JP2007272540A - Data distributing method and data distributing system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To mutually transfer data configuring a large amount of data between a plurality of nodes, and to effectively distribute a large amount of data by dynamically changing a transfer route. <P>SOLUTION: A data distributing method distributes data divided into a plurality of pieces of data. In the data distributing method, a route node device 1 transmits distribution information to terminal node devices 3. The terminal node device 3 requests the distribution to the route node device 1 after deciding the divided data to be first distributed to the route node device 1 based on the distribution information. The terminal node device 3 decides the other terminal node devices 3 to distribute the other divided data based on the distribution information, and the distribution is requested to the other terminal node devices 3. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a data distribution method and a data distribution system, and in particular, a data distribution method and a data distribution system for efficiently distributing a large amount of data by cooperative operation of a plurality of nodes in a network system in which a plurality of nodes are connected. About.

  Basic software (operating system: OS) and its application software used in a computer are regularly updated and distributed to users from the viewpoint of improving functions, correcting defects, security measures, and the like. That is, such software is packaged so that it can be easily installed and set in a computer, released to a distribution server on a network, and downloaded by a user. Examples of such packaged software include image files for CD-ROM and DVD. For this reason, the file capacity is extremely large.

  In various business operations, manuals, instructions, commercial videos (so-called content) using moving image are often used. New videos are also created and distributed regularly. Examples of such a moving image include a manual moving image that expresses an operation procedure, a moving image that expresses a detailed portion, a manual that uses high-definition images, and a commercial that requires appealing power. This file size is extremely large.

As described above, there is a great demand for regularly distributing a large-capacity file from a distribution server to many bases (nodes, that is, network computers), and a method for efficiently performing this is desired. For example, it is known that a user who has obtained certain information using a user's profile distributes the information to users with similar profiles (see Patent Document 1). Also, for example, it is known that a client that acquires and stores content from a content distribution server distributes the content to other clients in accordance with destination designation information added to the content (Patent Document 2). reference).
JP 2002-298007 A JP 2003-345692 A

  As described above, when a large file is distributed from a distribution server to a large number of nodes, there are the following problems.

  That is, the distribution server sets an upper limit on the number of nodes that can be connected simultaneously. Therefore, a node that has made a download request after the download request has exceeded the upper limit waits for execution of its own download until any of the nodes that are currently downloading ends the download. Therefore, it is necessary to increase download facilities and network bandwidth in proportion to the increase in the number of users.

  Further, if the execution speeds of all networks are both 10 Mbps for both upstream (transmission direction) and downstream (reception direction), and the upper limit (number of simultaneous connections) of the distribution server is “3”, then 1 The downlink bandwidth per node is 3.3 (= 10/3) Mbps. This is not efficient when viewed from a node connected in a 10 Mbps band because only a part of the node is used. In addition, the upstream bandwidth of the node is not utilized for distributing large files.

  On the other hand, as described above, when transferring information between users or distributing contents between clients, there are the following problems. That is, when transferring information between users, a user profile must be used, and when distributing content between clients, destination designation information added to the content must be used. These profiles and destination designation information are fixed and cannot be changed dynamically. For this reason, the route of information or content distribution cannot be changed depending on which node is actually distributed with which information or content, or depending on the situation of the network.

  The present invention relates to a data distribution method for efficiently distributing large-capacity data by transferring the data constituting the large-capacity data among a plurality of nodes and dynamically changing the transfer route. The purpose is to provide.

  In addition, the present invention transfers data constituting a large amount of data between a plurality of nodes, and dynamically changes the transfer route, thereby efficiently distributing large amounts of data. An object is to provide a distribution system.

  The data distribution method of the present invention comprises a root node device that is a source of data distribution, a plurality of terminal node devices to which the data is distributed, and a network that connects them, and divides the data into a plurality of divisions This is a data distribution method in a data distribution apparatus that divides and distributes data. In the data distribution method of the present invention, the root node device includes at least a node ID assigned to the terminal node device, a total number of the plurality of terminal node devices, and a total number of divided data obtained by dividing the data. Distribution information is transmitted to each of the plurality of end node devices. Each of the plurality of terminal node devices determines division data that the terminal node device first requests distribution to the root node device based on the distribution information, and distributes the divided data to the root node device. Request. The root node device distributes the divided data to each of the plurality of terminal node devices in response to the distribution data distribution request. Each of the plurality of terminal node devices determines, based on the distribution information, which of the other terminal node devices requests distribution of divided data other than the divided data distributed from the root node device; The distribution of the divided data is requested to the determined other end node device.

  Preferably, in an embodiment of the data distribution method of the present invention, when the root node device distributes the data, prior to this, the distribution information is sent to each of the plurality of end node devices. Send.

  Preferably, in one embodiment of the data distribution method of the present invention, each of the plurality of terminal node devices measures a distribution speed at which divided data is distributed to each of the other terminal node devices, and the distribution speed Based on the above, the determined other end node device is changed to the end node device having a high distribution speed, and the changed end node device is requested to distribute the divided data.

  Preferably, in one embodiment of the data distribution method of the present invention, the root node device sets a distribution time, which is a time when the terminal node device starts a request for distribution of the data, together with the distribution information. And each of the plurality of terminal node devices starts a request for distribution of the data to the root node device after the distribution time elapses.

  Preferably, in one embodiment of the data distribution method of the present invention, the root node device calculates a hash value for each of the divided data obtained by dividing the data, and the calculated hash value is used as the distribution information. Is transmitted to each of the plurality of terminal node devices, and each of the plurality of terminal node devices calculates a hash value for the divided data received from the root node device, and the hash value and the distribution information Are compared with the hash values of the received divided data included in the received data, and if they match, the received divided data is managed as correct data.

