on assembly, and as can be seen, GCM has following characteristics from the example of Fig. 1:
Supporting reflex: GCM supports introspection and request mechanism simultaneously.
Hierarchical structure: grid application is made up of a lot of distinct programs usually.Each program is made up of littler program element again.This is the structure of a multi-level recurrence.GCM is the model of a separationization.
The extensibility of model: the GCM model allows the adding of new characteristic.Make that the GCM model is an extendible model.
Adaptivity: the GCM model is a kind of adaptive model for grid and its modular construction of facing.
At mutual mutual emphasis and the difficult point of different grid middlewares, we have proposed the scheme based on assembly, realized mutual mutually one of different grid middleware general, can expand, high performance method.
Summary of the invention
The object of the present invention is to provide a kind of mutual exchange method of grid that all is suitable for for all grid middlewares, to eliminate the problem of detached island of information that can't intercommunication forms owing to different grid middlewares.This method is different with the mutual exchange method of existing grid middleware, it does not solve problem from certain or certain several grid middlewares, but at first be three layers basic structure with grid dividing, take out the necessary functional module of all grid middlewares then, with modular assemblyization the most basic, general in the grid, realize the mutual mutually of all grids by assembly.
Core of the present invention has been to adopt Grid Component Model (GCM) component model to realize mutually mutual between the different grid middlewares.GCM is a kind of light-duty dynamic component model, utilizes GCM to realize that the modularization of grid nucleus module can obtain higher performance.
Described method realizes according to following steps on grid:
Step (1) is grid dividing following three layers of basic structure:
Resource layer is made of many computing nodes and memory node;
The nucleus module layer is based upon on the described resource layer, is made of many computing units and storage unit, and described computing unit and storage unit are the abstract of physical resource;
The mesh-managing layer, it comprises job manager and workflow engine at least, is based upon on the described nucleus module layer, is the grid application service;
Step (2) utilizes grid assembly model GCM that described mesh-managing layer is carried out modularization according to the following steps;
Step (2.1) turns to following four modules described mesh-managing layer is abstract: execution module, data module, information module and security module, constitute a grid middleware, wherein,
Execution module is the core of described grid middleware, and it is responsible for deployment, execution, scheduling and the life cycle management of grid work, also monitors simultaneously and coordinates different grid works;
Data module is all data of mesh-managing, for the user provides a unified view, carries out following operation:
Storage administration turns to a disk all data resources are virtual, and basic file operation is provided simultaneously;
Data run, the user uploads the input data that grid work needs, the result data of download grid operation;
Information module is controlled whole gridding resource, offers user's Service Source comprising physical resource and grid, adopts expandable mark language XML to represent resource;
Security module adopts grid security framework GSI to carry out grid authentication;
Step (2.2) is packaged as sub-component to described execution module, data module, information module and security module with described GCM grid assembly model according to the following steps:
Step (2.2.1), described four modules separately assembly turn to sub-component;
Step (2.2.2), the mode that external unified interface after the execution module modularization is described with operation is expressed as: import the order that needs to carry out into, obtain the file that needs to carry out from the virtual data space, insert each parameter of gridding task, comprising the state of the gridding task of file, the execution of input, the gridding task that obtains, and the execution result that obtains from the virtual data space;
Step (2.2.3), the mode that the external unified interface after the data module modularization is described with operation is expressed as: download the file into this locality to the virtual data space, from the virtual data space, return all grid files, create a directory, delete the file in virtual data space and the catalogue in deletion virtual data space in the virtual data space with described XML form from local upload file;
Step (2.2.4), the mode that the external unified interface after the information module modularization is described with operation is expressed as: register a new gridding resource that adds, and browse already present resource in the grid;
Step (2.2.5), the external unified interface after the security module componentization is the user rs authentication interface;
Step (3), clicking step makes up the resulting sub-component of step (2), form a composite component, so that execution, data, information and the safe sub-component of the different grid middleware correspondences of each component call, to realize the intercommunication between the different grid middlewares;
Step (3.1), the user calls security component and carries out the grid checking by the GSI security system;
Step (3.2), the information assembly of user inquiring grid obtains available gridding resource;
Step (3.3), the user calls data package, and required input file is uploaded in the grid virtual data space;
Step (3.4), the user uploads the file that carry out, inserts the gridding task parameters needed;
Step (3.5), the user calls executive module and carries out gridding task;
Step (3.6), the user obtains the state of task by executive module;
Step (3.7), task are carried out and are finished, and the user obtains output file.
