CN100588197C - Gridding emulation method and its emulator - Google Patents

Abstract

The invention relates to a network simulation method which comprises: information distribution, network solid construction, analogue program generator, analogue program operation, and analogue resultoutput. The simulator with said invention comprises a information distribution module, a network grid solid construction module, a simulation program generation module, a simulation program operationmodule, and a simulation result output module. The simulator builds models of user solid, mission solid, resource solid, message catalog solid, distributor solid, network solid and statistic analyzessolid, to support accurately simulating the network grid application environment. The simulator uses the support of disperse mission simulated basic structure, via independent process simulates each solid in said network grid system, while the process cycles repeated, it simulates the solid motion and processes the information of external solid. The simulator has better user interface and abundantmode supports, while it can supply relative data statistic and analyze tool to effectively reflect the simulation result.

Description

Grid simulation method and simulator thereof

Technical Field

The invention belongs to the technical field of grid systems and simulation thereof, and particularly relates to a grid simulation method and a simulator thereof, which are suitable for the research of simulation evaluation of grid architecture, scheduling strategy, application execution performance and the like.

Background

With the continuous development and updating of information technology, people have more and more demands on computing power and data storage capacity, and meanwhile, closer cooperation and resource sharing are needed among the people. At the heart of grid technology is the sharing and efficient aggregation of computing resources, information resources, storage resources, and a wide variety of applications, services, and decision support systems across a wide area.

Designing a grid system is a very complex system engineering that requires many considerations. For example, the effectiveness of resource management and scheduling in a grid is critical to the success or failure of a grid system; the heterogeneity, security and network performance brought by wide-area sharing of resources are not negligible; in addition, fault tolerance, scalability, and adaptability are all considerations that a grid designer needs to consider. Therefore, grid system designers need to ensure that it is feasible and performs efficiently as desired before actually deploying a newly designed system.

In general, a real grid environment is dynamically changing, uncontrollable. In such an environment, a system designer has little ability to explore all of the performance issues of the system design. Taking the operation of the grid application as an example, the grid application is operated on a virtual machine, the virtual machine does not exist before the operation, and only when the application is actually operated, the grid establishes a virtual machine in real time according to the request of the user for operating the application. In general, the details of the virtual machine, such as the topology between nodes, the node type, the information about the owner of the node, etc., cannot be known, and the virtual machine is also dynamically changed, which is reflected in both the dynamic leaving and joining of resources and the dynamic entering and exiting of applications. If the same application program is operated by two requests, the latter application is almost impossible to repeat the original operation process, and the uncontrollable performance and the unrepeatability of the grid system are fully reflected.

The advent of grid simulators solved many of the above problems. The simulator is used for simulating a grid operating environment, and the operating environment is controllable and reusable. We investigated different problems in this simulated environment, such as feasibility and performance issues. In the simulated environment, various application scenes in the real environment can be simulated more truly by configuring different parameters, so that the simulation process is more realistic; more importantly, the same application program can be repeatedly operated in the same environment without limit, which is the basis of performance analysis comparison, and by analyzing the result of the test on the simulator, a grid researcher can select the design scheme and improve the system design.

Disclosure of Invention

The invention aims to provide a grid simulation method which can simulate a grid system with a complex architecture through software and provide a simulation tool for the research of organization architecture, scheduling strategy, application execution performance and the like of the grid system; the invention also provides a simulator realized according to the method.

The invention provides a grid simulation method, which comprises the following steps:

(1) setting basic configuration information of a simulation grid, wherein the configuration content comprises simulation parameter configuration, user definition, task definition, resource definition, information directory configuration, scheduler configuration, network configuration and statistical analysis configuration;