  The data distribution system of the present invention comprises a root node device that is a source of data distribution, a plurality of end node devices to which the data is distributed, and a network that connects them, and divides the data into a plurality of divisions Distribute into data. In the data distribution system, the root node device includes distribution information including at least a node ID assigned to the terminal node device, a total number of the plurality of terminal node devices, and a total number of divided data obtained by dividing the data. , A means for transmitting to each of the plurality of terminal node devices, and distributing the divided data to each of the plurality of terminal node devices in response to a request for distribution of the divided data from each of the plurality of terminal node devices. Means. Further, each of the plurality of terminal node devices determines divided data that the terminal node device first requests the root node device to distribute based on the distribution information, and distributes the divided data to the root node Means for requesting the apparatus are provided.

  According to the data distribution method and data distribution system of the present invention, the terminal node device requests the root node device to distribute the divided data that first requests the root node device based on the distribution information, and further distributes the data. Based on the information, with respect to the divided data other than the divided data distributed from the root node device, the distribution is requested to other end node devices.

  As a result, the download requests are not concentrated on the root node device but distributed to the plurality of terminal node devices, and it is possible to use almost the entire upstream and downstream bands assigned to each of the plurality of terminal node devices. It can be used effectively for distributing large files. Therefore, the divided data can be efficiently distributed from the other end node devices to the end node device. On the other hand, there is no need to use destination designation information added to a user profile or content. Therefore, the distribution route can be changed according to which divided data is distributed to which end node device, and the total number of the plurality of end node devices and the total number of divided data which are elements of the distribution information can be changed. The distribution route of the divided data can be changed according to the change. Therefore, an optimal distribution route can be selected without intervention of the system administrator in accordance with both the scale of the data distribution system and the size (capacity) of data to be distributed. In addition to the above, the root node device does not need to set an upper limit on the number of nodes that can be connected simultaneously. Therefore, even a terminal node device that has made a download request later than the other terminal node devices does not need to wait for its own download execution until the downloading of the other terminal node devices is completed.

  According to an embodiment of the present invention, distribution information is transmitted to the end node device when data is distributed. As a result, the end node device can obtain the distribution information when the route node device desires distribution, and can dynamically change the data distribution route based on the distribution information. A route can be selected.

  Further, according to one embodiment of the present invention, based on the distribution speed at which the divided data is distributed, the terminal node device having a high distribution speed is requested to distribute the divided data. As a result, the terminal node device can dynamically change the data distribution route according to the network status, and as a result, can select the optimum distribution route.

  Also, according to one embodiment of the present invention, the distribution time is transmitted together with the distribution information, and the end node device initiates a data distribution request to the root node device after the distribution time has elapsed. Thereby, after the download request is distributed to the root node device and the terminal node device, the download request to the root node device can be further distributed in time.

  Further, according to one embodiment of the present invention, the end node device determines whether the received divided data is correct by comparing the hash value of the received divided data with the hash value received in advance. Accordingly, it is possible to improve security in data distribution by utilizing the distribution of large-capacity data divided into divided data.

  FIG. 1 is a configuration diagram of a data distribution system of the present invention, and FIG. 2 is an explanatory diagram of a root node device of the present invention.

  The data distribution system according to the present invention includes a root node device 1, a plurality of end node devices 3, and a network 2 connecting them. The network 2 is composed of the Internet, for example. That is, the network 2 is constructed by connecting many LANs (Local Area Networks) and WANs (Wide Area Networks) to each other, and the physical topology cannot be clearly understood.

  The root node device 1 is a computer or a distribution server that distributes, that is, transmits (that is, a distribution source) a large-capacity file 16 that is data, and includes a terminal node management unit 11, a file distribution unit 12, and a data management unit 13. Consists of. Each of the terminal node devices 3 is a computer that distributes, that is, receives (that is, becomes a distribution destination) the large-capacity file 16, and includes a status report unit 31, a file distribution request unit 32, and a file distribution unit 33. The In order to distinguish a plurality of terminal node devices 3 from each other, a code such as A or B is added to distinguish them. The same applies to the status reporting means 31 and the like of the terminal node device 3.

  The data distribution system of the present invention distributes data by dividing it into a plurality of distribution unit data (divided data). Specifically, distribution is performed after a single large-capacity file 16 is divided into a plurality of small-capacity files (hereinafter referred to as divided files) 161. Accordingly, the large capacity file 16 is one piece of data (software) to be distributed, and the divided file 161 is a unit of distribution (download or transfer). The size of the divided file 161 is determined in advance. In the middle of distribution of the large capacity file 16, a plurality of divided files 161 are distributed on the network 2 (a plurality of terminal node devices 3 thereof). The large-capacity file 16 includes, for example, a program such as an OS, content such as video, and the like, but may be any file that can be divided into a plurality of transfer units, and is not particularly limited.

  In the data distribution system of the present invention, the root node device 1 distributes the divided file 161 constituting the large capacity file 16 to the terminal node device 3. At the same time, the terminal node device 3 that has already received the divided file 161 distributes the received divided file 161 to the other terminal node device 3. Thus, many end node devices 3 finally distribute all the divided files 161, that is, the large capacity files 16, to all the end node devices 3 while cooperating with each other.

  When desiring to distribute the large-capacity file 16, in the terminal node device 3, the status report means 31 or the file distribution request means 32 accesses the terminal node management means 11 of the root node device 1 in advance and receives information indicating that it wants to receive it. Access information. In this example, the access information includes, for example, its IP address and port number. Each terminal node device 3 includes access information to the root node device 1 in advance.