The invention has the advantages that: versatility, the mutual exchange method based on assembly that is proposed among the present invention are not to be suitable at some or certain several grid middleware, but to all grid middlewares method in common all; Extensibility, when new grid middleware occurred, this method also can well adapt to new middleware, can make original middleware and new middleware mutual easily; Dynamic, for the execution of each grid work, this method can be called the gridding resource in a plurality of different grids; High-performance, because this method employing is the GCM component model, the lightweight characteristic of this assembly makes the expense of being brought by assembly be reduced to minimum, thereby keeps the original high-performance of grid.
Description of drawings
Fig. 1: GCM component model example
Fig. 2: the three layer model of grid
Fig. 3: utilize GCM at CGSP, GT4, the Componentized grid work of carrying out on the EGEE gLite
Fig. 4: the execution under the many grid platforms of grid work
Fig. 5: the execution flow process of grid work under the present invention
Fig. 6: example-dna sequence dna comparison
Embodiment
As can be seen, GCM is very suitable as the component model of different grid middleware intercommunications from the characteristics of GCM.This is because at the mutual mutual difficult point that runs into of the different grid middlewares of mentioning in preamble, GCM as mutual interactive component model, can both be solved:
1) GCM itself can be taken as be a kind ofly use gridding resource, dispose grid work, agreement that execution and control mesh are used.Simultaneously, GCM provides some additional functions, for example life cycle management, interface management;
2) solve the diversity that grid is realized, current common practices is to write the intercommunication that realizes different grid middlewares with the driver of its mutual grid system for each grid system.GCM has a unified sets of interfaces.Can realize intercommunication by calling this unified sets of interfaces between the different grids;
3) from the level of grid, GCM is deployed on all grids.All grid systems can abstractly be some nucleus modules, and GCM is with these nucleus module modularizations.Therefore, GCM is deployed on all grid systems, does not need to do any change for existing grid system;
More than three promises do not changing on the basis that grid environment and middleware are arranged now, use the GCM component model to realize the intercommunication of different grid middlewares.GCM dynamically disposes simultaneously, and this has more increased the dirigibility of grid application.And GCM itself is the component model of lightweight, and the influence to grid system performance that is brought by GCM can drop to minimum.
Though grid middleware has nothing in common with each other, observing codes and standards separately, different separately realizations is also arranged.But generally speaking, all at present grid middlewares can abstractly be four module: execution module, data module, information module and security module and three levels: mesh-managing layer, nucleus module layer and resource layer.
Through the investigation to existing all grid systems, we can therefrom take out four most crucial modules of grid middleware.Below four modules of Jie Shaoing are the nucleus modules that all have of all grid middlewares now, if system has had the basic function that these four modules just can be finished grid.
1. execution module
Execution module is the core of whole grid middleware.Be responsible for deployment, execution, scheduling and the life cycle management of grid work.Execution module also needs to monitor and coordinates different grid works simultaneously.
2. data module
The execution of any operation all be unable to do without input data and output data.For grid work, the data volume of input and output is quite big.Data module is exactly to be all data of mesh-managing.Grid is an isomery, dynamic environment.Grid data management is different with traditional data management, needs the transparence of deal with data, problem such as virtual, to offer unified view of user.
Data management comprises the basic function of two aspects:
Storage administration: the bottom grid resource unified management with all isomeries makes it to become a unified resource.Memory management functions is a disk with thousands of data resource virtualizings, and from user's angle, what see is a great data resource of capacity.Storage administration is similar to the file system function of operating system, and it provides basic file operation.
Data transport: upload/data download is the committed step that grid work is carried out.Grid user is uploaded the needed input data of grid work by the data transport function, grid work is downloaded the input data, when operation is complete, grid work uploads to output data in the grid system by data transport, and the user can download the output result for the scientific research analysis by data transport once more.
3. information module
Information module is finished the function of resource discovering.Here said resource comprises that physical resource and grid can offer user's service.Physical resource comprises computational resource and storage resources again.Information module is being controlled the resource in the whole grid, and the normal operation of grid is played crucial effects.The difficult point of the integration of the information module of different grid middlewares is that different grid middlewares has different separately resource representation modes.Have some middlewares to use URL to represent a resource, the use XML document that has is represented resource.Because XML is a kind of extendible SGML, in order to adapt to existing grid middleware, can satisfy following demand simultaneously, the present invention has adopted the expression mode based on the XML document resource
4. security module
Existing grid middleware adopts GSI (Grid Security Infrastructure) security framework mostly.The GSI framework is the integration of multiple existing safety technique.