(2) constructing corresponding grid entities according to various configuration information set in the step (1), and establishing network connection among the entities according to network connection information among the entities in network configuration to form an organization structure of a simulation grid; the grid entity comprises a user entity, a task entity, a resource entity, an information directory entity, a scheduler entity, a network entity and a statistical analysis entity; wherein,

the user entity carries out configuration and initialization according to the user definition information; after the simulation starts, the user entity submits the tasks to be executed to the corresponding schedulers in sequence, and finishes after the tasks are all completed and the processing result is received;

the task entity is represented by using a static class structure; the task entity defines various basic attributes required by task execution according to the task definition information, including code quantity of task execution, data quantity required to be processed and data quantity of processed results, and defines the execution conditions of the task on resources;

the resource entity is configured and initialized according to the resource definition information; registering related information of resources to the information directory; receiving a request for inquiring the dynamic information of the resources and returning the current information of the resources; receiving the task from the scheduler, executing the task according to the local scheduling strategy, and returning the execution result of the task to the user; updating the resource change condition of the information directory to the information directory during the task execution period; this process is repeated until an end message is received;

the information directory entity is configured and initialized according to the information directory configuration information; it receives the registration information from the resource, organizes the relevant data of the available resource in the system; according to the setting of the resource updating mode, the information of the resources is actively or passively updated, and the change condition of each resource is dynamically maintained; the information directory responds to the query from the scheduler, searches the resources meeting the conditions according to the corresponding algorithm, and returns an available resource list; this process is repeated until an end message is received;

the scheduler entity carries out configuration and initialization according to the scheduler configuration information; the scheduler entity receives the user request and inquires available resource information to the information directory to obtain an available resource list; inquiring dynamic information from the resources to obtain current information of the resources; selecting resources according to a scheduling algorithm, and submitting tasks to the selected resources; this process is repeated until an end message is received;

the network entity comprises a router, a network flow generator and a network link, and is configured and initialized according to the network configuration information; the router receives the data packet on the link, and sends out the data packet after carrying out routing selection; this process is repeated until an end message is received; the network flow generator generates background flow on the network according to the configuration condition until receiving the end message;

the statistical analysis entity carries out configuration and initialization according to the statistical analysis configuration information; the method collects and records various analog data, responds to the statistical requirements of various data in the simulation process, performs statistics on various index quantities and outputs statistical results; responding to the analysis requirement of the simulation process, carrying out quantitative analysis on the simulation performance and outputting an analysis result; this process is repeated until an end message is received;

(3) generating simulation program codes of the entities according to the configuration information and the network connection information of the entities, and finally combining the simulation entities and the related network connection simulation codes thereof into a complete executable program;

(4) executing the simulation program on a computer system to simulate the operation of the grid system;

(5) and collecting simulation data, counting, analyzing and outputting a simulation result.

The grid simulator realized according to the method comprises an information configuration module, a grid entity construction module, a simulation program generation module, a simulation program execution module and a simulation result output module; wherein,

the information configuration module is used for acquiring simulation parameters and various configuration information and providing the information to corresponding modules in the grid entity construction module and consists of a simulation parameter configuration module, a user definition module, a task definition module, a resource definition module, an information directory configuration module, a scheduler configuration module, a network configuration module and a statistical analysis configuration module; wherein,

the user definition module provides the user configuration information to the user entity construction module to construct a user entity for simulating the user in the grid;

the task definition module provides the task configuration information to the task entity construction module to construct a task entity for simulating tasks in the grid;

the information directory configuration module provides the information directory configuration information to the information directory entity construction module to construct an information directory entity for simulating an information directory in a grid;

the network configuration module provides the network configuration information to the network entity construction module to construct a network entity for simulating the network in the grid;

the scheduler configuration module provides the scheduler configuration information to the scheduler entity construction module to construct a scheduler entity for simulating the scheduler in the grid;

the resource definition module provides the resource configuration information to the resource entity construction module to construct a resource entity for simulating the resources in the grid;

the statistical analysis configuration module provides the statistical analysis configuration information to the statistical analysis entity construction module to construct a statistical analysis entity for obtaining simulation information in the grid simulation process;

the grid entity construction module consists of a user entity construction module, a task entity construction module, a resource entity construction module, an information directory entity construction module, a scheduler entity construction module, a network entity construction module and a statistic analysis entity construction module; wherein,

