JP4718963B2 - Data transmission system - Google Patents

Description

  The present invention relates to a data transmission system, and more particularly to a data transmission system for smoothly transferring data between a plurality of computers connected via a network.

  In a power transmission system, various nodes (servers, PCs, etc.) are connected to a network such as an IP router network, an MPLS network, and an Ethernet (registered trademark) network, and distributed processing is performed by these nodes. Yes. However, when distributed processing is executed in an environment in which various types of networks and nodes are mixed, various loads and failures are assumed. Therefore, at present, improvement of QoS (communication service quality) and high reliability for distributed processing are required.

As a technique for improving QoS and achieving high reliability for distributed processing, for example, a technique as disclosed in Patent Document 1 is known. In this patent document 1, the reliability of distributed processing is improved by performing communication between nodes such as delivery confirmation, retransmission of a request message to a processing execution node, or reacquisition of a response message to a processing request node. I try to plan.
JP 2005-56260 A

  However, the method of achieving high reliability of distributed processing by performing communication between the processing request node and the processing execution node is realized only on the premise that the transmission path functions normally. In particular, in Patent Document 1, since various messages are exchanged and delivery confirmation is performed between nodes using a single transmission path, if a transmission path becomes unusable due to some sort of malfunction in the transmission path. The problem that communication between nodes becomes impossible occurs.

  SUMMARY An advantage of some aspects of the invention is that it provides a data transmission system capable of increasing the reliability of data communication performed between a plurality of computers.

To achieve this object, the data transmission system according to the first aspect of the present invention performs data transmission / reception by packet communication with a plurality of nodes performing distributed processing connected to each other via a transmission path. In the data transmission system, a plurality of the transmission lines are provided between the nodes, an IP address corresponding to each of the transmission lines is provided for each node, and the requested processing is performed. The first IP address assigning means for assigning the representative IP address of the IP addresses provided in the receiving side node provided with the application to be executed to the data body unit to indicate the receiving side node, and the IP address correspond to each other Data duplicating means for duplicating the data based on an address table described in an attached state, and the data duplicating means Therefore, second IP address assigning means for assigning different IP addresses of the receiving node based on the address table to each header of the plurality of data obtained as a result of duplication of the data, A data transmission means for transmitting the data to the receiving node according to the IP address assigned by the second IP address assignment means, and a plurality of data transmitted by the data transmission means possessed by another node are received. A data receiving means and a later-arrival data discarding means for discarding later-arrived data when the plurality of identical data are received by the data receiving means are provided in each of the nodes, together from the received first-arrival data removing IP address of the header, the application of which nodes Generated or identification Shinano a data is to be restored to the original communication data by joining the order of brushes sequence number.

Therefore, each of a plurality of data obtained as a result of data duplication by the data duplicating means is transmitted to the receiving side node via different transmission paths according to the IP address added to each data.

As described above, according to the data transmission system of the first aspect, since the transmission side node transmits a plurality of the same data to the reception side node via different transmission paths, it is assumed that a plurality of transmission paths Even if a failure occurs in any one of the transmission paths, the data can reach the second node through another transmission path. Thereby, the reliability of data transmission can be increased.

  Hereinafter, the configuration of the present invention will be described in detail based on the best mode shown in the drawings.

1 to 6 show an embodiment of a data transmission system of the present invention. In the present embodiment, transmission / reception between a first node (transmission side node) and a second node (reception side node) will be described as an example. The data transmission system 1 of this embodiment exchanges communication data between a processing request node 4 functioning as a first node and a processing execution node 5 functioning as a second node that are connected to each other via a transmission path. Is to do. The transmission paths 2 and 3 are provided between the processing request node 4 and the processing execution node 5, and the processing execution node 5 is provided with an IP address corresponding to the transmission paths in a one-to-one manner. Based on the application 12 functioning as a first IP address assigning means for assigning the representative IP address of the IP addresses to the main part of the communication data, and the address table 27 described with the IP addresses associated with each other The transmission control unit 17 that functions as a data replication unit that replicates communication data, and based on the address table 27 for each header of a plurality of communication data obtained as a result of replication of communication data by the transmission control unit 17 A transmission control unit 17 functioning as second IP address assigning means for assigning different IP addresses to each other, and A transmission execution unit 19 that functions as a data transmission unit that transmits communication data to the process execution node 5 according to the IP address assigned by the transmission control unit 17 is provided in the process request node 4 and transmitted by the transmission execution unit 19. A reception execution unit 33 that functions as a data reception unit that receives a plurality of data, and a reception control that functions as a late arrival data discard unit that discards late arrival data when the reception execution unit 33 receives a plurality of identical data. The unit 31 is provided in the process execution node 5. As will be described later, the processing request node 4 also includes the reception execution unit 33 and the reception control unit 18 similar to the reception execution unit 33 that functions as the data reception unit and the reception control unit 31 that functions as the late arrival data discard unit. Is provided. The processing execution node 5 also has the application 12 functioning as the first IP address assigning unit, the transmission control unit 17 functioning as the data duplicating unit, the transmission control unit 17 functioning as the second IP address assigning unit, and data transmission. An application 13, a transmission control unit 30, and a transmission execution unit 32 similar to the transmission execution unit 19 functioning as means are provided.

  As shown in FIG. 1, the data transmission system 1 includes a process request node 4 and a process execution node 5 that are connected to each other via transmission paths 2 and 3. The processing request node 4 includes a CPU 6 and a hard disk 7. Further, the process execution node 5 includes a CPU 8 and a hard disk 9. Distributed processing is performed between these nodes.

  A plurality of types of applications are installed in the hard disks 7 and 9 in each of the process request node 4 and the process execution node 5. Input devices 10 and 11 are connected to the CPUs 6 and 8, respectively, and the above application is activated in response to signals input from the input devices 10 and 11.

  When performing distributed processing between the processing request node 4 and the processing execution node 5, for example, the application 12 operating on the CPU 6 issues a processing request, and the application 13 of the CPU 8 that receives the processing request receives the processing request node. The processing is executed on the communication data sent from 4 according to the content of the processing request. When the processing is completed, the application 13 transmits the processing result to the application 12. In the present embodiment, the case where distributed processing is performed by the application 12 of the processing request node 4 and the application 13 of the processing execution node 5 will be described as an example. Of course, it is possible to use a plurality of types of applications existing in the node 4 and a plurality of types of applications existing in the processing execution node 5. Even in such a case, the present invention is naturally applicable. Needless to say, the present invention can be applied to a case where three or more nodes are connected to each other via a network and distributed processing is performed using all of these nodes.

  Hereinafter, the configuration of the processing request node 4 will be described. The CPU 6 executes a program stored in a large-capacity storage unit such as the hard disk 7 to thereby execute a file system 14, an input / output interface 15, a parameter management unit 16, a transmission control unit 17, a reception control unit 18, a transmission execution unit 19, It functions as the reception execution unit 20 and the NFS server 21.

  The file system 14 is a function provided in the OS of the processing request node 4, reads predetermined data from the hard disk 7, and forms a parameter holding directory 22. The parameter holding directory 22 is a storage area for storing parameters set for the transmission control unit 17 and the reception control unit 18. The parameter holding directory 22 includes the transmission control unit 17 and the reception control unit. The parameters used by 18 are stored step by step according to the hierarchical structure of the directory. In this embodiment, as the “OS”, an OS that can use JAVA (registered trademark), such as Linux or Windows XP (registered trademark), can be used.