Whole grid also has the gridding resource (comprising physical resource and service) of bottom except these four nucleus modules, also have the administration module on upper strata.So a typical grid system can be divided into one three layers system, as shown in Figure 2.
1. resource layer: resource layer is made of thousands of computing nodes and memory node;
2. nucleus module layer: the nucleus module layer is based upon on the resource layer basis, is made of many computing units and storage unit.Here said computing unit and storage unit are the abstract of physical resource;
3. mesh-managing layer: such as job manager, workflow engine etc.This one deck mainly is based upon on the nucleus module layer, for grid application provides a more convenient and powerful function;
These three layers can be used the GCM component model to carry out modularization respectively, realize the mutual mutually of grid middleware.
1. resource layer is mutual mutually: with all computing nodes and back end respectively assembly turn to computation module and data package.This is equivalent to pack one deck on each gridding resource, makes different gridding resources have unified use-pattern.Like this, any one grid application can be used the grid system that is made up by different grid middlewares by this unified use-pattern.This is a kind of pattern the most flexibly.
2. the nucleus module layer is mutual mutually: with the nucleus module modularization of grid.This one deck mainly is computing unit and the memory unit packageization with grid than resource layer abstraction level height.
3. the mesh-managing layer is mutual mutually: be the mesh-managing layer on the nucleus module layer.The abstraction level of mesh-managing layer is higher than nucleus module layer.Carry out the characteristics that different grid middlewares can be farthest brought into play in modularization at this one deck.
Intercommunication between the grid middleware of being realized by the modularization of different levels respectively has quality.The present invention be directed to the mesh-managing layer and use the GCM component model to carry out the modularization of grid nucleus module, to realize the intercommunication between the different grid middlewares.This method has plurality of advantages.At first, this method has very big dirigibility.It can make full use of the advantage of GCM and each different grid middlewares.The grid developer can use the controller among the GCM to control various assembly, and they can maximize favourable factors and minimize unfavourable ones from the grid middleware of different characteristics simultaneously.For example, the data module of CGSP provides a virtual data management module, and this is that the GridFTP of Globus Toolkit is not available.A grid application can be calculated on the computing unit of Globus Toolkit, utilizes the virtual easily memory function of CGSP to realize the virtual input and output of data simultaneously.Simultaneously, the intercommunication of the grid middleware that the modularization of grid nucleus module is realized can also be played the effect of maximizing favourable factors and minimizing unfavourable ones to different grids.For example, the grid of a bioinformatics, the scheduling strategy that job scheduler adopts are used bioinformatics has good effect, if this grid is used as other purposes, this dispatching method may obtain very poor result.The scheduling strategy of upper strata grid administration and supervision authorities has been avoided in the modularization of nucleus module, can be according to the concrete configuration schedules strategy that should be used for.Thereby obtain very high performance.
The present invention carries out modularization with the mesh-managing layer in the grid three layer model, utilize the GCM component model that execution module, data module, information module and the security module of grid are packaged as assembly, each assembly can call execution, data, information and the security module of different grid middleware correspondences, to realize the intercommunication between the different grid middlewares.
The modularization process is:
With the mesh-managing layer of a general grid middleware abstract be previously described grid four big nucleus modules.
2. be sub-component with grid four big nucleus module modularizations.Each module after the modularization need expose the interface of standard, all has unified interface after four nucleus module modularizations of different middlewares.Below be the detail of four modular assemblyizatioies and the standard interface of required exposure.
The modularization of execution module
Execution module be responsible for execution, the task status of a gridding task inquiry, the task suitable parameters is provided and obtains return results etc.So after the execution module modularization with different grid middlewares, the interface that executive module is unified is:
Interface Describe
public void ExecCommand(String cmd); Import the order that needs execution into
public void GetProgram(String src); Obtain executable file from the virtual data space
public void FillParameter(DataStrategy[]param); Insert the gridding task parameter, comprise input file
public void Invoke(); Carry out gridding task
public String getStatus(); Obtain the state of gridding task
public void getResult(String localDir); Obtain execution result from the virtual data space
The unified interface of table 1 executive module
The modularization of data module
Data module is responsible for the grid data ADMINISTRATION SUBSYSTEM, need provide virtualized storage for the user, and each user has catalogue and the file of oneself.The user can upload, file in download, creates virtual folder, inquiry file and catalogue, deleted file and catalogue etc.After the data module modularization of different grid middlewares, the data package unified interface is:
Interface Describe
public boolean upload(String localPath,String remotePath,Boolean overwrite); From local upload file to the virtual data space
public boolean download(String localPath,String remotePath,Boolean overwrite); Download the file into this locality from the virtual data space