the user entity construction module is used for generating a user entity for describing the behavior of the user;

the task entity construction module is used for generating task entities and is used for describing user tasks and various attributes thereof;

the resource entity construction module is used for generating resource entities for describing the computing resources and the storage capacity of the system;

the information directory entity construction module is used for generating an information directory entity for providing and maintaining various information of resources;

the scheduler entity construction module is used for generating a scheduler entity for completing resource allocation when the task is submitted;

the network entity construction module is used for generating a network entity, comprises a router, a network flow generator and a network link and is used for completing data transmission in the system;

the statistical analysis entity construction module is used for generating a statistical analysis entity and is used for acquiring relevant simulation data in the whole simulation process and carrying out statistics and analysis;

the simulation program generation module generates simulation program codes of the entities according to the configuration information and the network connection information of the entities and combines the simulation entities and the related network connection simulation codes into a complete executable program;

the simulation program execution module executes the simulation program on the computer system by utilizing the support of the discrete event simulation infrastructure to simulate the operation of the grid system;

and the simulation result output module collects simulation data, counts and analyzes the simulation result and outputs the simulation result.

The grid simulator of the invention has the following characteristics: (1) a good user interface helps a user to quickly model; (2) abundant simulation elements support various types of entity modeling; (3) allowing a user to participate in model design and expanding the existing model library; (4) the simulation data statistics and the automatic simulation data analysis are detailed, and the simulation result is visually reflected through a chart.

Drawings

FIG. 1 is a flow chart of a grid simulation method of the present invention;

FIG. 2 is a flow chart of simulating a user entity according to the present invention;

FIG. 3 is a flow chart of simulating resource entities according to the present invention;

FIG. 4 is a flow chart of the present invention for modeling information directory entities;

FIG. 5 is a flow chart of a simulation scheduler entity of the present invention;

FIG. 6 is a flow chart of a simulated network router entity of the present invention;

FIG. 7 is a flow chart of a process of a statistical analysis entity of the present invention;

FIG. 8 is a timing diagram of message processing between simulation entities of the present invention;

FIG. 9 is a block framework diagram of the grid simulator of the present invention;

FIG. 10 is a flow chart of the use of the grid simulator of the present invention.

Detailed Description

The invention is described in detail below with reference to the figures and examples.

As shown in fig. 1, the method of the present invention comprises the steps of:

the first step is as follows: setting basic configuration information of the simulation grid, wherein the configuration content comprises simulation parameter configuration, user definition, task definition, resource definition, information directory configuration, scheduler configuration, network configuration and statistical analysis configuration. The simulation parameter configuration includes the number of times the simulation process is repeatedly performed. The utility configuration information includes the number of users, the scheduler and the selection of the information directory. The task configuration information includes the number of tasks of each user, the attributes of the tasks (the size of the tasks, the size of data to be processed for executing the tasks, the size of result data obtained after the tasks are executed), and the association information between the tasks. The resource configuration information includes the number of resources, and the attributes of the resources (the type of the resource, the number of processors owned by the resource, the arithmetic capability of each processor, and the storage capability of the resource). The information directory configuration information includes the setting condition of each service layer in the information directory and the aggregation mode thereof. The scheduler configuration information includes the scheduling algorithm used by the scheduler. The network configuration information includes definitions of routers and network traffic generators in the system, their network connections to the various entities, associated routing algorithms, and specification of network traffic generation patterns. The statistical analysis configuration information comprises a user-defined statistical index quantity and statistical method, a statistical output mode and the specification of an analysis algorithm of statistical data.

The second step is that: various entities in the grid system are constructed, and network connection among related entities is established to form an organization framework of the grid system. The specific method comprises the following steps:

and constructing corresponding grid entities according to various configuration information set in the first step, and establishing network connection among the entities according to network connection information among the entities in the network configuration, thereby forming an organization structure of the simulation grid. The grid entity comprises a user entity, a task entity, a resource entity, an information directory entity, a scheduler entity, a network entity and a statistical analysis entity.

