Summary of the invention
The object of the present invention is to provide a kind of city bus information management and scheduling decision back-up system based on grid, this system has Sharing degree height, collaborative work is strong, the system expandability is strong characteristics.
City bus information management and scheduling decision back-up system based on grid provided by the invention is characterized in that: this system comprises master server and is positioned at autonomous territory two parts of distributed grid computing system;
Master server comprises grid portal, workflow engine and authentication center;
Grid portal is responsible for providing the inlet of user capture grid environment;
Workflow engine is responsible for finishing the workflow instance flow process from abstract grid work flow model to concrete grid work flow, obtains the concrete subjob carried out information, at last subjob is distributed to corresponding territory job manager and is carried out actual operation execution;
Authentication center is responsible for verifying and authorize corresponding System Privileges to the subscriber identity information that is received by grid portal;
Comprise domain information service module, territory job manager, domain manager and resource node cluster respectively in the autonomous territory;
The resource node cluster is finished the structure of gridding resource and storage, encapsulation, registration, management, monitoring and the execution of decision-making resource, is the place of realizing the communications policy service;
The information of Grid Metadata in the autonomous territory is in charge of and is safeguarded to the domain information service module, realizes the resource view in autonomous territory; The domain information service module provides the gridding information service by grid portal to the user, and real-time update gridding resource information; Each domain information service module is collected the node information on services of the resource node cluster report in this autonomous territory, and is periodic simultaneously by the resource information in each autonomous territory of the mutual communications exchange of point-to-point mode, to set up the resource mapping of the overall situation;
The territory job manager is responsible for the job request information that collection work stream engine provides, and request responds timely to user job to use the decision-making resource, calls the request execution that fulfils assignment of the resource node in the corresponding resource node cluster in autonomous the territory in; Simultaneously, in the process that operation is carried out, the territory job manager receives the real-time monitor message of service of resource node cluster, makes gridding resource and decision-making resource obtain reasonably to distribute to use by scheduling of resource, improves the balancing dynamic load in the whole autonomous territory;
Domain manager is in charge of the resource of Virtual Organization in the autonomous territory, understand the statistical information of resource in the Virtual Organization and the statistical information of single tissue, the unified interface of inquiry, access resources is provided for the user, coordinate the shared use of resource, and carry out the mutual of resource information with the resource node cluster of lower floor.
System of the present invention utilizes grid, geography information (GIS) technology, GPRS multiple technologies means such as (GPRS), solves a multinomial key technology difficult problem related in the intelligent public transport information system.With regard to the design of public transport information system, the present invention adopts grid to come standard to make up the traffic decision-making back-up system, various decision-making resources are packaged into mesh services, be distributed on each relatively independent grid computing node, carry out the cooperative scheduling of multinode, be based on the solution that present domestic bus dispatching system development survey and analysis of status is drawn.Its characteristics are:
(1) Sharing degree height
The present invention utilizes the powerful data storage technology of grid, well solved the seamless storage problem of mass GPS, GIS database, for the calculating storage of the needed large-scale dataset of decision support is laid a good foundation.From the information sharing angle, the present invention is that the form in autonomous territory manages with resource division.The bus dispatching subsystem of each bus subsidiary is as the autonomous territory of a grid, preliminary preliminary treatment is done to its own a large amount of original transport information separately in each autonomous territory, and as required data processed is packaged into various mesh services, this structure has effectively been broken information isomerism bottleneck, provides high-level more fast and effectively decision information to the policymaker.From system integration angle, the present invention has broken each self-enclosed state of traditional bus dispatching subsystem, by implementing the resource-sharing between unified standard realization heterogeneous system, (being the autonomous territory of grid) closely combined by the service call of loose coupling, realized the conception of intelligent bus information platform.