public org.w3c.dom.Document List(); List all grid files, return with the XML form
public boolean mkDir(String dir); Create a directory in the virtual data space
public boolean delFile(String file); The file in deletion virtual data space
public Boolean delDir(String dir); The catalogue in deletion virtual data space
The unified interface of table 2 data package
The modularization of information module
Information module is controlled whole gridding resource (physical resource and service).Information module must provide the registration and the query function of resource.After the information module modularization of different grid middlewares, the information assembly unified interface is:
Interface Describe
public void Register a new gridding resource
registerResource(Document xmlDom);
public Vector browseResource(); Browse already present resource in the grid
The unified interface of table 3 information assembly
The modularization of security module
Security module control user rs authentication.After the security module componentization of different grid middlewares, the security component unified interface is:
Interface Describe
public Boolean authenticate(); User rs authentication
The unified interface of table 4 security component
3. according to grid application different sub-components is combined as composite component.
4. whole composite component is disposed, carried out.
The grid nucleus module of a grid work after by modularization called ChinaGrid SupportPlatform (CGSP) respectively, Globus Too among Fig. 3] mesh module of kit 4 (GT4) and EGEE gLite, realized the intercommunication between different grids.
To just the atom component that whole each step of grid application is called can be made up after the modularization of grid nucleus module, form a composite component.Then whole composite component is disposed and carried out.
For a typical grid work, experience following steps:
1. the user calls the checking of security component by grid.
2. inquiry gridding information assembly obtains available gridding resource.
3. call data package, needed input file is uploaded in the grid virtual data space.
4. upload executable file.
5. insert all parameters.
6. call executive module and carry out gridding task.
7. obtain the state of task by executive module.
8. task is carried out and is finished, and obtains output file.
Fig. 4 has shown the execution flow process of a grid work.Can be clearly seen that by Fig. 4 our method will be carried out in the flow process all and all regard assembly (sub-component and composite component) as.Because each assembly is separate, all separate complementary interference of executive module, data package, information assembly and security component, these assemblies can call the corresponding nucleus module of different grid middlewares, thereby the different piece that has realized an operation is carried out on different grid middlewares, has realized the mutual mutually of different grid middlewares by the GCM component model.
Fig. 5 has provided the execution flow process of a grid work under the present invention, and table 5 has been described this flow process in conjunction with the interface of preceding each nucleus module listed in the text.
1. initialization security component, executive module, data package and information assembly.2. carry out user rs authentication by security component.3. use data package to upload user data to the virtual data space.4. use executive module that executable program is packaged as grid application, and upload required executable file.5. packaged grid application is registered to the gridding information module by information assembly.6. insert the parameter that grid application needs by executive module.7. execution grid application.8. check task status by executive module.If task do not finish return the 8th the step 10. tasks complete, obtain output file by data package.
Table 5 grid work is carried out flow process and is described
Hardware environment: 1GHz CPU, 512M internal memory, 1G hard disk, the 10M network bandwidth
Software environment: Linux Red Hat F3 system, Java SDK 1.5-04 environment
Use the present invention mainly in two steps: to determine the grid assembly of needs and deployment, execution according to grid application.
Determine the number of the sub-component that needs call according to the flow process of concrete grid application.Each sub-component is got up according to concrete grid application path combination, form a composite component.The composite gridding assembly that combines is disposed execution according to the code requirement of GCM.
Example with dna sequence dna comparison in the bioinformatics illustrates mutual method mutually below.Whole flow process is as follows: the user arrives the grid system identity verification by security component, after by authentication, the user can obtain available resources in the grid by information assembly, the user obtains the source document of the dna sequence dna of required comparison by data package then, the dna sequence dna comparison is a parallel procedure, is divided into two roads and carries out simultaneously.Article one, road, executive module calls the CAP3 application program, and (this program is the bioinformatics program, the Tigr that below mentions, Blast also is the bioinformatics program), the output file that CAP3 was handled passes to the virtual data space by data package then, executive module calls Blast and output file that will be just now as input, calls data package at last again the result of Blast is passed to the virtual data space.Another road also is so, and different is, that call at the beginning is not CAP3, but Tigr.Security component of whole requirements of process, an information assembly, five data assemblies and four executive modules.Each component call be module in the different grid middleware.These atom component are become composite component according to the path combination of this application, as Fig. 6.Then, the composite component of combination is disposed execution according to the code requirement of GCM.