  For example, as shown in FIG. 2, the parameter holding directory 22 includes a “templateInfo” directory 23 indicating a template set for an application, a “nodeInfo” directory 24 indicating a node, and an “apInfo” directory 25 indicating an application. , And a system directory (not shown) indicating the entire system are arranged in parallel in the highest hierarchy, and the representative IP address of another node and the representative IP address of another node are assigned to the hierarchy one level below the “nodeInfo” directory 24. The plurality of node IP address directories shown are arranged in parallel, and the “1” directory indicated by the ID number of the plurality of templates set for the application is placed one level below the “templateInfo” directory 23. "2" directory and "3" directory A plurality of template ID directories are arranged in parallel, and a plurality of application priorities, such as a “100” directory and a “500” directory, in which the priority of the application is indicated numerically in the hierarchy one level below the “apInfo” directory 25 The directory is arranged in parallel, and in the hierarchy one level below the application priority directory, individual applications such as the “CountupAP” directory, “MessageAP” directory, and “TestAP” directory that indicate applications belonging to that priority by application name Directories are arranged in a tree structure. A parameter setting file “#properties” is provided in each of the system directory, node IP address directory, template ID directory, application priority directory, and application individual directory. Parameters set for the control units 17 and 30 and the reception control units 18 and 31 are described.

  Specifically, the system directory parameter setting file (hereinafter referred to as “system-wide parameter setting file”) has the same data communication behavior in the entire system, that is, in all nodes connected to the network. The parameters to do are described.

  The parameter setting file (hereinafter referred to as “node parameter setting file”) in each node IP address directory describes parameters set for each node. For example, a parameter setting file for a node belonging to the node IP address directory whose IP address is “192.168.7.11” describes parameters set for the processing request node 4.

  Parameters set for each template are described in the parameter setting file (hereinafter referred to as “template parameter setting file”) in each template ID directory. For example, in the template parameter setting file in the “1” directory, parameters set for the template with the ID number “1” are described.

  The parameter setting file in each application priority directory (hereinafter referred to as “application priority parameter setting file”) describes parameters set for applications belonging to the priority. For example, the parameter setting file for the application priority in the “100” directory describes parameters set for the applications with the priority “100” (“CountupAP” and “MessageAP”).

  A parameter setting file (hereinafter referred to as “application parameter setting file”) in each application individual directory describes parameters set for each application. For example, parameters set for an application named “MessageAP” are described in the application parameter setting file in the “MessageAP” directory.

  Note that the above-mentioned “plural templates set for an application” is set by the number of processing execution nodes when, for example, one application requests distributed processing to a plurality of processing execution nodes. Indicates a template. The “application priority” is given to each data when the application performs data transmission. The template parameter setting file, the application parameter setting file, the application priority parameter setting file, and the node parameter setting file may not be set arbitrarily, but the system-wide parameter setting file is an NFS server. By using a network sharing function such as 21 or NFS client 36, a common one is always set through all nodes connected via the network. In the present embodiment, the template parameter setting file includes a parameter indicating that data communication is performed using both the transmission paths 2 and 3 (hereinafter referred to as “two-route communication”). In the other parameter setting file, an example in which a parameter indicating that data communication is performed using only the transmission path 2 (hereinafter referred to as “one-route communication”) is described. I will explain. In addition, when the parameter indicating the above two-route communication and the parameter or communication data for giving the node a delivery confirmation function for confirming whether or not the communication data has been delivered are not delivered to the predetermined node, the communication data is retransmitted. The reliability of data communication and the guarantee of QoS can also be achieved by using a parameter for giving a retransmission function to a node together.

  The application 12 generates communication data used in the distributed processing based on a predetermined program and data stored in the hard disk 7 in response to the distributed processing start signal input from the input device 10 to the CPU 6. Further, a processing request and a representative IP address of the processing execution node 5 are assigned to the memory area of the main body of the communication data. In the present embodiment, the application 12 assigns the representative IP address of the processing execution node 5 to the memory area of the communication data main body. However, the present invention is not limited to this. The representative IP address may be assigned.

  The application 12 inputs communication data to the transmission control unit 17 via the input / output interface 15. The application 12 inputs a signal to the parameter management unit 16 when the distributed processing start signal is input. When the signal is input from the application 12, the parameter management unit 16 reads all the parameter setting files from the parameter holding directory 22 of the file system 14, and stores them in the memory 26 step by step according to the hierarchical structure of the directory 22. Hold on.

  The template parameter setting file, the application parameter setting file, the application priority parameter setting file, the node parameter setting file, and the system-wide parameter setting file are previously assigned ranks. In other words, ranks are assigned in advance to parameters belonging to the template unit, application unit, application priority unit, node unit, and system unit. Specifically, the template parameter setting file is ranked first, the application parameter setting file is ranked second, the application priority parameter setting file is ranked third, and the node parameter setting file is ranked. Is ranked fourth, the ranking of the parameter setting file for the entire system is ranked fifth, and a ranking table 17 a in which the relationship between the parameter setting file and the ranking is described is held in the memory area of the transmission control unit 17.

  When the communication data is input, the transmission control unit 17 reads the rank table 17a from the memory area, and ranks described in the rank table 17a and attributes of the communication data (for example, template ID, application name) The parameter setting file is designated from the parameter setting files held in the memory 26 based on the priority, transmission IP address, and reception IP address), and a parameter acquisition request signal is input to the parameter management unit 16.

  When the transmission control unit 17 designates a parameter setting file on the memory 26, first, the transmission control unit 17 determines whether or not a template parameter setting file corresponding to the template ID of the attribute of communication data exists in the memory 26. To identify. If it exists, specify the template parameter setting file. If it does not exist, it is determined whether or not an application parameter setting file corresponding to the application name of the attribute of the communication data exists in the memory 26. If it exists, specify the parameter setting file for the application. If it does not exist, it is identified whether or not an application priority parameter setting file corresponding to the priority of the attribute of the communication data exists in the memory 26. If it exists, specify the application priority parameter setting file. If it does not exist, it is determined whether or not a node parameter setting file corresponding to the transmission IP address and the reception IP address exists in the memory 26. If it exists, specify the parameter setting file for that node. If it does not exist, specify the parameter setting file for the entire system.

  As shown in the flowchart of FIG. 3, the parameter management unit 16 reads all the parameter setting files from the parameter holding directory 22 of the file system 14 (S2) when the application 12 is started (S1). The file is held in the memory 26 (S3). The parameter management unit 16 constantly monitors whether or not any request signal is input from the transmission control unit 17 (S4), and is in a standby state while no request signal is input (S4; No) (S5). When a request signal is input from the transmission control unit 17 (S4; Yes), the parameter management unit 16 identifies whether the request signal is a parameter acquisition request signal indicating a parameter acquisition request (S6). . Then, when the parameter management unit 16 identifies that the input request signal is a parameter acquisition request signal (S6; Yes), the parameter management unit 16 acquires a parameter from the parameter setting file designated by the transmission control unit based on the parameter acquisition request signal. It is extracted and set for the transmission control unit 17 and the reception control unit 18 (S7). When the parameter management unit 16 identifies that the input request signal is not a parameter acquisition request signal (S6; No), the parameter management unit 16 identifies whether the request signal is an update request signal instructing parameter updating. (S8). When the parameter management unit 16 identifies that the input request signal is an update request signal (S8; Yes), the parameter management unit 16 newly reads all the parameter setting files from the parameter holding directory 22 of the file system 14 and reads them. The stored file is held on the memory 26 (S2). When the parameter management unit 16 identifies that the input request signal is not an update request signal (S8; No), the parameter management unit 16 identifies whether the request signal is an end request signal for instructing the end of the parameter setting system 1 (S9). When the parameter management unit 16 identifies that the input request signal is an end request signal (S9; Yes), all the processing ends. If it is determined that the input request signal is not an end request signal (S9; No), it is continued to monitor whether or not any request signal is input from the transmission control unit 17 (S4).