In the grid simulation method of the present invention, the behavior of the user includes: and selecting a task scheduler, transmitting a task request and receiving a processing result of the task. In the user entity, a user requests to perform one or more tasks. For the case of multiple tasks, the tasks are submitted for execution in sequence according to the sequential association among the tasks. As shown in fig. 2, the user entity is configured and initialized according to the user defined information. After the simulation starts, the user entity submits the tasks to be executed to the corresponding schedulers in sequence, and finishes after the tasks are all completed and the processing result is received.

In the grid simulation method of the present invention, the static depiction defining an application in the grid comprises two parts: problems and schedulers and algorithms used to solve the problems; and assuming that the simulation process is performed at the task level, the problem is defined by a set of associated tasks, while the scheduler and its algorithms define the relevant processes for resource selection and task scheduling. The task entities are represented using a static class structure. The task entity defines various basic attributes required by task execution according to the task definition information, including the code amount of the task execution, the data amount required to be processed and the data amount of the processed result. Furthermore, the execution conditions of the task on the resource (resource type, computing power, storage power, etc.) are also defined. The execution association between tasks is described by a directed acyclic graph that a user can automatically generate by a scripting tool by configuring the relevant input files.

In the grid simulation method of the present invention, resource entities are used to describe the computing resources of the system and its storage capabilities. Assuming that the performance of a computing resource is described by its computing power, each computing resource may own one or more computing nodes, which in turn may own multiple processor units. As shown in fig. 3, the resource entities are configured and initialized according to the resource definition information. After initialization, each resource entity performs the following operations: registering related information of resources to an information directory; receiving a request for inquiring the dynamic information of the resources and returning the current information of the resources; receiving the task from the scheduler, executing the task according to the local scheduling strategy, and returning the execution result of the task to the user; and updating the resource change condition of the information directory to the information directory during the execution of the task. This process is repeated until an end message is received. The execution time of a task on a resource entity is determined by the following factors: the number of floating point operations contained by the task, the local scheduling policy of the resources, the computational power of the nodes, and the impact of the I/O processing process on the execution of the task. The resource entity may also receive a message from the scheduler to remove a task that is waiting or executing.

In order to more accurately reflect the process of collecting and maintaining the global information in the real grid system, in the grid simulation method, an information directory is abstracted into three parts according to the realized functions of the information directory: the first layer is an information service, which classifies and integrates information from the bottom layer to form the use information of local resources; the second layer is an aggregated information service which is used for aggregating the resource information of a plurality of information services to form an information directory and providing a larger-range information source for the query request; the third layer is a directory service, which is query request oriented, providing query, discovery, and location of resource information. In addition, the directory service can also directly send a request to the information service to inquire the related information of the local resources. As shown in fig. 4, the information directory entity is configured and initialized according to the information directory configuration information. It receives the registration information from the resource, organizes the relevant data of the available resource in the system; according to the setting of the resource updating mode, the information of the resources is actively or passively updated, and the change condition of each resource is dynamically maintained; the information directory responds to the query from the scheduler, searches the resources meeting the conditions according to the corresponding algorithm, and returns an available resource list. This process is repeated until an end message is received.

The scheduler dynamically queries the information directory according to the scheduling policy, locates appropriate resources, and submits tasks. The scheduling policy of the simulation process may be defined by the user, various different types of policies may be implemented by the scheduler, the policies may include various constraints on tasks, users, and resources, and the scheduler entity may react to the real-time grid load by querying the information directory. As shown in fig. 5, the scheduler entity is configured and initialized according to the scheduler configuration information. The core of the system is a scheduling algorithm, the system provides a built-in scheduling algorithm for use, and the scheduling algorithm can be customized by a user. The scheduler needs to query the information directory to obtain the information of the available resources in the current system, and using the information, the scheduler dynamically selects the resources and performs task scheduling. The dispatcher inquires available resource information from the information directory after receiving a user request to obtain an available resource list; inquiring dynamic information from the resources to obtain current information of the resources; and selecting resources according to a scheduling algorithm, and submitting tasks to the selected resources. This process is repeated until an end message is received.