Claims (1)

1. based on the mutual exchange method of the grid middleware of assembly, it is characterized in that described method realizes according to following steps on grid:
Step (1) is grid dividing following three layers of basic structure:
Resource layer is made of many computing nodes and memory node;
The nucleus module layer is based upon on the described resource layer, is made of many computing units and storage unit, and described computing unit and storage unit are the abstract of physical resource;
The mesh-managing layer, it comprises job manager and workflow engine at least, is based upon on the described nucleus module layer, is the grid application service;
Step (2) utilizes grid assembly model GCM that described mesh-managing layer is carried out modularization according to the following steps;
Step (2.1) turns to following four modules described mesh-managing layer is abstract: execution module, data module, information module and security module, constitute a grid middleware, wherein,
Execution module is the core of described grid middleware, and it is responsible for deployment, execution, scheduling and the life cycle management of grid work, also monitors simultaneously and coordinates different grid works;
Data module is all data of mesh-managing, for the user provides a unified view, carries out following operation:
Storage administration turns to a disk all data resources are virtual, and basic file operation is provided simultaneously;
Data run, the user uploads the input data that grid work needs, the result data of download grid operation;
Information module is controlled whole gridding resource, offers user's Service Source comprising physical resource and grid, adopts expandable mark language XML to represent resource;
Security module adopts grid security framework GSI to carry out grid authentication;
Step (2.2) is packaged as sub-component to described execution module, data module, information module and security module with described GCM grid assembly model according to the following steps:
Step (2.2.1), described four modules separately assembly turn to sub-component;
Step (2.2.2), the mode that external unified interface after the execution module modularization is described with operation is expressed as: import the order that needs to carry out into, obtain the file that needs to carry out from the virtual data space, insert each parameter of gridding task, comprising the state of the gridding task of file, the execution of input, the gridding task that obtains, and the execution result that obtains from the virtual data space;
Step (2.2.3), the mode that the external unified interface after the data module modularization is described with operation is expressed as: download the file into this locality to the virtual data space, from the virtual data space, return all grid files, create a directory, delete the file in virtual data space and the catalogue in deletion virtual data space in the virtual data space with described XML form from local upload file;
Step (2.2.4), the mode that the external unified interface after the information module modularization is described with operation is expressed as: register a new gridding resource that adds, and browse already present resource in the grid;
Step (2.2.5), the external unified interface after the security module componentization is the user rs authentication interface;
Step (3), clicking step makes up the resulting sub-component of step (2), form a composite component, so that execution, data, information and the safe sub-component of the different grid middleware correspondences of each component call, to realize the intercommunication between the different grid middlewares;
Step (3.1), the user calls security component and carries out the grid checking by the GSI security system;
Step (3.2), the information assembly of user inquiring grid obtains available gridding resource;
Step (3.3), the user calls data package, and required input file is uploaded in the grid virtual data space;
Step (3.4), the user uploads the file that carry out, inserts the gridding task parameters needed;
Step (3.5), the user calls executive module and carries out gridding task;
Step (3.6), the user obtains the state of task by executive module;
Step (3.7), task are carried out and are finished, and the user obtains output file.
CNB2008100574104A 2008-02-01 2008-02-01 Based on the mutual exchange method of the grid middleware of assembly Expired - Fee Related CN100538650C (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CNB2008100574104A CN100538650C (en) 2008-02-01 2008-02-01 Based on the mutual exchange method of the grid middleware of assembly