  In the root node device 1, the end node managing means 11 receives the access information from the plurality of end node devices 3 prior to the delivery of the large capacity file 16, as shown in FIG. A node ID table 14 is created to manage the end node device 3. The large-capacity file 16 is distributed to (only) the terminal node device 3 whose access information is registered in the node ID table 14. FIG. 3A shows an example of the node ID table 14 of the present invention. The node ID table 14 stores an IP address and a port number (that is, the access information) for each node ID. The node ID is unique in the data distribution system, and is assigned to each terminal node device 3 by the terminal node management unit 11. That is, the terminal node management unit 11 issues node IDs to the terminal node device 3 in the order in which the access information is received.

  The terminal node management means 11 sorts the created node ID table 14 in ascending order of IP addresses (numbers thereof) periodically or sequentially at predetermined time intervals as shown in FIG. 3B. . If the IP addresses are the same, they are sorted in ascending order of port numbers (numbers of numbers). In this sorting process, node IDs are assigned again in ascending order. Thus, the node ID table 14 is sorted by IP address and port number when used (distributed).

  In this example, the terminal node management unit 11 receives the access information from all the terminal node devices 3 and when the final sorting of the node ID table 14 is completed, the terminal node device 3 (the status report unit 31 or the file distribution) Each of the requesting means 32) is notified of the finally assigned node ID. That is, the end node management unit 11 refers to the node ID table 14 to obtain the IP address of the end node device 3, further obtains the corresponding node ID, and notifies the IP address of the node ID. To do.

  For example, assume that a final node ID table 14 is obtained as shown in FIG. In this case, the terminal node device 3 has 1000 nodes in total. When the IP address of the terminal node device 3 is “10.8.2.5”, the node ID is “00125”. That is, the terminal node device 3 having the IP address “10.8.2.5” has the node ID “00125” for distribution of the large capacity file 16. This node ID is valid only for distribution of the large-capacity file 16.

  In the root node device 1, a large capacity file 16 to be distributed is registered in advance in the data management means 13 in accordance with a user instruction input. As shown in FIG. 2, the data management unit 13 divides the registered large-capacity file 16 into a plurality of divided files (also referred to as samples) 161 in advance of the distribution. The size of the divided file 161 is, for example, a predetermined data capacity. The data management means 13 adds a file name to each divided file 161 divided. The file name includes at least a divided file number and is unique in the data distribution system. The divided file numbers are numbers that are consecutive in the order of the divided files 161 below, with the number of the first divided file 161 in the original large-capacity file 16 being “1”. Note that the division of the large-capacity file 16 may be performed at any time before distribution.

  As shown in FIG. 2, the data management unit 13 generates a file name table 17 for the large capacity file 16 (that is, all the divided files 161). For this purpose, the data management means 13 obtains a hash value for each of the divided files 161. FIG. 4A shows an example of the file name table 17 of the present invention. The file name table 17 stores the hash value for each file name. The large-capacity file 16 and the corresponding file name table 17 constitute registration data 15.

  In FIG. 4A, for example, it is assumed that a large-capacity file 16 has a file name “sample.dat” and a capacity of 1 gigabyte. The data management means 13 divides this into 1000 (n = 1000) divided files 161 for each megabyte, obtains a hash value for each divided file 161, and generates a file name table 17. The file name of the divided file 161 is, for example, “sample # 1”. Each of the terminal node devices 3 includes a file name “sample.dat” of the large capacity file 16 and a value of the number of files n in advance.

  The data management unit 13 (or the file distribution unit 12, the same applies hereinafter) generates a list 18 corresponding to each terminal node device 3 based on the node ID table 14 and the file name table 17. This generation process is executed, for example, when the creation of the node ID tables 14 for all the terminal node devices 3 is finished. The list 18 includes at least the node ID assigned to the terminal node device 3, the total number of the plurality of terminal node devices 3, and the total number of divided files 161 obtained by dividing the large capacity file 16. The root node device 1 transmits the list 18 to each of the plurality of end node devices 3.

  FIG. 5 shows an example of the list 18 of the present invention. The list 18 includes a node information part 181 for storing node information, a file name table part 182 corresponding to the file name table 17, and a node ID table part 183 corresponding to the node ID table 14. The data management means 13 refers to the node ID table 14 and extracts the maximum node ID number “01000” and generates a node information portion 181 based on this and the node ID “00125” of the terminal node device 3. To do. Further, the data management means 13 acquires the file name table 17 and generates a file name table portion 182 based on the file name table 17. Further, the data management means 13 refers to the node ID table 14 to obtain access information (node ID, IP address, port number) of a plurality of adjacent end node devices 3 and generates a node ID table portion 183. . That is, the node ID table portion 183 is generated using “IP address” such as “next node ID” and “next node ID” of the “node ID”. The node ID table portion 183 may be composed of the node ID table 14.

  The data management means 13 returns the generated list 18 to the end node device 3 as a response to the file request. At this time, in this example, the data management means 13 notifies the distribution time together with the list 18. The distribution time is a time that determines how long the terminal node device 3 starts to acquire the divided file 161 after the list 18 is acquired (received). For example, the distribution time is determined as a relative time after receiving the list 18, but may be determined as an absolute time.

  In the terminal node device 3, the file distribution request unit 32 receives the list 18 from the root node device 1, and based on this, which is the divided file 161 that it should obtain first from the root node device 1. Determine whether. That is, the file number of the divided file 161 to be acquired first from the root node device 1 is calculated. The number (number of acquisitions) of the divided files 161 to be acquired is one or more. For this calculation process, the file distribution request means 32 (node) assigned to the end node device 3 in the list 18 (hereinafter referred to as own node ID) and the total number of divided files 161 (hereinafter referred to as file). Number) and the total number of terminal node devices 3 (hereinafter referred to as the number of nodes). Based on the calculation result, each of the plurality of end node devices 3 requests the root node device 1 to distribute the divided file 161.