The network entities comprise routers, network traffic generators and network links, and different network models can be organized by different network entities. As shown in fig. 6, the network entity is configured and initialized according to the network configuration information. The router receives the data packet on the link, and sends out the data packet after routing. This process is repeated until an end message is received. The network traffic generator generates background traffic on the network according to the configuration until the end message is received. The connections between the related entities are made over network links. If there are multiple data to be transmitted simultaneously over the network, the system assumes that they share the link in a time-shared manner. The transmission time of the data depends on the size of the data, the bandwidth of the network and the delay.

The statistical analysis entity provides support for collecting various data in the simulation process, various basic state data are recorded in the simulation process, index quantities built in the system and defined by a user are visually output in a chart form, and an analysis result is given through an analysis strategy and an algorithm. As shown in fig. 7, the statistical analysis entity is configured and initialized according to the statistical analysis configuration information. The statistical analysis entity collects various analog data and records the analog data in a proper form; responding to the statistical requirements of various data in the simulation process, performing statistics on various index quantities and outputting statistical results; and responding to the analysis requirement of the simulation process, quantitatively analyzing the simulation performance and outputting an analysis result. This process is repeated until an end message is received.

The third step: generating simulation program codes according to the configuration information and the network connection information of the entities, wherein the simulation program codes comprise corresponding simulation program codes of user entity simulation, task entity simulation, resource entity simulation, scheduler entity simulation, information directory entity simulation, network entity simulation and statistical analysis entity simulation, and finally combining all module entities and related network connection simulation codes thereof into a complete executable program.

Simulation program code refers to program code that is generated according to a simulation parameter configuration module and various grid entity construction modules and can be executed on a simulation infrastructure.

The fourth step: the simulation program is executed on a computer system (including various systems such as a PC, a workstation and a cluster), and the operation of the grid system is simulated. The execution of the simulation program needs the support of a discrete event simulation infrastructure, which is a process-oriented discrete event simulation package. Based on this simulation package, the various entities in the simulation process perform their respective functions. The simulation package generates an independently running thread for each entity according to the simulation program, and the behavior of the entity is described in its thread body method. In addition, the discrete event simulation package can provide simulation time points when each entity executes system behaviors and various controls of the simulation process in time. By acquiring such information, the requirements for grid simulation and statistical analysis thereof can be fulfilled.

As shown in fig. 8, the time sequence of message processing between entities when the emulator code executes is as follows:

(1) the resource registers to the information directory, and the information directory entity receives the registration information from the resource to form a resource directory;

(2) the user submits the task requests to the scheduler entity in sequence according to the configuration;

(3) the dispatcher inquires the information directory and obtains an available resource list returned by the information directory;

(4) respectively inquiring the current state of the resources in the available resource list;

(5) according to the returned dynamic information of the resources and the scheduling strategy of the scheduler, resource allocation and task scheduling are carried out, and the tasks are assigned to proper resources to be executed;

(6) after the task is executed, the execution result of the task is returned to the user entity;

(7) in the process of task execution on the simulation resource, the statistical analysis entity collects various current simulation data from the resource entity for subsequent statistical and analysis processes;

(8) in the simulation process, some new resources may be available, and some resources are invalid, so that the performance of the resources is constantly changed, which requires the information directory entity to update the resource information in time to reflect the change condition of the resources.

The fifth step: and D, collecting the simulation data generated in the step four, counting, analyzing and outputting a simulation result. Respectively carrying out statistics according to the simulation operation result and various index quantities configured by the system; analyzing the specified simulation data and the statistical result according to a related analysis algorithm; and finally, outputting the result according to the requirement.

As shown in fig. 9, the grid simulator implemented according to the above method is composed of an information configuration module, a grid entity construction module, a simulation program generation module, a simulation program execution module, and a simulation result output module.