  As shown in the flowchart of FIG. 4, when communication data is input (S101), the transmission control unit 17 first identifies whether or not a processing request is given to the communication data (S102). When the processing request is not given (S102; No), it enters a standby state (S103). If the processing request has been given (S102; Yes), it is next identified whether or not the communication is new (S103). If it is not new communication (S104; No), the transmission control unit 17 and the reception control unit 18 use existing parameters (S105). In the case of new communication (S104; Yes), the transmission control unit 17 confirms the presence of the parameter setting file with the first rank in the memory 26 (S106), and the parameter setting file with the first rank exists in the memory 26. If not (S107; No), the presence of the parameter setting file of the next order is confirmed (S108). When the parameter setting file exists (S107; Yes), the transmission control unit 17 and the reception control unit 18 acquire the parameters described in the parameter setting file as parameters used for communication (S109). The transmission control unit 17 and the reception control unit 18 execute processing based on the acquired parameters (S110). The transmission control unit 17 identifies the type of request input from the application 12, and updates the currently set parameter when the request input from the application 12 is an update request (S111; Yes). (S112). When the request input from the application 12 is not an update request (S111; No) and the request input from the application 12 is an end request (S113; Yes), all the processes are ended and the request is input from the application 12. If the request is not an end request (S113; No), it is identified again whether or not a processing request is given to the communication data (S102).

  The transmission control unit 17 holds an address table 27 in a memory area. In the address table 27, for example, the representative IP address set to the process execution node 5 and the sub IP address are described in association with each other. For example, the representative IP address corresponds to the transmission path 2, and the sub IP address corresponds to the transmission path 3. The transmission control unit 17 reads the address table 27 from the memory area when the parameter is set by the parameter management unit 16. Then, the communication data is duplicated based on this address table 27. As a result, communication data is generated as many as the number of transmission paths. That is, in this embodiment, two pieces of communication data are generated.

  The transmission control unit 17 packetizes two pieces of communication data and assigns a sequence number indicating the order of the packets to the packet header for each piece of communication data. Further, the representative IP address is assigned to the header of each packet of one communication data, and the sub IP address is assigned to the header of each packet of the other communication data.

The transmission execution unit 19 transmits the packet to which the representative IP address is assigned to the processing execution node 5 using the transmission path 2 according to the communication protocol 40, for example, TCP / IP, and the packet to which the sub IP address is assigned to the communication protocol. 40 to the processing execution node 5 using the transmission path 3.

The reception control unit 18 determines whether the packet received by the reception execution unit 20 is a late arrival packet. That is, the reception control unit 18 has a log file 28 that records the sequence number of the first-arrival packet, reads the sequence number in the header of the packet received by the reception execution unit 20, and the sequence number exists in the log file 28. Check whether or not. If it does not exist in the log file 28, it is determined that the received packet is the first packet that has not been received in the past, and the sequence number is written in the log file 28. As a result, when the same packet is received next time, it can be confirmed from the log file 28 that the packet is a late arrival. On the other hand, if the sequence number of the received packet is already in the log file 28, it is determined that the received packet is the same as the last received packet and discarded. Further, the reception control unit 18 removes the IP address from the packet header and connects the packets according to the sequence number to restore the communication data. Each packet includes the application names of the nodes 4 and 5 that perform transmission / reception, and the reception control unit 18 generates an application of which node (in this case, the application 13 because the process execution node 5 sends a reply). Each packet is connected and restored while identifying whether the data is correct. Then, the communication data is sent to the application 12 via the input / output interface 15. The application 12, the input / output interface 15, and the reception control unit 18 recognize the representative IP address of the processing request node 4, and in the communication data received by the reception execution unit 20, the representative IP address is included in the main body of the communication data. The processing is executed only for the communication data to which is assigned.

  The NFS server 21 is installed in the hard disk 7 as a part of the OS function of the processing request node 4 or as a program, and is read from the hard disk 7 and activated when the CPU 6 is activated. By using a communication function (not shown), the parameter holding directory 22 of the file system 14 is disclosed to the remote NFS client 36.

  Next, the configuration of the process execution node 5 will be described. The CPU 8 executes a program stored in a large-capacity storage unit such as the hard disk 9 to thereby execute an input / output interface 29, a transmission control unit 30, a reception control unit 31, a transmission execution unit 32, a reception execution unit 33, and a parameter management unit 34. , Function as a file system 35 and an NFS client 36. The CPU 6 and the CPU 8 are the input / output interface 29, the transmission control unit 30, the reception control unit 31, the transmission execution unit 32, the reception execution unit 33, the parameter management unit 34, and the file system 35. Since it has the same function as the transmission control unit 17, the reception control unit 18, the transmission execution unit 19, the reception execution unit 20, the parameter management unit 16, and the file system 14, the following description is repeated. Description is omitted, and only functions different from the functions described above will be described.

  The NFS client 36 is installed in the hard disk 9 as a part of the OS function of the process execution node 5 or as a program, and is read from the hard disk 9 and activated when the CPU 8 is activated. By communicating with the NFS server 21 of the processing request node 4, the parameter holding directory 22 disclosed by the NFS server 21 is mounted on the mount point of the file system 35.

  The parameter management unit 34 reads all the parameter setting files from the parameter holding directory 22 of the file system 35 when the signal is input from the application 13 and holds them in the memory 37. Then, the parameter management unit 34 sets parameters for the transmission control unit 30 and the reception control unit 31 in response to the signal input from the input device 11.

  The reception control unit 31 determines whether or not the packet received by the reception execution unit 33 is a late arrival using the log file 38, discards the late arrival and discards the first arrival packet. Data is restored and input to the application 13 via the input / output interface 29. The application 13, the input / output interface 29, and the reception control unit 31 recognize the representative IP address of the process execution node 5, and in the communication data received by the reception execution unit 31, the representative IP address is included in the main body of the communication data. The processing is executed only for the communication data to which is assigned.

  The application 13 executes processing according to the content of the processing request given to the communication data. When the processing is completed, communication data is generated, and the processing result and the representative IP address of the processing request node 4 are assigned to the memory area of the main body of the communication data. The application 13 inputs communication data to the transmission control unit 30 via the input / output interface 29.

  The transmission control unit 30 holds the address table 39 in the memory area. In the address table 39, for example, the representative IP address set to the processing request node 4 and the sub IP address are described in association with each other. The representative IP address corresponds to the transmission path 2, and the sub IP address corresponds to the transmission path 3. The transmission control unit 30 reads the address table 39 from the memory area when communication data is input from the application 13. The communication data is copied based on the address table 39. As a result, communication data is generated as many as the number of transmission paths. That is, in the present embodiment, two identical communication data are generated.

  The transmission control unit 30 packetizes two pieces of communication data and assigns a sequence number indicating the order of the packets to the packet header for each piece of communication data. Further, the representative IP address is assigned to the header of each packet of one communication data, and the sub IP address is assigned to the header of each packet of the other communication data.

  The transmission execution unit 32 transmits the packet with the representative IP address to the processing request node 4 using the transmission path 2 according to the communication protocol (for example, TCP / IP), and communicates the packet with the sub IP address. It transmits to the processing request node 4 using the transmission path 3 according to the protocol.

  Next, the flow of processing in the processing request node 4 will be described with reference to the flowchart of FIG. The application 12 generates communication data (S201). The application 12 gives a processing request to the memory area of communication data (S202). The transmission control unit 17 determines a parameter with reference to the attribute of the communication data (S203). If the parameter indicates two-route communication (S204; Yes), the communication data is duplicated (S205). Each communication data is packetized (S206). A sequence number indicating the order of the packets is assigned to the packet header for each communication data (S207). Based on the address table 27, the representative IP address of the processing execution node 5 is assigned to the header of each packet of one communication data, and the sub IP address of the processing execution node 5 is assigned to the header of each packet of the other communication data ( S208). On the other hand, if the parameter indicates one-route communication (S204; No), the communication data is packetized (S209). A sequence number indicating the packet order is assigned to the header of each packet (S210). The representative IP address of the processing execution node 5 is given to the header of each packet (S211). The transmission execution unit 19 performs data communication according to the representative IP address or sub IP address assigned to the header of each packet. Specifically, the packet with the representative IP address assigned to the header is sent to the processing execution node 5 via the transmission path 2, and the packet with the sub IP address assigned to the header passes through the transmission path 3. Is sent to the process execution node 5 (S212). In the reception control unit 18, the communication data returned from the process execution node 5 is not received by the reception execution unit 20 (S213; No), and a predetermined time limit has not elapsed (S214; No). A standby state is entered (S215). If the communication data returned from the process execution node 5 is not received by the reception execution unit 20 (S213; No), and a predetermined time limit has passed (S214; Yes), error processing is performed (S216). Terminate the process. On the other hand, when the communication data returned from the process execution node 5 is received by the reception execution unit 20 (S213; Yes), the reception control unit 18 logs whether the packet received by the reception execution unit 20 is a late arrival or not. If it is determined using the file 28 and it is a late arrival packet (S217; Yes), the packet is discarded (S218). If it is a first-arrival packet (S217; No), the IP address is removed from the header of the packet (S219). Then, the packets are connected according to the sequence number, and restored to the communication data generated by the application 13 of the process execution node 5 (S220). The reception control unit 18 inputs the communication data to the input / output interface 16 (S221). The input / output interface 16 extracts the processing result in the application 13 from the communication data (S222), and sends the processing result to the application 12 (S223).