(2) collaborative work is strong
The present invention is real-time, the dynamic traffic system of trans-sectoral affair, can be dynamically and visit the service of distributed miscellaneous service function smoothly, and utilize the grid work flowable state to be integrated into new operation flow as required, and needn't be concerned about affiliated function, geographical position, internal structure, the execution details of concrete Executive Module, improved the collaborative work ability of system greatly.
(3) system expandability is strong
Because each functional module all adopts the form of mesh services to develop in the invention, and adopt the grid work flow model to make up, increased the loose coupling degree of system greatly.The user can finish required difference in functionality by the reorganization configuration to basic module, and initiate functional module can be called easily by the encapsulation of mesh services form, has avoided problem in traditional closed intelligent transportation system, the problem includes: the repeated encoding problem.
Embodiment
The present invention solves two key issues emphatically: the one, how in existing grid environment, dispose the traffic application service, i.e. the service network deployment of formatting; The 2nd, how the service of having disposed is registered, searches for, creates, carried out to support public transport decision-making, i.e. grid service instantiation process.
Service network of the present invention is formatted to dispose and is related to five parts: resource, operation, information service, the management of autonomous territory and man-machine interaction.Wherein, resource is meant computational resource, storage resources, public transport simulation software and the public transport Legacy System resource that need use, public transport isomeric data resource etc. in grid; Operation refer to mesh services in grid environment from instantiation, service assignment, carry out the process that obtains result of calculation; Resource registering and the search procedure in the grid environment finished in information service; The management of autonomous territory is finished in the autonomous territory of resource in the grid environment between management and autonomous territory alternately; Man-machine interaction then be in service registry, selection, implementation in order to realize the user that specific user's job requirements is taked and the reciprocation of grid environment, realize by user's space and grid portal 1.1.
Below by by embodiment the present invention being described in further detail, but following examples only are illustrative, and protection scope of the present invention is not subjected to the restriction of these embodiment.
As shown in Figure 1, system of the present invention comprisemaster server 1 and be positioned at the distributed grid computing system autonomous territory 2.1,2.2 ..., 2.n two parts, n is a positive integer, represents the number in autonomous territory.
Master server 1 comprises grid portal 1.1, workflow engine 1.2 and authentication center 1.3.
Grid portal 1.1 is responsible for providing the inlet of user capture grid environment, comprises user's space, application program entry and user instruction three parts that enter the mouth.Wherein, the concrete Authorized operation (comprise that the user logins, user profile inquiry, user operation records etc.) of user in the grid environment is in charge of and is monitored to user's space, so that each different user can obtain personalized customize services according to self-demand; Application program entry offers submission of user from writing the operation-interface of application program, the user uses this interface will be certainly writing application deployment behind grid, can directly utilize the computational resource of grid and the Distributed Calculation that storage resources carries out this program, this function mainly is the checking at a large amount of simulation algorithms in the traffic science experiment; The user instruction inlet has then provided the interface of an existing communications policy mesh services of user capture system, the relevant grid decision service of a series of field of traffic that the user has developed by input service call instruction (as service name, data source address etc.) calling system is finished self needed decision support function.
Workflow engine 1.2 is mainly used in the instantiation flow process of finishing abstract grid work flow model and the execution of managing concrete grid work flow.It receives the abstract grid work flow model that the user provides from grid portal 1.1, (wherein abstract grid work flow model belongs to Problem Areas to the workflow instance flow process of concrete grid work flow to finish abstract grid work flow model, concrete grid work flow belongs to solution space, workflow instanceization is promptly set up the process of mapping to abstract grid work flow model and concrete grid work flow by the matching process in territory under striding, finish the mapping of separating of problem), obtain the concrete subjob carried out information and (comprise autonomous territory, operation procedure place, the operation physical address, the sequence of operation etc.), at last subjob is distributed to corresponding territory job manager and carried out actual operation execution.
When the user provide be not direct abstract grid work flow model the time, workflow engine 1.2 can also help the user to set up abstract grid work flow model.At this moment, it receives the decision task information that the user provides from grid portal 1.1, provides the graphics edition interface to help the user to set up abstract grid work flow model.