The information configuration module is used for acquiring the simulation parameters and various configuration information and providing the information to corresponding modules in the grid entity construction module. The information configuration module consists of a simulation parameter configuration module, a user definition module, a task definition module, a resource definition module, an information directory configuration module, a scheduler configuration module, a network configuration module and a statistic analysis configuration module; wherein,

the user definition module provides the user configuration information to the user entity construction module to construct a user entity for simulating the user in the grid;

the task definition module provides the task configuration information to the task entity construction module to construct a task entity for simulating tasks in the grid;

the information directory configuration module provides the information directory configuration information to the information directory entity construction module to construct an information directory entity for simulating an information directory in a grid;

the network configuration module provides the network configuration information to the network entity construction module to construct a network entity for simulating the network in the grid;

the scheduler configuration module provides the scheduler configuration information to the scheduler entity construction module to construct a scheduler entity for simulating the scheduler in the grid;

the resource definition module provides the resource configuration information to the resource entity construction module to construct a resource entity for simulating the resources in the grid;

the statistical analysis configuration module provides the statistical analysis configuration information to the statistical analysis entity construction module to construct a statistical analysis entity, which is used for acquiring simulation information in the grid simulation process.

The grid entity construction module consists of a user entity construction module, a task entity construction module, a resource entity construction module, an information directory entity construction module, a scheduler entity construction module, a network entity construction module and a statistic analysis entity construction module; wherein,

the user entity construction module is used for generating a user entity for describing the behavior of the user;

the task entity construction module is used for generating task entities and is used for describing user tasks and various attributes thereof;

the resource entity construction module is used for generating resource entities for describing the computing resources and the storage capacity of the system;

the information directory entity construction module is used for generating an information directory entity for providing and maintaining various information of resources;

the scheduler entity construction module is used for generating a scheduler entity for completing resource allocation when the task is submitted;

the network entity construction module is used for generating a network entity, comprises a router, a network flow generator and a network link and is used for completing data transmission in the system;

the statistical analysis entity construction module is used for generating a statistical analysis entity for acquiring relevant simulation data in the whole simulation process and carrying out statistics and analysis.

The simulation program generation module generates simulation program codes according to the configuration information and the network connection information of the entities, wherein the simulation program codes comprise corresponding simulation program codes of user entity simulation, task entity simulation, resource entity simulation, scheduler entity simulation, information directory entity simulation, network entity simulation and statistical analysis entity simulation, and finally all module entities and related network connection simulation codes are combined into a complete executable program.

The simulation program execution module executes the simulation program on the computer system by utilizing the support of the discrete event simulation infrastructure to simulate the operation of the grid system.

And the simulation result output module collects simulation data, counts and analyzes the simulation result and outputs the simulation result.

Example (c):

the configuration and operation of the grid simulator of the present invention is described in detail below with reference to fig. 10, using an example:

(1) configuring user, resource and network conditions

On a user, resource and network configuration page, 3 users and 10 resources are created, 2 routers are created, the first two users are connected with one router, the third user is connected with the other router, the routers are connected, and all the resources are respectively connected with the two routers; each resource is configured to be composed of 3 computing nodes, wherein two computing nodes respectively comprise 4 CPUs, the other computing node comprises 2 CPUs, the computing capacity of each CPU is 377MIPS, and the sharing mode of the CPUs is time-sharing; the routing algorithm of the router is configured to be FIFO, and the bandwidth of the network link is 10 Mb/s.

(2) Configuring user tasks

And configuring the tasks of the created users in a user configuration page. The following tasks are configured for each user: the task instruction is 5000 multiplied by 0.8 to 5000 multiplied by 1.2 million pieces in execution length, 12.5 multiplied by 0.8 to 12.5 multiplied by 1.2 kilobytes of data needing to be processed in execution, and 30 multiplied by 0.8 to 30 multiplied by 1.2 kilobytes of processed data.

(3) Configuration information directory

Creating two information service entities and an aggregated information service entity on an information service configuration page; and configuring a hierarchical information directory, namely, associating the first four resources with one information service entity, associating the other six resources with another information service entity, and associating the two information service entities with an aggregated information service entity.