  Next, the flow of processing in the processing execution node 5 will be described with reference to the flowchart of FIG. The reception control unit 31 receives the packet at the reception execution unit 33 (S301), determines whether the packet is a late arrival packet or not using the log file 38, and if it is a late arrival packet (S302; Yes) ), The packet is discarded (S303). If it is a first-arrival packet (S302; No), the IP address is removed from the header of the packet (S304). Then, the packets are connected according to the sequence number to restore the communication data generated by the application 12 of the processing request node 4 (S305). The reception control unit 31 inputs the communication data to the input / output interface 29 (S306). The input / output interface 29 extracts the processing request of the application 12 from the communication data (S307), and sends the processing request to the application 13 (S308). The application 13 executes processing according to the content of the processing request (S309). When the processing is completed (S310), the application 13 generates communication data (S311) and assigns the processing result to the memory area of the communication data (S312). The transmission control unit 30 determines the parameter with reference to the attribute of the communication data (S313). If the parameter indicates two-route communication (S314; Yes), the communication data is duplicated (S315). Each communication data is packetized (S316). A sequence number indicating the order of the packets is given to the packet header for each communication data (S317). Based on the address table, the representative IP address of the processing request node 4 is assigned to the header of each packet of one communication data, and the sub IP address of the processing request node 4 is assigned to the header of each packet of the other communication data (S318). ). On the other hand, when the parameter indicates one-route communication (S314; No), the communication data is packetized (S319). A sequence number indicating the order of the packets is given to the header of each packet (S320). The representative IP address of the processing request node 4 is assigned to the header of each packet (S321). The transmission execution unit 19 performs data communication according to the representative IP address or sub IP address assigned to the header of each packet. Specifically, a packet with a representative IP address assigned to the header is sent to the processing request node 4 via the transmission path 2, and a packet with a sub IP address assigned to the header passes through the transmission path 3. Is sent to the processing request node 4 (S322).

  As described above, according to the data transmission system 1 of the present invention, the processing request node 4 transmits a plurality of identical communication data to the processing execution node 5 via the transmission paths 2 and 3, so that transmission is temporarily performed. Even if a failure occurs in one of the transmission paths 2 and 3, the communication data can reach the processing execution node 5 through the other transmission path. As a result, the reliability of data transmission can be increased, and distributed processing using the processing request node 4 and the processing execution node 5 can be reliably executed.

  The above-described embodiment is an example of a preferred embodiment of the present invention, but is not limited thereto, and various modifications can be made without departing from the scope of the present invention. For example, in the above embodiment, the process request node 4 and the process execution node 5 are connected by the transmission lines 2 and 3, but the present invention is not limited to this, and the two nodes may be connected by three or more transmission lines. . In this case, IP addresses corresponding to the number of transmission paths are provided in the receiving side node.

  In the above-described embodiment, the data communication between the process request node 4 and the process execution node 5 has been described as an example. Even in this case, the present invention can be applied. In this case, each node is provided with an IP address different from each other by the number of transmission paths connected to the node, and an address table in which all destination IP addresses are described. For example, when the processing request node 4 and the node A (not shown) are connected by three transmission lines, and the processing request node 4 performs data communication not only with the processing execution node 5 but also with the node A, the node A Are provided with a representative IP address and two sub IP addresses, and an address table in which these IP addresses are described is provided in the processing request node 4. In addition to the representative IP address and sub IP address used for data communication with the processing execution node 5, the processing request node 4 is provided with a representative IP address used for data communication with the node A and two sub IP addresses. An address table in which these IP addresses are described is provided in the node A.

  An example in which the data transmission system of the present invention is applied to remote operation maintenance in a power transmission system will be described below.

  With the progress of electric power liberalization, the importance of cooperation (interconnectivity) between various electric power devices, applications, and even between electric power companies will increase. In order to realize such cooperation, standardization (unification) of information communication architecture (platform) is necessary. In addition, when many devices such as distributed power sources are connected to the grid, autonomous distributed control (see C. Rehtanz, Autonomous Systems and Intelligent Agents in Power System Control and Protection, Springer, 2003) may be effective. Therefore, in order to realize interoperability, distributed real-time network architecture, which is an information communication architecture that applies international standard information model of power equipment and general-purpose Internet technology, and mobile agent technology capable of autonomous distributed processing, etc. (DRNA: Distributed Real-time Computer Network Architecture) is being developed (Reference; "Development of Distributed Real-Time Network Architecture (DRNA) (Part 1)-Outline of Architecture Requirements and Functions-") R02013, March 2003). DRNA has a structure consisting of four elements as shown in Fig. 7, and each component has the following functions.

1. Application program and information model layer: Object model of power system equipment and application program used for supervisory control.
2. Advanced communication function layer: In general higher level protocols, it operates autonomously in response to changes in the situation and provides a basis for various communication forms. The advanced communication function includes “advanced communication function middleware” mainly responsible for the communication function and “agent platform” responsible for information processing by the agent. In this embodiment, advanced communication function middleware will be described.
3. Communication and communication functional layer: Basic communication and quality of service (QoS) control based on IP (Internet Protocol) technology in general lower protocol.
4). Communication management / security function: Supports the proper operation of each element of DRNA.

  In the current power communication network, a communication network is configured for each application of equipment operation / maintenance system such as system monitoring and power supply operation. However, in the future, from the viewpoint of efficient use of the communication network and cost reduction, It is desirable to integrate the data transmission of supervisory control systems (ie applications), that is, to build an integrated IP network (see; “Future prospects of power communication networks in the power liberalization era”) Research Report of Central Research Institute of Electric Power R02003, 2002 (December) (Reference: “Effects and Issues of Electricity Liberalization on Electric Power Communication Networks” Research Report of Central Research Institute of Electric Power R02002, January 2003). At this time, when using an IP network, QoS control is required to ensure communication quality. In DRNA, IP-level QoS control is performed using the transfer communication function, but once set, it is difficult to change frequently according to the application and communication network conditions. The advanced communication function middleware complements this, and the QoS guarantee function that adapts to the situation operates.

The advanced communication function middleware provides a combination of appropriate functional elements according to the requirements of the host application. For this reason, the application only needs to set basic parameters, and it is not necessary to set what function is used for QoS guarantee. Up to now, real-time communication using autonomous priority control for mobile agent and object-to-object communication has been realized at the sending and receiving terminals (see; Development of distributed real-time network architecture (DRNA) (Part 7)-Detailed design of advanced communication functions and Implementation-Power Central Research Institute report Research report R03007, March 2004). However, integrated IP network (see; Development of distributed real-time network architecture (DRNA) (Part 5)-Concept of application of IP router / MPLS / Wide area Ethernet in transmission and communication function-, Research report of R & D Center for Electric Power Research, R02010, 2003 (March), because multiple applications are mixed,
a) Interference between same priority applications
b) Occupancy of communication resources by a small number of applications
c) Measures against the load on the communication network were an issue. In addition, it was a challenge to meet the requirements specific to the following power system operation and maintenance system.
d) Response to higher reliability For this purpose, the following functions were added to the advanced communication function middleware (see Fig. 8). Note that TCP / IP is adopted as the communication method in FIG. 8, but it can also be executed by UDP / IP.
・ QoS guarantee function (additional): Correspondence to the above items a and b ・ Cooperation with transmission communication function: Corresponding to the above item c ・ High reliability function: Corresponding to the above item d

  Next, we will describe the QoS guarantee, real-time communication, and advanced communication function middleware required for integrated IP networks, related research and issues.