  First, when the number of nodes is equal to the number of files, the file distribution request unit 32 sets the divided file 161 including the divided file number equal to the own node ID in the file name as the divided file 161 to be acquired first. Specifically, as shown in FIG. 12B, the end node device 3 uses the node number “01000” of the maximum node ID and the file name table portion 182 in the list 18 shown in FIG. The maximum value “01000” of “xxxxx” of the name “sample.xxxxx” is obtained and compared. As a result of the comparison, if both are equal, the divided file 161 having the file name “sample.00125” including the same divided file number as the own node ID in the list 18 is set as a distribution request target. Note that the list 18 illustrated in FIG. 5A illustrates an example of the list 18 when the number of nodes is equal to the number of files.

  Secondly, the file distribution request means 32 determines that the divided file number determined by (own node ID) / ((number of nodes) / (number of files)) is the file name when the number of nodes is larger than the number of files as a result of the comparison. The divided file 161 included in is a divided file 161 to be acquired first. Specifically, as shown in FIG. 12A, the end node device 3 obtains a natural number obtained by rounding up the decimal point of the value of (own node ID) / ((number of nodes) / (number of divided files)). The file name including the same divided file number is obtained. For example, in the list 18 (different from FIG. 5A), the node number of the maximum node ID is “01000”, the maximum value of “xxxxx” of the file name “sample.xxxxx” is “00250”, and the own node ID is Assuming that the division file number is “00125”, the division file 161 with the file name “sample.00032” is the target of the distribution request from 125 / (1000/250) = 32.

  Third, if the number of nodes is smaller than the number of files as a result of the comparison, the file delivery requesting unit 32 first (number of files) ÷ (number of nodes) to obtain (the number of divided files 161 to be obtained) Acquisition number), and then the offset is obtained by (number of files) ÷ (number of acquisitions). Further, the division file 161 whose division file number is determined based on these is first obtained. 161.

  Specifically, as shown in FIG. 12C, a natural number obtained by rounding up the decimal point of the calculation result value of (number of files) / (number of nodes) is calculated from its own node ID, and the same number as those is obtained. Find a numbered file name. For example, as described above, the node number of the maximum node ID is “00250”, the maximum value of “xxxxx” of the file name “sample.xxxxx” is “01000”, and the divided file number of the own node ID is “00125”. Then, the number of acquisitions is 1000/250 = 4, and the offset is 1000/4 = 250. Accordingly, with the same division file number as that of the own node ID as an initial value (reference), the number of offsets is incremented by the number of acquisitions. That is, the division file 161 having the file names “sample.00125”, “sample.00375”, “sample.00625”, and “sample.00875” is targeted for distribution. When the number of files is exceeded, the file is returned to the first sample.00001 and the divided files 161 corresponding to the number of acquisitions are targeted for distribution.

  The file distribution means 12 of the root node device 1 that has received the distribution request, for example, the divided file 161 having the file name “sample.00125” (hereinafter, this is also referred to as the divided file sample.00125, and the other divided files 161 are also the same). Is distributed to the terminal node device 3. The terminal node device 3 that has received the divided file sample.00125 obtains the hash value of the file, obtains the hash value of the file name “sample.00125” from the file name table portion 182 of the list 18, and compares the two. .

  As a result of this comparison, if both are the same, the end node device 3 notifies its own data management means (not shown), holds the divided file sample.00125 as correct data, and other end node devices. 3 is managed on the assumption that the divided file 161 is already possessed. This data management means has the same function as the data management means 13 of the root node device 1. That is, the end node device 3 creates a file name table (not shown) similar to that of the root node device 1 based on each of the received divided files 161, and thereby manages the (correctly) received divided files 161. To do.

  Both the root node device 1 and the end node device 3 encrypt the communication route itself or the communication data in response from the root node device 1 to the end node device 3, and maintain the confidentiality of the communication. Specifically, in this example, in order to maintain data accuracy by comparing hash values, transmission of the list 18 from the root node device 1 is encrypted using, for example, SSL. Thereby, the confidentiality of this communication can be maintained.

  When the confidentiality of one side can be maintained in this way, in the transmission / reception of the divided file 161, neither the root node device 1 nor the end node device 3 performs encryption by SSL or the like. That is, the transmission / reception of the list 18 is performed by encrypted communication, and the transmission / reception of the divided file 161 is performed by communication without encryption (encryption is omitted). By omitting the encryption, it is possible to avoid an increase in cost for the encryption process. Even in this case, both the root node device 1 and the terminal node device 3 may perform encryption such as SSL on the file receiving side.

  When each of the plurality of terminal node devices 3 performs the above processing, all the “divided files 161” existing in the root node device 1 during the distribution of the large capacity file 16 (divided file 161) The plurality of terminal node devices 3 exist in a distributed manner. An example of this state is shown in FIG.

  FIG. 6 is an explanatory diagram of data distribution processing according to the present invention. The example of FIG. 6 corresponds to the list 18 shown in FIG. That is, FIG. 6 shows an example of distribution of the divided file 161 when the number of nodes is equal to the number of files. Specifically, in this case, the end node device 3 sends the first divided data 161 to the root node device 1. The distribution of the request is requested, and the distribution state is indicated.