(4) Selecting task scheduling policy

And selecting the available resource rotation strategy for scheduling in the task scheduling strategy option.

(5) Configuring simulation parameters and statistical analysis parameters

Configuring the repetition times of the simulation process to be 2 on a simulation parameter configuration page; the configuration statistical parameter is a default statistical quantity provided by the system.

(6) Generating simulation code according to configuration

On the code generation page, the "generate" button is clicked to generate a simulation code.

(7) Executing simulation code

And clicking an 'execution' button on the code execution page to execute the simulation code.

(8) Statistical analysis and output of results

And in the simulation result page, the statistical results of various index quantities in the simulation process and the related analysis conclusion of the simulation grid are checked.

The above description is only for the best mode of carrying out the invention, and the method for carrying out the invention is not limited thereto, and any modification within the field of the invention without departing from the spirit of the invention should be construed as being within the scope of the invention.

Claims (1)

1. A grid simulation method comprises the following steps:

(1) setting basic configuration information of a simulation grid, wherein the configuration content comprises simulation parameter configuration, user definition, task definition, resource definition, information directory configuration, scheduler configuration, network configuration and statistical analysis configuration;

(2) constructing corresponding grid entities according to various configuration information set in the step (1), and establishing network connection among the entities according to network connection information among the entities in network configuration to form an organization structure of a simulation grid; the grid entity comprises a user entity, a task entity, a resource entity, an information directory entity, a scheduler entity, a network entity and a statistical analysis entity; wherein,

the user entity carries out configuration and initialization according to the user definition information; after the simulation starts, the user entity submits the tasks to be executed to the corresponding scheduler entities in sequence, and the tasks are finished after all the tasks are finished and the processing results are received;

the task entity is represented by using a static class structure; the task entity defines various basic attributes required by task execution according to the task definition information, including code quantity of task execution, data quantity required to be processed and data quantity of processed results, and defines the execution conditions of the task on resources;

the resource entity is configured and initialized according to the resource definition information, and registers the related information of the resource to the information directory entity; the resource entity receives the request for inquiring the dynamic information of the resource, returns the current information of the resource, receives the task from the scheduler entity, executes the task according to the local scheduling strategy, returns the execution result of the task to the user entity, updates the resource change condition of the resource to the information directory entity during the task execution period, and repeats the process until receiving the ending message;

the information directory entity is configured and initialized according to the information directory configuration information; the information directory entity receives the registration information from the resource entity, organizes the relevant data of the available resources in the system, actively or passively updates the information of the resources according to the setting of the resource updating mode, dynamically maintains the change condition of each resource, responds to the query from the scheduler entity, searches the resources meeting the conditions according to the corresponding algorithm, returns the available resource list, and repeats the process until receiving the ending message;

the scheduler entity carries out configuration and initialization according to the scheduler configuration information; the scheduler entity receives the user request and inquires available resource information to the information directory entity to obtain an available resource list, inquires dynamic information to the resources to obtain the current information of the resources, selects the resources according to a scheduling algorithm, submits tasks to the selected resources, and repeats the process until receiving an end message;

the network entity comprises a router, a network flow generator and a network link, and is configured and initialized according to the network configuration information; the router receives the data packet on the link, sends out the data packet after carrying on the routing selection, repeat this process, until receiving and ending the message; the network flow generator generates background flow on the network according to the configuration condition until receiving the end message;

the statistical analysis entity carries out configuration and initialization according to the statistical analysis configuration information; the statistical analysis entity collects and records various analog data, responds to the statistical requirements of various data in the simulation process, counts various index quantities and outputs statistical results, responds to the analysis requirements of the simulation process, quantitatively analyzes the simulation performance and outputs the analysis results, and repeats the process until receiving the end message;

(3) generating simulation program codes of the entities according to the configuration information and the network connection information of the entities, and finally combining the simulation entities and the related network connection simulation codes thereof into a complete executable program;

(4) executing the simulation program on a computer system to simulate the operation of the grid system;

(5) and collecting simulation data, counting, analyzing and outputting a simulation result.

Images