The integrated IP network has the following characteristics.
-Many applications are mixed: By mixing applications, various information flows occur, and the importance of controlling the communication of each application increases.
・ Large-scale: The backbone network, branch network, and regional network are interconnected, and various terminals (servers, PCs, etc.) are connected to increase the importance of network management including unified terminals. .
・ Heterogeneous technology mixed network: IP router network, MPLS network and Ethernet network are mixed, and unified QoS control is required.

As the integration of networks progresses and becomes more complex, various loads and failures are assumed, so the quality required by applications (QoS: communication service quality) and the requirement for high reliability are diversified.
With respect to QoS, so far, QoS is guaranteed (QoS guaranteed) for applications by performing QoS control (QoS control) only in the communication device. For example, DiffServ (Differentiate Services) (see K. Nichols, S. Blake, F. Baker and D. Black: “Definition of the Differentiated Services Field (DS Field) in the IPv4 and IPv6 Headers”, RFC 2474, 1998) ( Reference: QoS guarantee by "DiffServ delay characteristics and application to power communication network", Research Report R00026, April 2001). The following items can be listed as QoS.
・ Delay ・ Delay fluctuation (jitter)
-Bandwidth and loss rate As described above, QoS required by applications varies. In order for applications to communicate, a terminal such as a server and an IP network are required.
In QoS control, resources such as priority, line bandwidth, and buffer are assigned to communication that should be given priority. Focusing on the timing of resource allocation, QoS control methods can be classified into the following two types.
-Provisioning method: A method that assigns in advance, assuming traffic conditions in advance.
Signaling method: A method for allocating resources for each individual request from an application.

There are three main QoS control methods in the network protocol layer as follows.
ST2 (Internet Stream Protocol Version 2) (see L. Delgrossi and L. Berger (Editors): “Internet Stream Protocol Version 2 (ST2) Protocol Specification-Version ST2 +”, RFC1819, 1995 August): End done in signaling mode It is a protocol for performing real-time quality assurance with -to-End.
• IntServ (Integrated Service) (see; R. Braden, D. Clark and S. Shenker: “Integrated Services in the Internet Architecture: an Overview”, RFC1633, June 1994): A signaling method and QoS for IP protocol flows. It is excellent in that it can be controlled.
DiffServ: A method that can be realized by either a signaling method or a provisioning method. Compared with the above two methods, control is performed in units of flow sets.

  ST2 and IntServ are not widely used because all communication devices on the communication path need to be supported. DiffServ is inferior in that control is performed with a flow set compared to the other two methods, but it is easy to implement. At present, the provisioning method on the communication network side, the method of performing QoS control based on the DSCP value that is the priority of the IP level is the mainstream. In DiffServ, DSCP is specified at the edge node. By developing this and specifying the DSCP value in the application layer of the terminal and performing autonomous QoS guarantee according to the situation instead of the signaling method, it can be expected to realize high-quality communication on a packet basis.

As for the high reliability in the terminal, the following items can be listed as requests from the application.
• Two-route transmission • Continuous transmission • Re-transmission • Delivery confirmation Functions that satisfy these requirements must be able to be used in conjunction with QoS guarantee functions.

When implementing autonomous QoS guarantee, priority control is the most important. In developing advanced communication function middleware, we have developed our own priority control and confirmed its effectiveness. However, if there are too many occupied bands, priority control will be affected. The same thing happens with DiffServ. In DiffServ, research to introduce a subsystem called Bandwidth Broker has been conducted to compensate for this. Since this corresponds to the signaling method, in order not to impair the autonomy, a function of the provisioning method in the terminal instead of this is required. Its functions are shaping and policing, and there are studies on the network layer such as UDP and TCP as methods applicable to the IP network, but research on the application layer is still in progress. With regard to resource allocation, research on SUN Grid and globus in the grid field can be cited, but information processing technology is more important than communication technology.
In addition, research and development from the OS (Operating System) aspect is also underway, and there is a network layer for shaping / policing functions (see Raj Rajkumar, Kanaka Juvva, Anastasio Molano and Shui Oikawa, Resource Kernels: A Resource-Centric Approach to Real-Time Systems, In Proceedings of the SPIE / ACM Conference on Multimedia Computing and Networking, 1998.January (see Takashi Okumura, Daniel Mosse, Virtualizing Network I / O on End-Host Operating System: Operating System Support for Network Control and Resource Protection, Trans. On Computer, IEEE, Vol. 53, No. 10, pp. 1303-1316, 2004. October). Requires OS changes and extensions.
Since the control function in the network layer is often dependent on the OS, a system realized in the application layer is desirable for constructing a system by a multi-vendor required for the supervisory control system.

  In power system facility operation and maintenance communications, not only QoS but also high reliability is required. In fields other than electric power, there are few approaches to high reliability technology required in the electric power field. In the electric power field, there is RNA (Real-time computer Network Architecture) developed by the Central Electric Power Council. RNA provides functions of two-route transmission, delivery confirmation, retransmission, and late arrival discard. However, a priority control function related to QoS guarantee is not provided. Also, agent communication is not supported. Details will be described later.

In order to satisfy the requirements for QoS guarantee and high reliability in end-to-end communication, it is necessary to develop communication middleware installed in the terminal and to have the following functions on the terminal side.
・ Priority control (transmission / reception)
・ Shaping / policing / Aggregation / deployment / High reliability

When realizing these, the communication middleware needs to cooperate with the QoS control function of the communication device that constitutes the transmission communication network. For this purpose, the communication dollarware needs the following functions.
-Linking function with transfer communication function Priority control is for dealing with communication congestion and can be said to be a QoS guarantee function in a narrow sense. Shaping, policing, and aggregation are for controlling the communication timing of the amount of information related to transmission and reception to prevent interference between applications having the same priority, occupation of communication processing capability by a small number of applications, and the like. High reliability is to cope with communication path interruption and packet loss such as two-route transmission and retransmission. Cooperation with transmission communication is an important function for coordinating with QoS control of a communication device.

In a power system supervisory control system, it is important to guarantee communication delay in order to realize reliable supervisory control. Here, in this embodiment, communication in which a delay is guaranteed is referred to as “real-time communication”.
In DRNA, mobile agent technology capable of autonomous decentralized processing and distributed object technology as its basic technology can be applied, and the utilization of these technologies can be expected to enhance the efficiency and efficiency of monitoring and control work. Distributed object technology is an evolution of RPC. By using the distributed object technology, it is possible to communicate in the same format when exchanging information with other terminals in the information format that applications send and receive within their own terminals, and the efficiency of developing applications that perform distributed processing Can be achieved. Major methods applied in applications that perform distributed processing include CORBA (Common Object Request Broker Architecture) (see http://www.omg.org) and Java RMI (Remote Method Invocation) (http: // www .sun.com). CORBA and Java RMI are implemented using application layer protocols. However, these do not consider real-time communication. In the development of DRNA, we are examining middleware for advanced communication functions located in the application layer, aiming at real-time communication.

  The advanced communication function middleware is capable of not only streaming, which is a conventional monitoring information transmission method, but also inter-object communication and agent communication capable of transmitting advanced information such as atypical data. Agent communication realizes not only data but also a program that performs processing while moving the program itself to each monitoring control device, that is, autonomously performs processing. Thereby, the application can easily realize autonomous distributed control by utilizing agent communication. The advanced communication function middleware can use TCP, Java RMI, and UDP as lower protocols. In the advanced communication function middleware that has been implemented so far, the following two types of communication methods and “3 types of priority control and information sharing” are implemented as autonomous QoS guarantee functions.

-Communication method: Agent communication, communication between objects.
Autonomous QoS guarantee function: priority control (static priority processing, dynamic priority processing, process activation order control), information sharing.