  In FIG. 6, the divided file 161 having the file name “sample #i” is distributed and held in the terminal node device 3A. Here, the node ID of the terminal node device 3A is “00125”, and the file name “sample #i” is “sample.00125”. Similarly, the node IDs of the end node devices 3B, 3C, and 3D are “00126”, “00127”, and “00128”, respectively, and the divided files “sample.00126”, “sample.00127”, and “sample.00128”, respectively. "Is distributed and held.

  In this example, as described above, the distribution time is transmitted together with the list 18. Therefore, all the terminal node devices 3 do not request the root node device 1 to distribute the divided file 161 at the same time. For example, when 1000 terminal node devices 3 exist, 50 terminal node devices 3 each make a distribution request to the root node device 1 according to the distribution time set by shifting in time. As a result, even if the bandwidth of the network 2 of the root node device 1 is small or the number of terminal node devices 3 that the root node device 1 can simultaneously deliver is limited, the distribution to the root node device 1 is possible. It is possible to prevent requests from being concentrated.

  In the data distribution system of this example, paying attention to the continuity of IP addresses, this is used. That is, an IP address is often assigned to one organization as a whole by an organization that manages the IP address. For this reason, IP addresses having the same high-order bits often have a close logical network topology. In other words, the other end node device 3 having the IP address next to its own IP address exists in the vicinity of the end node device 3 (for example, in the same organization), and other divided files 161 (continuous) There is a high possibility that the divided file 161) to be held is held. Therefore, as shown in FIG. 7, the end node device 3 distributes the other divided file 161 not from the root node device 1 but from another end node device 3. FIG. 7 is an explanatory diagram of data distribution processing according to the present invention, and corresponds to FIG.

  In FIG. 7, the end node device 3 sequentially acquires other divided files 161 using a plurality of pieces of access information (such as IP addresses) such as “next node” and “next node” in the list 18. For example, the terminal node device 3 (3A) that has received the divided file sample.00125 with the same number as its own node ID “00125” from the root node device 1 has the node ID “00126” (next node ID) from the list 18. The IP address of the terminal node device 3 (3B) is obtained. This is an IP address whose number is adjacent to the IP address of its own node. Accordingly, the end node device 3A requests the end node device 3B to distribute the divided file sample.00126. At this time, the terminal node device 3B having the node ID “00126” has already acquired the divided file sample.00126, and distributes it to the terminal node device 3A.

  The terminal node device 3A with the node ID “00125” measures the average download time while receiving the divided file sample.00126. Thereafter, as described above, the end node device 3A performs the check process using the hash value, registers the divided file sample.00126 in its own data management means (not shown), and has acquired this. The average download time is registered, and the divided file sample.00126 is managed.

  Next, in the same manner as described above, the end node device 3A having the node ID “00125” obtains the IP address of the end node device 3C having the “node ID”, that is, the node ID “00127” from the list 18, In response to the request, distribution of the divided file sample.00127 next to the divided file sample.00126 is requested. At this time, the terminal node device 3C with the node ID “00127” has already acquired the divided file sample.00127.

  Therefore, the terminal node device 3A having the node ID “00125” measures the average download time while receiving the divided file sample.00127. Thereafter, as described above, the end node device 3A performs the check process using the hash value, registers the divided file sample.00127 in its own data management means (not shown), and “acquired”. The average download time is registered, and the divided file sample.00127 is managed.

  The terminal node device 3 sequentially acquires (distributes) the plurality of divided files 161 by an arbitrary number of times (specified number of times) specified in advance. The specified number of times can be determined empirically, and is, for example, about 1/10 of the number of files. The terminal node device 3 measures and manages the average download time in each distribution. For example, as shown in FIG. 5B, a download time portion 184 is added to the list 18 (the other portions 182 and 183 are not shown).

  As a result, the end node device 3 determines, based on the list 18, which other end node device 3 is connected in a good line state of the network 2 in order to obtain the divided file 161. Further, when receiving the distribution of the other divided file 161 from the same end node device 3, the end node device 3 refers to the previous average download time, and changes the average download time to a certain condition set in advance. When applicable, or intentionally even if there is no significant change in the average download time, the distribution request destination is changed to another terminal device 3 every several distributions. Thereby, it is possible to prevent distribution requests from being concentrated on a specific terminal node device 3.

  The other end node device 3 (3A) to which the divided file 161 is distributed from the end node device 3 (3B) is notified to the end node device 3B of the average download time (or an average download based on this) as a completion notification. Speed). The end node device 3B that has received this average download time manages the average download time of the divided file 161 as described above, and compares it with that of other end node devices 3 such as “next node” and “next node”. As described above, when a certain condition is met (corresponding) (for example, when it is slower than a predetermined speed), the other divided file 161 is acquired from the root node device 1 instead of the end node device 3. .

  Further, the end node device 3A having the node ID “00125”, for example, downloads the average download time from the end node device 3B having the node ID “00126” and the average download from the end node device 3C having the node ID “00127”. The time is compared (assuming that downloads from both sides are made in advance), and the next and subsequent divided file 161 is requested to be delivered to the faster end node device 3. For example, as shown by “x” in FIG. 7, when the average download time from the end node device 3C is slower than a predetermined condition (preliminarily determined empirically) than that from the end node device 3B, the end node device 3A As shown by a dotted line in FIG. 7, the destination of subsequent distribution is changed to the terminal node device 3B, and distribution file 161 is received from now on. Even if the average download time between the two is not more than a predetermined condition, the distribution request destination is alternately changed every several times.