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CNB2008100574104A CN100538650C (en) 2008-02-01 2008-02-01 Based on the mutual exchange method of the grid middleware of assembly

Publications (2)

Publication Number Publication Date
CN101226491A true CN101226491A (en) 2008-07-23
CN100538650C CN100538650C (en) 2009-09-09

Family

ID=39858504

Family Applications (1)

Application Number Title Priority Date Filing Date
CNB2008100574104A Expired - Fee Related CN100538650C (en) 2008-02-01 2008-02-01 Based on the mutual exchange method of the grid middleware of assembly

Country Status (1)

Country Link
CN (1) CN100538650C (en)

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101488872B (en) * 2009-01-09 2011-02-09 哈尔滨工业大学 Biological information computing grid system
CN102033777A (en) * 2010-09-17 2011-04-27 中国资源卫星应用中心 Distributed type job scheduling engine based on ICE (internet communication engine)
CN101539861B (en) * 2009-05-04 2012-05-09 江西省电力信息通讯有限公司 Tool for graphical design and verification of general workflow
CN103003830A (en) * 2010-07-20 2013-03-27 国际商业机器公司 Managing and optimizing workflows among computer applications
CN103023983A (en) * 2011-11-24 2013-04-03 卡巴斯基实验室封闭式股份公司 System and method for distributing processing of computer security tasks
CN109949858A (en) * 2019-02-26 2019-06-28 杭州来布科技有限公司 A kind of original document is converted into activation document method
CN110162515A (en) * 2019-04-30 2019-08-23 中国科学院深圳先进技术研究院 A kind of uncoupled elastic data warehouse schema
CN111027221A (en) * 2019-12-13 2020-04-17 中国人民解放军空军工程大学 Modular comprehensive avionics simulation training system architecture based on components

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101488872B (en) * 2009-01-09 2011-02-09 哈尔滨工业大学 Biological information computing grid system
CN101539861B (en) * 2009-05-04 2012-05-09 江西省电力信息通讯有限公司 Tool for graphical design and verification of general workflow
CN103003830A (en) * 2010-07-20 2013-03-27 国际商业机器公司 Managing and optimizing workflows among computer applications
CN103003830B (en) * 2010-07-20 2016-05-11 国际商业机器公司 The method and system of the workflow between management and optimizing computer application
CN102033777A (en) * 2010-09-17 2011-04-27 中国资源卫星应用中心 Distributed type job scheduling engine based on ICE (internet communication engine)
CN102033777B (en) * 2010-09-17 2013-03-20 中国资源卫星应用中心 Distributed type job scheduling engine based on ICE (internet communication engine)
CN103023983A (en) * 2011-11-24 2013-04-03 卡巴斯基实验室封闭式股份公司 System and method for distributing processing of computer security tasks
CN103023983B (en) * 2011-11-24 2015-08-26 卡巴斯基实验室封闭式股份公司 For the system of the process of distributed computer safe task
CN109949858A (en) * 2019-02-26 2019-06-28 杭州来布科技有限公司 A kind of original document is converted into activation document method
CN110162515A (en) * 2019-04-30 2019-08-23 中国科学院深圳先进技术研究院 A kind of uncoupled elastic data warehouse schema
CN111027221A (en) * 2019-12-13 2020-04-17 中国人民解放军空军工程大学 Modular comprehensive avionics simulation training system architecture based on components
CN111027221B (en) * 2019-12-13 2022-04-08 中国人民解放军空军工程大学 Modular comprehensive avionics simulation training system based on components

Also Published As

Publication number Publication date
CN100538650C (en) 2009-09-09

Similar Documents

Publication Publication Date Title
CN100538650C (en) Based on the mutual exchange method of the grid middleware of assembly
CN100530160C (en) Architecture for distributed computing system and automated design, deployment, and management of distributed applications
US20200004570A1 (en) Dynamically scaled hyperconverged system
US10057377B2 (en) Dynamic resolution of servers in a distributed environment
CN101276271A (en) Method and interceptor system facing to tangent plane programming
CN101946258A (en) Model based deployment of computer based business process on dedicated hardware
US20060173894A1 (en) System and methods for capturing structure of data models using entity patterns
CN102523225A (en) Decoupling method in distributed loosely-coupled framework and distributed loosely-coupled framework platform
CN105204863B (en) Unlawful data auditing system
Bal et al. Real-world distributed computer with ibis
KR101056682B1 (en) A weapon simulation system and the same method based on the component
KR20080038350A (en) Component architecture
CN102323879A (en) Complete network middleware operation support platform and integration method of heterogeneous middleware
CN102523101A (en) Machine manager service fabric
CN102314358A (en) Method for deploying conventional applications on cloud platform in SOA (service oriented architecture) way
Li et al. Research on distributed architecture based on SOA
CN103455647A (en) Reflective-type object-oriented simulation engine
Tarricone et al. Grid computing for electromagnetics
US7636911B2 (en) System and methods for capturing structure of data models using entity patterns
CN104735111A (en) Method for achieving ESB by adopting WEBSERVICE as uniform interface
CN101861576A (en) Network operating system
CN103631645A (en) Digital family middleware system based on digital medical treatment
CN100412793C (en) Integrated framework for web container in J2EE application server
KR101319815B1 (en) Cloud computing-based App implementing system using distributed multi-mobile terminal
CN103677842A (en) Software tool configuration type integrated expansion calling method and system

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

Granted publication date: 20090909

Termination date: 20170201

CF01 Termination of patent right due to non-payment of annual fee