  The processing procedure of inter-object communication when using priority control is as follows. However, the QoS guarantee request includes a processing time limit and an expected processing time on the receiving side.

1) The information that the application passes to the advanced communication function middleware when sending data is only the name of the application, the QoS guarantee request, and the data that is actually sent.
2) The advanced communication function middleware automatically sets the necessary functions according to the QoS guarantee request. At the time of transmission and reception, processing with higher priority or processing with a short time until the end deadline is preferentially processed (static priority processing, dynamic priority processing).
3) When receiving data, the advanced communication function middleware calls the application and delivers the data (process startup sequence control).

As a result, the processing time of the low-priority communication increases as the number of processes increases, but the high-priority communication for which the processing time limit is guaranteed can complete the processing in a substantially constant time, and the real-time property of the communication can be secured.
The priority control corresponds to agent communication, inter-object communication, and streaming communication. Real-time communication is guaranteed by controlling transmission / reception processing based on QoS guarantee requests.

FIG. 9 shows an overview of transmission priority control using agent communication as an example. The priority control for transmission includes static priority processing and dynamic priority processing. For example, there is a method in which static priority processing is applied to an agent for express processing and certain processing, and dynamic priority processing is used for an agent of best effort (BE) processing. The certainty type process always requires a process having real-time properties with a processing time limit, and the express type process always requires the highest priority process among the certainty type processes.
Static priority processing consists of a priority adjuster and a transmission subcomponent. When a plurality of data is simultaneously passed, the priority order controller refers to the processing deadline and adjusts the order of data passed to the transmission subcomponent. Next, the transmission subcomponent performs data transmission.

  The dynamic priority process includes a priority discriminator and a transmission subcomponent. A transmission subcomponent is installed for each priority. The transmission subcomponent itself does not perform scheduling. In the dynamic priority process, the sub-components are operated in parallel, and the operating time allocation of the sub-components is adjusted by checking the margin for the processing time limit of the data staying in the sub-components. This adjustment is performed by lengthening the operation time allocation of subcomponents with no margin, or vice versa. Static priority processing and dynamic priority processing are described in the document "Development of Distributed Real-Time Network Architecture (DRNA) (Part 7)-Detailed Design and Implementation of Advanced Communication Functions-, Research Report R03007, March 2004" ”And the document,“ Priority Processing Method for Mobile Agents for Distributed Control, Research Report of Central Research Institute of Electric Power Industry, R01010, April 2002 ”have been expanded to handle priority levels in order to support advanced communication function middleware. ing.

  FIG. 10 shows an outline of process activation order control, which is a reception priority control function, taking agent communication as an example. The process activation order control corresponds to agent communication, inter-object communication, and streaming communication. When data is passed to the process startup sequence control, the process startup sequence control uses the priority of QoS guarantee parameters, the processing deadline, and the expected processing time on the receiving side to balance the margin of processing time and priority of each data. Efficient scheduling that takes into account However, the express processing agent always starts with the highest priority. The process activation order control sequentially activates agents. The details of the process startup sequence control are based on the method described in the document “Proposal of dynamic scheduling function to guarantee real-time performance of distributed processing, Research Report R01021 of the Central Research Institute of Electric Power, April 2002”. An extension is being implemented to prevent a single agent from occupying the CPU of the terminal.

In the research on advanced communication function middleware, the overall structure (refer to: “Development of distributed real-time network architecture (DRNA) (Part 2)-Composition of advanced communication function and design of external interface specification-”), Central Research Institute of Electric Power Research Institute R02005, March 2003), priority control to realize real-time communication (see; "Study on adaptive protocol layer in power IP communication network", Central Research Institute of Electric Power Research R99006, April 2000) (see; Proposal of dynamic scheduling function to guarantee real-time performance of distributed processing ”, Research report R01021 of the Central Research Institute of Electric Power, April 2002) and information sharing (see“ Development of distributed real-time network architecture (DRNA) 3) -Investigation of application of information sharing method and cache in advanced communication function- ", Central Research Institute of Electric Power Research R02009, March 2003) There, has been promoting the implementation (see; the development of distributed real-time network architecture (DRNA) (Part 7) - the detailed design and implementation of advanced communication function -, Central Research Institute of Electric Power reported research report R03007, 3 May 2004). The study so far is aimed at the case where applications with different priorities communicate with each other on a 1: 1 terminal, and each communication is guaranteed in a situation where multiple applications are mixed in an integrated IP network. It was difficult.
As described above, end-to-end QoS guarantees can be made by adding functions such as shaping, a highly reliable function, and a function for linking with a communication device to the advanced communication function middleware. For that purpose, it is necessary to have the following characteristics for QoS guarantee function and high reliability based on the discussions so far.
・ Functions such as shaping perform autonomous control in the same way as priority control.
-High reliability has the functions provided by RNA and supports agent communication.
In order to realize cooperation with the communication device, the following processing is required.
-ToS connection and communication device can specify ToS value and port number. The ToS value contains DSCP (DiffServ Code Point) in the upper 3 bits.

The advanced communication function middleware has added functions such as shaping, a highly reliable function, and a link function with the transmission communication function. The additional function is located in the application layer as shown in FIG. 11 (the thick frame in the figure indicates the additional function), and performs communication control according to the data processing deadline and priority. As a result, the following functions are arranged in the application layer. In addition to the functions related to quality and high reliability, a remote processing call function is added. The remote processing call function can be used for connection with the agent platform.
・ Communication method (agent communication, inter-object communication, streaming communication)
・ QoS guarantee function ・ High reliability function ・ Linkage function with communication function ・ Remote processing call function (advanced communication RPC)
When an application communicates, a processing time limit is given to the application layer together with data, and each function performs the following processing, thereby realizing communication with guaranteed end-to-end QoS and high reliability requirements. .
-Perform priority control, shaping, and aggregation suitable for real-time processing such as transmission of monitoring information.
・ Perform high-reliability transmission peculiar to power, such as 2-route transmission or single-route transmission.
-Specify the QoS control parameters for linking with the QoS control function of the communication device.
-Transmit streaming, objects, and agents passed from the application layer. In the prototype, TCP, UDP, and Java RMI can be used as lower protocols.

In order for the communication device constituting the integrated IP network and the advanced communication function middleware on the terminal to cooperate, the following items must be realized.
-Support for mixed heterogeneous technology networks: QoS control can be performed without depending on technologies such as MPLS and Ethernet.
・ No need to modify the communication device: The linkage function can be realized with the terminal function.
-Unified linkage control: Control information of each terminal can be shared, and communication devices do not need to correspond to individual terminals.

  This linkage function focuses on the IP header that can be referenced by each device and uses the ToS value in the IP header. By specifying the ToS value at the transmission terminal, it is not necessary to modify the communication device, and it is not necessary to set the communication device each time an application is added. In order to enable detailed QoS control, in addition to the ToS value, the linkage function specifies the port number of the transport layer. Thereby, for example, cooperation with SSH which is a security function is possible.

  FIG. 12 shows an outline of communication processing using the cooperation function. The linkage function holds a conversion table of priority and port number / ToS value. On the transmission side, transmission processing is performed by specifying the ToS and port number based on the conversion table. The receiving side can receive from the port number described in the conversion table.

FIG. 13 shows the outline of QoS control parameter setting processing of the communication device when the substation server and the control station server communicate via the IP network in which the wide area Ethernet network and the MPLS network are integrated. Indicates.
FIG. 6A shows a case where there is no linkage function, and the first-stage communication device of each network sets priorities to the VLAN-tag and MPLS label of the QoS control parameter Ethernet based on the IP address. The last-stage device removes the VLAN-tag and MPLS labels. The communication device can be controlled only in units of IP addresses and needs to be set.
Figure (b) shows the case where there is a linkage function. By specifying from the sending terminal, each network can set the Ethernet VLAN-tag and MPLS label based on this ToS value in each network. The QoS specified by the transmitting terminal can be provided to the application. In this case, each communication device only needs to set QoS control for each ToS value, and it is not necessary to add or change settings for each terminal or application added.