Authentication center 1.3 is responsible for verifying and authorize corresponding System Privileges to user's identity.It receives the certificate of certification information that the user provides from grid portal 1.1, confirms whether authentication is successful, determines then that as success the user logins qualification and preserves certificate, as the then return authentication failure information of failing.
Autonomous territory 2.1,2.2 ..., comprise respectively among the 2.n domain information service module 3.1,3.2 ..., 3.n, territory job manager 4.1,4.2 ..., 4.n, domain manager 5.1,5.2 ..., 5.n, and resource node cluster 6.1,6.2 ..., 6.n.
Resource node cluster 6.1,6.2 ..., comprise a plurality of resource node a of identical or different number in the 6.n, resource node a can be divided into computational resource node and storage resources node two classes, wherein the computational resource node refers to the resource node that calculates needed hardware system (as CPU, internal memory etc.) and software platform can be provided, and the storage resources node refers to provide the resource node of needed hardware system of storage (as memory, buffer memory) and software platform (as Various types of data base management system, file management system).The structure of all kinds of resource node a is provided with node service container b and some gridding resource h and decision-making resource j all as shown in Figure 2 in it.Be provided with nodal information service module c, node resource service module d, node resource management services module e, grid file host-host protocol (GridFTP) transmission service module f and monitoring resource service module g in the node service container b.
Node service container b is responsible for providing on each node in grid environment that a mesh services is installed, disposed, the environment of operation, monitoring, and provides support for the resource real-time status that monitors grid node.Any node resource service module d must be configured among the node service container b could be searched with call.
The function of nodal information service module c is similar to the domain information service module, and it safeguards local node resource or service metadata information, on the node of resource registering in the autonomous territory, makes that the resource of node or information on services are discerned by grid.For the service that is deployed in the node service container b, nodal information service module c collects its information automatically; Then import resource information by hand for the resource that is not deployed in the node service container b by resource provider.Nodal information service module c will periodically register or collect the information of local node to the report of domain information service module simultaneously.
Node resource service module d is responsible for making resource to call for the territory job manager with the service form of loose coupling to gridding resource h and the encapsulation of decision-making resource j service.Node resource service module d comprises computing node resource service module and memory node resource service module.Wherein computing node resource service module mainly is responsible for providing the calculation task in the grid service instantiation process to calculate required computing environment, and memory node resource service module mainly is responsible for providing the memory space in the calculation task desired data source in the grid service instantiation process.
The function of node resource management services module e is similar to domain manager, is in charge of and coordinates all kinds of gridding resources and decision-making resource of this intranodal, and the real-time resource control information of acceptance domain manager is to assist to finish the resource management requirement of the grid overall situation simultaneously.
Transmission service module (f) adopts GridFTP (grid file host-host protocol), it is responsible in the process of fulfiling assignment needed data transport service between storage resources node and computational resource node, comprises two parts of transmission client module and index server module.The transmission client module mainly is responsible for the foundation of transfer of data between storage resources node and computational resource node, third party's transmission, the re-transmission that makes mistakes in service instantiation process, is destroyed and create or the like; The index server module is responsible for providing the positional information at data place, comprise nodal information catalogue module and two parts of node file index module, wherein, the nodal information catalogue module is responsible for the simple files function management of storage resources node, and node file index module is responsible for the statistic record of storage resources nodal information state, as sink information of the releasing news of the statistic record of file access information, file, file etc.;
The static properties and the dynamic performance parameter of resource monitored on monitoring resource service module g was responsible for when the traffic grid work each cluster, to understand the load and the ruuning situation of whole mesh network network, in order to the usefulness of scheduling.These dynamic and static information of the periodic real-time update of monitoring resource service module g are carried out the reference information of autonomous territory internal burden balance as the territory job manager.Simultaneously, the resource performance that monitoring resource service module g collects same application example utilizes situation, ruuning situation, storage resources and instrument resource ground operating position etc. as computational resource, so that operating single grid application example is carried out Performance Evaluation, resource requirement and consumption situation in the reflection example running are with the dynamic assignment and the scheduling of guiding work stream engine 1.2.