  In this case, if the IP addresses of the end node devices 3 that are slower than the predetermined condition are continuous, all the end node devices 3 and subsequent nodes having adjacent node IDs are affected by the late end node device 3. As a result, there is a possibility that delivery of the divided file 161 to the whole will be delayed. Therefore, in this example, the terminal node device 3 issues a distribution request to the root node device 1 having all the divided files 161 when the IP addresses of the terminal node devices 3 that are slower than a predetermined condition are continuous. .

  For this reason, when the terminal node device 3 issues a distribution request for the divided file 161 to the other terminal node device 3, the terminal node device 3 notifies the average download time when the node receives the distribution from the terminal node device 3. The terminal node device 3 is connected to the average download time of the own node received from the other terminal node device 3 and other terminal node devices 3 having a relationship of “next node ID” and “next node ID” to the own node ID. If the “next node ID” and “average node download time of the terminal node device 3 that is next node ID” are slower than a predetermined condition compared with the average download time, a distribution request is made to the root node device 1. .

  As described above, it is also conceivable that there is a difference in the degree of distribution of the divided file 161 among the plurality of terminal node devices 3. Therefore, in this example, each terminal node device 3 generates an acquired list (not shown) that includes the acquired status of the divided file 161 and the distribution band of the current node at that time, and uses this as a route. By notifying the terminal node management means 11 of the node device 1, a status report is periodically made every predetermined time.

  When the terminal node device 3 sequentially acquires the split file 161 from the other terminal node device 3, the current node upload status (distribution band) of the own node acquired as a completion notification from the other terminal node device 3, and the split file 161 Is periodically notified to the root node device 1 at predetermined time intervals. The root node device 1 receives the notification, the acquisition of the split file 161 is completed, and the access information of the end node device 3 that has a margin in the current upload status is notified from the end node device 3 Is returned as a response. The terminal node device 3 changes the request for distribution of the split file 161 to the root node device 1 to the notified request for distribution to the terminal node device 3. As a result, concentration of distribution requests to the root node device 1 can be prevented.

  When there is a terminal node device 3 that has acquired all the divided files 161 in the status report from the terminal node device 3, the root node device 1 receives the status report after that point. In response, the node ID and IP address of the terminal node device 3 that has been completed and has a small distribution band are returned as a response. The terminal node device 3 that has received this response uses the node ID and IP address in the response to issue a distribution file 161 distribution request to the terminal node device 3 instead of the root node device 1. As a result, concentration of distribution requests to the root node device 1 can be prevented. In practice, the means for changing the various delivery request destinations described above are executed randomly (ie, flexibly or appropriately) in combination.

  When the terminal node device 3 receives all the divided files 161, the terminal node device 3 integrates these files and reconfigures them into a large-capacity file 16 that was originally targeted for distribution. Thereafter, the terminal node device 3 continuously notifies the root node device 1 of the status, and the distribution of the divided file 161 is distributed to the other terminal node devices 3 that have not yet acquired all the divided files 161. If a request is received, it is delivered. When the root node device 1 receives the status notification from the end node device 3 and all the end node devices 3 have completed the file acquisition, based on this, all of the plurality of end node devices 3 Send a completion response. When each of the plurality of terminal node devices 3 receives the completion notification, each of the terminal node devices 3 determines that the distribution of the large capacity file 16 to all the terminal node devices 3 is completed, and stops the status notification to the root node device 1. .

  Hereinafter, a data distribution method in the data distribution system of the present invention will be described with reference to the flowcharts shown in FIGS.

  The terminal node device 3 that desires to distribute the large-capacity file 16 registers its access information in the root node device 1. In response to this, the root node device 1 registers and manages the terminal node device 3 as shown in FIG. FIG. 8 is an end node management processing flow of the present invention.

  In FIG. 8, the terminal node management means 11 of the root node device 1 is in a state of waiting for receiving the information of the terminal node device (terminal node) 3 (step S11), and newly receives the delivery of the large file 16. When the terminal device 3 to be transmitted transmits its own access information consisting of “IP address” and “port number”, it receives it and adds the access information to the node ID table 14 (step S12). The node ID table 14 is sorted (step S13).

  On the other hand, as shown in FIG. 9, after registering the large capacity file 16 in the data management means 13, the root node device 1 completes registration of access information from all the terminal node devices 3, and creates the node ID table 14. When is finished, a list 18 is created. FIG. 9 is a data management processing flow of the present invention.

  In FIG. 9, the data management means 13 of the root node device 1 reads the large file 16 in accordance with the user's instruction input (step S21), divides it into a plurality of divided files 161 (step S22), and the large file A file name table 17 for the file 16 is created (step S23). Thereby, registration data 15 is generated. Thereafter, the data management unit 13 obtains the number of files of the divided file 161 (step S24), determines the delivery time of the large-capacity file 16 according to the user's instruction input (step S25), and the end node management unit. The node ID table 14 is acquired from 11 (step S26), the list 18 is created based on these (step S27), and this is distributed to all the terminal node devices 3 via the network 2 (step S28).

  As described above, in the process of generating the list 18, the root node device 1 calculates a hash value for each of the divided files 161 obtained by dividing the large-capacity file 16 and uses this as a part of the list 18. Further, as described above, the root node device 1 determines a delivery time which is a time when the end node device 3 starts a request for delivery of the large capacity file 16. Therefore, the hash value and the distribution time are transmitted to each of the plurality of terminal node devices 3 together with the list 18.

  Furthermore, as shown in FIG. 10, the root node device 1 transmits root node information (own access information) to each of the plurality of terminal node devices 3. FIG. 10 is an initial registration process flow of the status report of the present invention.

  In FIG. 10, when the status report means 31 of the terminal node device 3 receives the IP address and port number from the root node device 1, it registers it in its own memory area (not shown) (step S31). .