  The QoS guarantee function (additional) controls communication timing in the application layer to prevent interference of applications with the same priority and occupation of communication resources.

Shaping suppresses the burst arrival of packets to the transmission destination by providing restrictions on data transmission. Advanced communication function middleware level window control can be realized.
In shaping, a unique marking method is performed using the processing deadline, and shaping and disposal are performed based on the result. It can be discarded using the data processing time limit and the allowable overtime of the control parameter.
Shaping can be controlled in units of application, destination (node name + application name), and source application priority (specific priority).

Policing prevents application overload by limiting the number of data received per unit time during reception. Further, in combination with the process activation order control, the real-time guarantee rate by the process activation order control can be improved.
Control is possible in units of receiving application, transmission source (node name + application name), and priority (specific priority) of the transmission source application.

When there is a lot of data and the overhead of the transmission process is large, the aggregation function aggregates and transmits the data in order to reduce this overhead.
As a control method, aggregated data is transmitted when the following conditions are satisfied. The aggregation upper limit number and unit time are set for each control unit.
-When the maximum number of aggregations has been reached-When the unit time has elapsed Control is possible in units of sending application, destination node name (representative IP address), and priority. The priority at each level is used as a control unit.

  The expansion function is arranged on the reception side, and when the data aggregated by the aggregation function is received on the transmission side, it is expanded and restored to the original data. Control parameters are not required, and are sequentially expanded in the order received.

The delivery confirmation function confirms that the transmission data has passed to the receiving terminal while maintaining the order and continuity. The receiving side sends back a delivery confirmation when receiving data. If the delivery confirmation does not return to the transmitting terminal for a certain period of time, the transmission is performed again.
Each packet is given a delivery confirmation sequence number for identification.

The retransmission function transmits the same data again when it cannot be confirmed that the transmission data has passed to the receiving terminal. No request for delivery confirmation is made from the sending terminal side. Each packet is given a retransmission sequence number for identification.
It is possible to specify the waiting time of the receiving terminal until a retransmission request is made and the number of retransmission requests.

The continuous transmission function duplicates the same data for one path and transmits it continuously between two terminals. A multiplexed data sequence number is assigned to each packet for identification.
You can specify the number of data to be replicated and the data transmission interval.

The two-route transmission function can select one of the following data transmission methods when there are two physical communication paths between two terminals.
-Establish two connections with the receiving terminal and send the same data to both paths at the same timing. A multiplexed data sequence number is assigned to each packet for identification.
-Establish only one connection with the receiving terminal and send data to that path. When the connection is disconnected, the other route is selected.
-Establish only one connection with the receiving terminal and send data to that path.
A conceptual diagram of the two-route data transmission method is shown in FIG. In the figure, “AP” is an abbreviation for “Application”, and “NIC” is an abbreviation for “Network Interface Card”.

  The late arrival discard function allows the receiving side terminal to pass the previously received data to the application among the received plurality of the same data if the transmitting side terminal assigns the multiplexed data sequence number to the data. Discard the arrival data. For identification of each packet, a multiplexed data sequence number and transmission / reception application names are used.

The case where the receiving terminal receives a plurality of the same data is as follows.
・ When data is transmitted to 2 routes using the 2 route transmission function ・ When multiple data is transmitted using the continuous transmission function

  Development of advanced communication function middleware prototypes for additional functions (see; Development of distributed real-time network architecture (DRNA) (Part 7)-Detailed design and implementation of advanced communication functions-Research report R03007, 2004 The result of mounting and evaluation in March) is shown below.

FIG. 15 shows the configuration of the evaluation system. Substation premises monitoring and control server (substation server), control station monitoring and control server (hereinafter referred to as control station server), and three PCs that simulate equipment maintenance servers as load generation sources were connected via an IP network. The specifications of the three PCs are the same, and the specifications are as follows.
CPU: Pentium (registered trademark) 4 2.8GHz
・ Memory: 1GB
・ Network interface card: 10/100 / 1000BASE-TX x 2 (connected to IP network with 100BASE-TX)

The configuration of the IP network was a two-line network using two CISCO Ethernet switches 2950. Physically, the control station server and the substation server are connected to both systems, and the equipment maintenance server is connected to one system (primary).
Various servers are equipped with middleware for advanced communication functions. This prototype is software developed by our company. Developed using Java language. The OS used was Linux. Details of each software are as follows.
-Java: SUN Java Standard Edition 1.4.2 (registered trademark)
-OS: Novel SuSE Linux 9.1 (kernel version is 2.6) (registered trademark)

  The evaluation system simulates communication using the following three types of applications.

-Monitoring and control application (A)
・ Maintenance (business support application) (B)
・ Maintenance (ITV application) (C)

  The application was assumed to have QoS requirements for delay, bandwidth and availability as shown in Table 1. Based on this, Table 2 and Table 3 show the results of determining properties and profiles. Based on delay requirements, bandwidth allocation, and high reliability requirements, the parameters were determined so that there would be a difference in preferential treatment. Based on this setting, real-time communication can be executed by designating the processing time limit requested by each application during communication. Table 1 shows application requests, Table 2 shows profile and property setting results, and Table 3 shows profiles for cooperation with communication devices.

  In the performance evaluation, the three types of applications described above performed periodic transmission of monitoring information from the substation to the control station. For the amount of information to be transmitted, the definition of the circuit breaker object included in the information model of IEC61850 was referred to, and this was used as data for 128 volumes. In the advanced communication function middleware prototype, a part of the information model of IEC61850 is implemented on the database. Table 4 shows data held by the circuit breaker object in this implementation. The total data size is 145 bytes, and when this is multiplied by 128, the result is 145 × 128 = 18560 bytes≈20,000 bytes, so the amount of data to be transmitted is 20,000 bytes. Table 4 shows an assumed configuration of the circuit breaker object.

  An experiment was conducted to measure the processing overhead of the proposed method. Table 5 shows basic communication characteristics (processing overhead). When using the priority control, shaping, policing, and high-reliability functions shown in Table 5 (a), using only the priority control function, and using an application with three types of settings when neither function is used The processing time from the start of transmission to the end of reception was measured. Inter-object communication using TCP was used as the communication method. As a result, Table 5 (b) was obtained. The ratio of processing time when the QoS guarantee function and the high reliability are applied (application A and application B) for communication between objects is 8.1: 12.2 [ms] ≈2: 3. This difference is the processing time required for the QoS guarantee and the high reliability function, and can be said to be small compared to the processing time of the inter-object communication itself. Therefore, it was confirmed that the QoS guarantee and the high reliability function can be applied to communication requiring a delay guarantee of about 1 second.

The result of evaluating the effect of QoS guarantee by advanced communication function middleware is described. When each of the two types of applications A and B was communicating in the direction from the substation to the control station shown in Table 6 (a), application C caused a load on one side (primary) of the communication network. In order to clarify the difference between applications A and B, we used communication between objects using TCP as the communication method. As a result, Table 6 (showing evaluation of QoS guarantee and high reliability) was obtained. The application C itself, which is a load on the communication network, continues to transmit data of 100 Mbps or more while the other two types are communicating, and as a result, the processing time increases and the processing time limit is exceeded. Although the delay of the application B having the same priority increases due to the influence of the application C, the influence on A having a high priority is small. In application B, the delay increased compared to the normal time, but the processing deadline was not exceeded.
For linkage with communication devices, as shown in Table 3, according to the priority of the application, the linkage function determines the port number of the control station server and the upper 3 bits of DSCP (DiffServ Code Point: ToS value) as specified in the profile. ) Specify a value and confirm that you are communicating.
Application A applies the two-route transmission function of the advanced communication function middleware and the late arrival discard to improve availability. It was confirmed that application A can continue communication even if a failure occurs in the transmission line of one system, and it can satisfy availability requirements.
From the above results, we confirmed the effectiveness of QoS guarantee and high reliability functions.