Gridding resource h and decision-making resource j provide the real resource of communications policy support service.Wherein, the grid equipment that gridding resource h refers to and field of traffic is irrelevant, first-class as computational resource, storage resources, network access device, transducer, shooting; The decision-making resource then refers to and the closely-related resource of communications policy, as forecasting traffic flow model, traffic information database, traffic simulation instrument etc.
Domain information service module 3.1,3.2 ..., 3.n is the important component part in the network of communication lines case system bottom architecture, be in charge of and safeguard the information of (comprising various resources and service metadata in the grid environment) of Grid Metadata in the autonomous territory, realize the resource view in autonomous territory, provide the gridding information service by grid portal 1.1 to the user, and real-time update gridding resource information.Each domain information service module is collected the node information on services of the nodal information service module c report in this autonomous territory, and is periodic simultaneously by the resource information in each autonomous territory of the mutual communications exchange of point-to-point mode, to set up the resource mapping of the overall situation.
Territory job manager 4.1,4.2 ..., 4.n is responsible for the job request information that collection work stream engine 1.2 provides, request responds timely to user job to use resource, calls the node corresponding resource service module d request execution that fulfils assignment in autonomous the territory in; Simultaneously, in the process that operation is carried out, the territory job manager receives the real-time monitor message of service of monitoring resource service module g, makes gridding resource and decision-making resource obtain reasonably to distribute to use by scheduling of resource, improves the balancing dynamic load in the whole autonomous territory.In the present invention, the territory job manager adopts the scheduling mechanism based on PBS, constitute by 3 modules, and be respectively client server, job scheduler and node actuator.Wherein client server is the core of entire domain job manager, is responsible for receiving various external request, realizes monitoring, distribution and management to resource; Job scheduler is responsible for carrying out the scheduling of task; The node actuator mainly is the detection of carrying out resource, and feeds back to client server.
Domain manager 5.1,5.2 ..., 5.n is in charge of the resource of Virtual Organization in the autonomous territory, understand the statistical information of resource in the Virtual Organization and the statistical information of single tissue, the unified simple interface of inquiry, access resources is provided for the user, coordinate the shared use of resource, and carry out the mutual of resource information with the node resource management services module e of lower floor.
Format after deployment finishes in service network, the present invention is based on the city bus information management of grid and scheduling decision back-up system and just can call the traffic application service and use to support communications policy.Its typical grid service instantiation process is as shown in Figure 3:
(1) subscriber authentication
(1.1) grid portal 1.1 is accepted the user certificate that the user provides, certificate information comprises Customs Assigned Number, name, login password, affiliated tissue etc., adopt X.509 the certificate of certification format standard to write, and transfer to authentication center 1.3 and be responsible for user's identity is verified;
(1.2) in this example, authentication center 1.3 mainly comprises Light Directory Access Protocol (LDAP) and the online certificate server of MyProxy.At first, authentication center 1.3 passes to ldap directory with the user certificate that receives, the passing user authentication information that writes down in this user certificate information and the ldap directory is compared, if compare successfully, just send the assertion message that an authentication is passed through to the user, if comparison is failed, then send the assertion message of an authentification failure to the user.Then, customer attribute information in the ldap directory query directory, certificate and attribute of user according to the user submits to send an attribute assertion message to the user, determine the authority of user capture system.Simultaneously, user oneself preserves and the process of management private cipher key and certificate file in order to simplify, and authentication center 1.3 uses the online certificate server of MyProxy, preserves and the leading subscriber certificate, for the user fetches, upgrades and create interim short-lived certificates;
(2) information on services coupling
(2.1) service request of domain information service module reception login user submission (wherein, service request information comprises: service name, service-number, address of service, serve port, service data source position etc., and encode) with the form of XML, according to service request information the information on services in the autonomous territory is searched for coupling, (present matching algorithm is mainly considered to mate by identical service name or service-number), and return the WSDL mesh services description document that satisfies the traffic mesh services in the autonomous territory of asking;
(2.2) the domain information service module is collected service status information among the domain node information service module c by the subscribing mode, with services selection, the distribution of support region job manager; Simultaneously, the domain information service module carries out data sync with the form of point-to-point (P2P) to the domain information service module in the adjacent autonomous territory and duplicates;
(3) workflow modeling
(3.1) workflow engine 1.2 provides a patterned editing interface assisted user to determine in the operation flow procedure relation (mating according to service function in service logic of using and the WSDL mesh services description document information that searched) between each service, carry out personalized abstract grid work flow model modeling, finish the foundation of one-to-one relationship between the service that searched and the activity in the workflow, and represent its concrete grid work flow with the form of BPEL (BPEL).