  The plurality of end node devices 3 that have received the list 18 request the root node device 1 to distribute the large capacity file 16. For this purpose, as shown in FIG. 11 and FIG. 12, each of the plurality of end node devices 3 has divided files 161 that the end node device 3 first requests distribution to the root node device 1 based on the list 18. Then, it requests the root node device 1 to distribute the divided file 161. At this time, each of the plurality of terminal node devices 3 starts the request after the elapse of the distribution time. FIG. 11 is a flowchart of the file distribution request process according to the present invention, and FIG. 12 is an explanatory diagram of the file distribution request process according to the present invention.

  In FIG. 11, the file distribution request means 32 of the terminal node device 3 checks whether or not the notified distribution time is reached (step S41), and if it is not the distribution time, repeats step S41. If it is the distribution time, the file distribution request means 32 compares the number of nodes with the number of files (step S42), and based on this, determines the number and file numbers of the divided files 161 to be acquired as follows. To do.

  That is, when the number of nodes is larger than the number of files, the file distribution request unit 32 sets the file number to be acquired as (own node ID) / ((number of nodes) / (number of files)) as shown in FIG. (Step S43), based on this, the root node device 1 is requested to distribute the divided file 161 having the file number (Step S48). In step S43, the decimal point is moved up (the same in step S45).

  If the number of nodes is equal to the number of files, the file distribution request means 32 sets the file number to be acquired to the same number as its own node ID as shown in FIG. 12B (step S44), and then executes step S48. To do.

  When the number of nodes is smaller than the number of files, the file distribution request means 32 calculates the number of files to be acquired (number of acquisitions) by (number of files) / (number of nodes) as shown in FIG. Step S45) Based on this, the offset of the file to be acquired (that is, the file number interval) is calculated by (number of files) ÷ (number of acquisition) (step S46), and the divided file to be acquired based on the above 161 is determined (step S47). That is, the file number having the same number as the own node ID is acquired, and further, the file number obtained by adding an offset to the own node ID until the number of acquisition is reached. Thereafter, step S48 is executed.

  In response to the request for distributing the divided file 161 from each of the plurality of terminal node devices 3 as described above, the root node device 1 distributes the requested divided file 161 to each of the plurality of terminal node devices 3.

  Each of the plurality of end node devices 3 calculates the hash value of the divided file 161 received from the root node device 1, and compares the hash value with the hash value of the received divided file 161 included in the list 18 If the two match, the received divided file 161 is managed as correct data.

  Thereafter, as shown in FIGS. 13A and 13B, each of the plurality of terminal node devices 3 distributes the divided file 161 that requests distribution and the distribution of the divided file 161 based on the list 18. To determine the other end node device 3 that requests. FIG. 13A is a flowchart of the file distribution request process of the present invention, and FIG. 13B is an explanatory diagram of the file distribution request process of the present invention.

  In FIG. 13A, the file distribution request means 32 of the terminal node device 3 compares the number of nodes with the number of files (step S51), and if the number of nodes is equal to or larger than the number of files, further acquisition is performed. It is checked whether the file number of the already divided file 161 is the maximum number of files (step S52). If it is not the maximum number, the file distribution request means 32 sets the divided file 161 having the file number next to the file number of the acquired divided file 161 as a candidate for the divided file 161 to be acquired (step S53). If the number is the maximum number in step S52, the file distribution request unit 32 sets the divided file 161 having the file number “1” as a candidate for the divided file 161 to be acquired (step S54).

  If the number of nodes is smaller than the number of files in step S51, the file distribution request unit 32 selects a divided file 161 having a file number next to the file number of the acquired divided file 161 as shown in FIG. The number of the divided files 161 to be acquired in order is the same as the number of offsets (step S55).

  Thereafter, as shown in FIG. 14, each of the plurality of end node devices 3 makes a request to the other end node device 3 that has decided to distribute the determined divided file 161. FIG. 14 is a file delivery request processing flow according to the present invention.

  In FIG. 14, the file distribution request means 32 of the terminal node device 3 checks whether or not the number of times is greater than or equal to the specified number (step S61). If the specified number of times is not exceeded, the file delivery requesting means 32 creates a delivery request for the divided file 161 to the terminal node device 3 having the next node ID (step S62), and the terminal node device 3 owns the divided file 161. Whether the divided file 161 is not owned or not is repeated (step S63). In step S63, when the divided file 161 is owned, the file distribution request unit 32 acquires it (step S64) and calculates an average download time (step S65).

  Thereafter, the file distribution request unit 32 checks whether or not the number of times is greater than or equal to the specified number of times (step S66). If it is not the specified number of times or more, the file distribution request means 32 creates a distribution file 161 distribution request to the terminal node device 3 having the next node ID (step S67), and the terminal node device 3 stores the divided file 161. It is checked whether or not the file is owned (step S68). If the divided file 161 is not owned, step S66 and subsequent steps are repeated. In step S68, if the divided file 161 is owned, the file distribution request unit 32 acquires it (step S69) and calculates the average download time (step S610).

  Thereafter, the file distribution request unit 32 checks whether or not the number of times is the specified number (step S611), and if it is not the specified number of times, repeats the above process up to the arbitrary number of node IDs (that is, until the specified number of times is reached). . If it is the specified number of times, the file distribution request means 32 compares the average download time and determines the ID (step S612).

  Each of the plurality of end node devices 3 repeats the process shown in FIG. 14 an arbitrary number of times, and then determines based on the status of the network 2, that is, the average download time, as shown in FIGS. A request is made to the other end node device 3 while dynamically changing the distribution request destination of the divided file 161. 15 and 16 show the file distribution request processing flow of the present invention, and both constitute a single file distribution request processing flow.