  Table 7 shows a comparison result between the advanced communication function middleware and RNA (Real-time computer Network Architecture). The advanced communication function middleware supports not only the high-reliability functions provided by RNA, but also priority control necessary for IP network-specific communication delay countermeasures, cooperation with communication devices, agent communication, and communication between objects. It is excellent in that it is. RNA has been introduced into surveillance control applications. On the other hand, the prototype of advanced communication function middleware was developed for the purpose of verifying advanced communication function middleware. It seems to be applied.

  Next, we will show how to use the advanced communication function middleware according to the QoS and high reliability requirements of various applications. Thereby, various functions provided by the advanced communication function middleware can be efficiently used.

The functions to be applied differ depending on the QoS guarantee and high reliability type required by the application. The types of requests and the functions to be applied are as follows.
・ Delay: priority control ・ Delay fluctuation: priority control and shaping, policing ・ Bandwidth: shaping, policing, aggregation / expansion ・ Loss rate: shaping, policing, aggregation / expansion, 2-route transmission, continuous transmission, retransmission ・ High reliability : 2-route transmission, continuous transmission, re-transmission, delivery confirmation When QoS of the transmission path is required, use the link function with the communication device.
The parameters given to the advanced communication function middleware are a profile, a property, and a communication parameter. Profiles and properties need to be determined in advance. However, it is necessary not to be an application user but to be set by the operation manager of the entire system through the network management / security function. The QoS guarantee parameter needs to be specified by the application at the time of communication request.
Profile: Specify the priority assigned to the application and the function that the application should use.
-Property: Specify the behavior for each function.
-QoS guarantee parameters: communication processing time limit, processing time on the receiving side

Each function can be flexibly combined according to the degree of congestion and quality of the communication network and the terminal. The combination is specified using a profile. FIG. 16 shows the order of use of various functions assumed. However, the same function cannot be used twice. Some functions are not used depending on the situation. Table 8 shows a list of properties. The advanced communication function middleware adopts the property as a parameter based on the following property specification priority. As shown in FIG. 17, the higher the priority, the higher the priority.
-Template property: Priority 5
-Application properties: Priority 4
-Application priority property: Priority 3
-Node property: Priority 2

The shaping, policing and aggregation / deployment functions can be controlled by grouping multiple applications, such as application priority units and node units, in the case of the above priority 3 and 2, and these can be easily controlled for all applications. Functions can be applied.
In the high reliability function, it is possible to set the retransmission interval and the number of retransmissions for each application, and more detailed settings can be made compared to the TCP controlled on a node basis.

  In order to guarantee the QoS required by the application, the following procedure is required (see FIG. 18).

1. Set application priority.
2. If there is a QoS requirement for the transmission path, compare it with the QoS control setting by the communication device. If it is necessary to add settings, add the priority of advanced communication function middleware and the assignment of ToS value and port number. For communication device settings, refer to the document "Development of Distributed Real-Time Network Architecture (DRNA) (Part 5)-IP router / MPLS / Wide-area Ethernet application concept for transmission and communication functions", Research Report R02010 , March 2003 ".
3. Select the required functions, determine the order so that communication bottlenecks do not occur during operation, and specify them in the profile.
4. Specify properties that specify the behavior of various functions.
5. During communication, the application specifies the delay that is the QoS guarantee parameter to the advanced communication function middleware.

The detailed procedure of each function is as follows.
(1) Delay Use priority control to prevent the effects of lower priority than the own application. The profile specifies the priority assigned to the application and the use of the priority control function. Application priority is set according to the type of application. Apply the priority control function to all applications. In parallel with this, the application specifies a processing time limit in the QoS guarantee parameter.
(2) Jitter Shaping and policing are used to cycle the transmission timing and reception timing and suppress fluctuations in application units.
In order to suppress the influence of an application having a low priority, it can be dealt with by using shaping, policing, and priority control for each low priority group.
(3) Bandwidth The number of data exchanged per unit time is controlled as a bandwidth unit using shaping, policing, and aggregation / development. Set properties based on the bandwidth required by the application.
(4) Loss rate Use the resend function when there is enough processing time to allow resending. In addition, TCP or Java RMI can be used as a lower protocol of the communication method, but the number of retransmissions can be specified only for each terminal.
(5) High reliability In response to high reliability requests, specify whether to apply high reliability functions and the functions to be used. Furthermore, the transmission path configuration (2-route (1 + 1), 2-route (1: 1), 1-route configuration) is selected according to the configuration of the IP network.
Two routes (1: 1) can also be realized by the route switching function provided by the transmission communication function.
(6) QoS of transmission line
In the profile, a communication quality class is designated for each application priority according to the required quality. Specify the transmission / reception port number and ToS value as the communication quality class. As a result, the communication apparatus performs priority control, shaping, policing, and path switching according to the contents set in advance in the communication apparatus based on these values.

  As described above, in this embodiment, the advanced communication function middleware that guarantees the QoS and the high reliability requirement in the application layer and the usage method thereof are shown. The advanced communication function middleware mainly includes QoS guarantee, high reliability, and linkage with transfer communication function. As a result, it is possible for the sending terminal, communication network, and receiving terminal to perform cooperative QoS guarantee and high reliability for message communication, agent communication, and streaming communication. It can satisfy the QoS guarantee and high reliability requirement of communication.

It is a functional block diagram which shows the principal part of the parameter setting system of this invention. It is a figure which shows the structure of the directory for parameter holding. It is a flowchart which shows the flow of a process in a parameter management part. It is a flowchart which shows the flow of a process in a communication control part. It is a flowchart which shows the flow of a process in a process request node. It is a flowchart which shows the flow of a process in a process execution node. It is a functional block diagram of the electric power system monitoring control system based on DRNA. It is a figure which shows the structure of an advanced communication function. It is a figure which shows the flow of the priority process for the QoS guarantee of a transmission side. It is a figure which shows the flow of the priority process for QoS guarantee on the receiving side. It is a figure which shows the structure of advanced communication function middleware. It is a figure which shows the outline | summary of the cooperation function with a transmission communication function. It is a figure which shows the behavior of a transmission communication function, (a) is a figure which shows the behavior of the transmission communication function when there is no cooperation function, (b) is a figure which shows the behavior of the transmission communication function when there is a cooperation function is there. It is a conceptual diagram of 2 route transmission. It is a figure which shows the structure of the simulation system used for evaluation. It is a figure which shows the use order of various functions. It is a figure which shows the hierarchical relationship of a property. It is a flowchart which shows the setting procedure of a control parameter.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Data transmission system 2, 3 Transmission path 4 Process request node 5 Process execution node 6, 8 CPU
7 Hard disk 12, 13 Application 16, 34 Parameter management unit 17, 30 Transmission control unit 17a Order table 18, 31 Reception control unit 19, 32 Transmission execution unit 20, 33 Reception execution unit 14 File system 22 Parameter holding directory 26 Memory 27 39 Address table

Claims (1)

In a data transmission system in which data exchange is performed by packet communication between a plurality of nodes performing distributed processing connected to each other via a transmission path, a plurality of the transmission paths are provided between the nodes, and the transmission paths For each node, an IP address corresponding to the one-to-one correspondence is provided, and the representative IP address among the IP addresses provided in the receiving node provided with the application for executing the requested processing is received. First IP address assigning means for assigning to the data main body to indicate a side node; data duplicating means for duplicating the data based on an address table described in a state where the IP addresses are associated with each other; For each header of the plurality of data obtained as a result of duplication of the data by the data duplicating means Second IP address assigning means for assigning different IP addresses of the receiving node based on the address table, and the data according to the IP address assigned by the second IP address assigning means A data transmitting means for transmitting to a node; a data receiving means for receiving a plurality of data transmitted by the data transmitting means possessed by another node; and when the plurality of identical data are received by the data receiving means Each of the nodes is provided with a later-arrival data discarding unit that discards the data of the received data, and the later-arrival data discarding unit removes the IP address of the header from the received first-arrival data, and which node has generated the application. whether the data identification Shinano by joining in the order of brush sequence number based on the communication de Data transmission system, characterized in that to restore the data.

Images