(3.2) workflow engine 1.2 obtains the concrete subjob carried out information (comprising autonomous territory, operation procedure place, operation physical address, the sequence of operation etc.) according to the corresponding relation of service and subtask among the BPEL, it is defined as the grid work that job description language (JDL) is described, and dynamically disposes the territory job manager.
(4) service assignment of operation
What the territory job manager was accepted workflow engine 1.2 transmissions is the grid work request JDL task description file of host-host protocol with Simple Object Access Protocol (SOAP), fulfils assignment to the distribution of service according to operation address, job title, operation and service binding information among the JDL.
(5) instantiation of service is disposed
(5.1) if define in the JDL task description file be the user submit to write application program certainly, the territory job manager is at first downloaded this program in user's space so, the resource packet holding tool that adopts Axis to provide carries out application service encapsulation (will be encapsulated into the Web service form with the application program of writing certainly of executable file EXE form compiling) to program, and according to the description of the WSDL in this Web service registration, installation, after service instantiation, again object code is signed, prevent to distort;
(5.2) if what define in the JDL task description file is the field of traffic service that system has developed, the territory job manager is directly located the position of this service and is carried out instantiation operation so.
(5.3) then, the territory job manager sends data transfer request to suitable data node resource service module d according to the demand information of data source in the job description; Simultaneously, performed service decomposition is become several calculation tasks parallel or serial, according to the service requirement and the specifying information of idling-resource (wherein service request information comprises the minimum cpu frequency of service, minimum CPU usage, the minimized storage space that service data source takies, intermediate data minimized storage space etc., idling-resource information comprises the cpu type of resource, the total frequency of the CPU of resource, the current utilance of the CPU of resource, the current storage size of utilizing of resource, the current memory size etc. that utilizes of resource), the computing node resource service module d that a selection part satisfies the minimum requirement of service participates in calculating, and calculation task is dispatched to the computing node resource service module d wait transfer of data that participates in calculating;
(6) visit of service data source
The autonomous territory of storage resources interior nodes has been disposed node service container b, the data transfer request that the acceptance region job manager sends, create back end resource service module d, data and file in the message reference memory nodes such as the Data Position that comprises according to request, transfer of data terminal point, and be transferred to the corresponding computing node of territory job manager appointment.Wherein data access is realized by the data access interface (OGSA-DAI) of Open Grid Services Architecture, comprises three class activities such as inquiry, expression conversion and transmission;
(7) transmission of service data source
Transmission service module f realizes the transfer of data between data source memory node resource service module d and computing node resource service module d, attribute according to definite transmission such as the actual loading of user's transmission demand information (as transmission minimum-rate, serious forgiveness etc.) and storage resources, network condition, replica management mechanism is set, and transfers data to the Practical Calculation node resource service module d of territory job manager distribution;
(8) Fu Wu execution and monitoring
(8.1) the appointment computing node service container b in the grid downloads code execution environments and registration is installed from the territory job manager, becomes the computing node resource service module d in the grid, and the transmission of waiting for desired data is to carry out calculation task;
(8.2) the mobile deployment agency of territory job manager establishment, be used to monitor distribution, execution and the last termination of the calculation task that belongs to same calculation procedure, it can be transferred to the client designated destination and carry out, also can change the position at any time, so that data are being provided to calculation task or when calculation task obtains data, are reducing data traffic the network as far as possible;
(8.3) each computational resource node resource service module d carries out calculation task according to the calculation procedure of disposing after obtaining calculated data, acts on behalf of condition report feedback and transfers to the territory job manager for user monitoring and management to moving to dispose at any time;
(9) service result storage
The territory job manager compiles result of calculation and transfers in the system database by transmission service module f, and the system service presentation layer adopts the C#.NET technology to write, and uses the stored data base of SQL Server database as final result.Collecting the result of calculation stage, move to dispose to act on behalf of and to duplicate or transfer to the SQL SERVER place main frame of depositing result data, control the result data stream that each resource provider sends effectively, and before the storage final result, again these data are handled.
(10) service result is showed
(ActiveX Data Objects ADO.NET) obtains the result of calculation that is stored in the SQL SERVER database, and the form by User represents to the user by the .NET data access interface for grid portal 1.1.
(11) autonomous territory management
In whole service instantiation process, domain manager all is responsible for memory node resource in the autonomous territory and computing node resource are managed, carry out the mutual of resource information with the node resource management services module e of lower floor, the withdrawing from and add of node resource in the management system is called SRM and managing computing resources.
In grid service instantiation process step (7), the present invention relates to the transfer of data of the transmission service module f of grid, its GridFTP transfer of data connects establishment and transmits constructive process as described in Figure 4:
(1) user at first logins transmission client by authentication, and transmission client reads user profile and sets up user data space;
(2) user submits data file transmission request in exclusive user data space, transmission client is by file logical name query node file index, find out the nodal directory that comprises this logical file, then shining upon the physical address of this logical file in the storage resources node by the nodal information catalogue returns to the user;
(3) user setup data transmission parameters, comprise the application for registration spatial cache, specify call back function, set cache size, set transmission mode, wire data type is set, be provided with server static/activity etc., set up the internodal transfer of data of transmission client and storage resources to reach an agreement on;
(4) data transport service of GridFTP startup self transmits data to destination node from source node.
It should be noted that safe mode with traditional normative document host-host protocol (FTP) different be the security mechanism that the GridFTP agreement comprises grid security architecture (GSI), therefore when connecting, need finish by the checking of legal certificate.Here we adopt function and interface that the commercial grid kit of Java (Java Commodity Grid Toolkit) is provided, generate and obtain letter of attorment, again it is packaged into and is stored the form that resource node is accepted, submit to the storage resources node verification.
In the step in grid service instantiation process (8), the present invention relates to the autonomous territory job scheduling of system.Autonomous territory job scheduling flow process in its job manager is as shown in Figure 5:
(1) in autonomous domain scheduling, job manager is submitted to operation in the job queue of local job scheduler, simultaneously the request of sending running job to client server;
(2) client server and job scheduler carry out periodical communication, after client server receives job request, and can be according to the demand notice job scheduler resource needed number and the resource type of operation;
(3) job scheduler and node actuator communicate, and survey each state of resources;
(4) job scheduler notice client server available resources are distributed desired resource by scheduling to operation simultaneously;
(5) client server with operation send to each general operation service (General RunningService, GRS) in, the execution of general operation service startup and monitoring task, and execution result returned to client server.
(6) general operation service is by the execution operation required data of data, services visit by the GridFTP transmission.
The present invention not only is confined to above-mentioned embodiment; persons skilled in the art are according to content disclosed by the invention; can adopt other multiple embodiment to implement the present invention; therefore; every employing project organization of the present invention and thinking; do some simple designs that change or change, all fall into the scope of protection of the invention.