  In FIG. 15, the file distribution request means 32 of the terminal node device 3 checks whether or not the number of times is greater than or equal to the designated number (step S71). If it is not the specified number of times or more, the file distribution request means 32 requests the terminal node device 3 having the determined node ID to distribute the divided file 161 (step S72), and from the terminal node device 3 as a response to this request. It is checked whether an alternative node ID has been presented (step S73). When the alternative node ID is presented, the file distribution request unit 32 changes the request destination of the divided file 161 to the terminal node device 3 having the alternative node ID (step S74), and thereafter, the steps after step S71. repeat.

  When the alternative node ID is not presented, the file distribution request means 32 notifies the average download time to the terminal node device 3 having the determined node ID (step S75), and owns the divided file 161. Whether or not it is present (step S76), and if not owned, step S71 and subsequent steps are repeated. If it is owned, the file distribution request unit 32 acquires the divided file 161 from the terminal node device 3 (step S77), and calculates the average download time (step S78).

  In FIG. 16, the file distribution request means 32 compares the calculated average download time with the average download time received from the other end node device 3, and checks whether the former is extremely late compared to the latter (step S79). ) If it is not extremely late, step S71 and subsequent steps are repeated. If it is extremely slow, the file delivery requesting means 32 changes the request destination for requesting delivery of the split file 161 to the root node device 1 (step S710), and notifies the root node device 1 of the acquisition status of the split file 161 ( Step S711).

  Thereafter, the file distribution request unit 32 checks whether or not all of the divided files 161 have been acquired (step S712). If all of the divided files 161 have not been acquired, step S71 and subsequent steps are repeated. When all the files are acquired, the file distribution request unit 32 combines the acquired divided files 161 to restore the original large-capacity file 16 (step S713), and indicates that all the divided files 161 have been acquired. The device 1 is notified (step S714), and thereafter, the distribution status to the other end node device 1 of the divided file 161 acquired by itself is periodically notified to the root node device 1 (step S715).

  On the other hand, when the root node device 1 receives the distribution request from the end node device 3 after starting the distribution of the divided file 161, as shown in FIG. 17, the root node device 1 performs the distribution or sends a notification instead of the distribution. Do. FIG. 17 is a file distribution processing flow in the root node of the present invention.

  In FIG. 17, when the file distribution unit 12 of the root node device 1 receives a request for distribution of the divided file 161 from the terminal node device 3 (step S81), the terminal node device 3 that has acquired all the divided files 161 exists. Whether or not the requested split file 161 is distributed to the end node device 3 (step S83), and if it exists, the end node device that has acquired all the split files 161. 3 is notified as an alternative node ID to the terminal node device 3 that made the request (step S84).

  Further, the root node device 1 manages the distribution status of the large-capacity file 16 according to the status report from the end node device 3 as shown in FIG. 18 in order to enable the notification in step S84 of FIG. To do. FIG. 18 is a terminal node management processing flow of the present invention.

  In FIG. 18, when receiving the information for status management from the end node device 3 (step S91), the end node management means 11 of the root node device 1 stores it in a predetermined memory area (not shown). Based on this, the distribution status of the large capacity file 16 is managed (step S92).

  The other end node device 3 requested to distribute the divided file 161 from the end node device 3 distributes the divided file 161 to the requesting end node device 3, as shown in FIG. FIG. 19 is a file delivery processing flow in the end node device 3 of the present invention.

  In FIG. 19, when the file distribution means 33 of the end node device 3 receives a request for distribution of the divided file 161 from the other end node device 3 (step S101), has the requested divided file 161 been acquired? (Step S102), if it has been acquired, the divided file 161 is distributed to the other end node device 3 (step S103). If it has not been acquired, it is not transferred to the other end node device 3. Acquisition is notified (step S104).

  As described above, the terminal node device 3 that has received all the divided files 161 merges the divided files 161 into one original file, and restores the original large-capacity file 16. Even if the terminal node device 3 has received all the divided files 161, the terminal node device 3 distributes the divided file 161 in response to a request from another terminal device 3. For this purpose, the terminal node device 3 holds all the divided files 161 in its own data management means (not shown) until, for example, the following completion response is received. Further, the end node device 3 periodically reports the reception status of the divided file 161 to the root node device 1 at a predetermined cycle.

  When the root node device 1 knows that all the end node devices 3 have received all of the divided files 161 based on this status report, it sends a completion response indicating this to the status report from the end node device 3. Notify them. The terminal node device 3 that has received this completion response ends the reception process of the large-capacity file 16.

  As described above, according to the present invention, in a data distribution method and system, a large file can be distributed efficiently by distributing download requests and using almost the entire network bandwidth. Furthermore, according to the present invention, the distribution route can be dynamically changed according to the distribution status of the divided data, the network status, the scale of the data distribution system, and the size of the data to be distributed. Therefore, according to the present invention, a large amount of data can be downloaded to all the terminal node devices in a minimum time without increasing the download facility and the network bandwidth.

It is a block diagram of the data delivery system of this invention. It is explanatory drawing of the root node apparatus of this invention. It is explanatory drawing of the node ID table of this invention. It is explanatory drawing of the file name table and node ID table of this invention. It is explanatory drawing of the list | wrist of this invention. It is explanatory drawing of the data delivery process of this invention. It is explanatory drawing of the data delivery process of this invention. It is a terminal node management processing flow of the present invention. It is a data management processing flow of this invention. It is an initial registration processing flow of the status report of this invention. It is a file delivery request processing flow of the present invention. It is explanatory drawing of the file delivery request | requirement process of this invention. It is the file delivery request processing flow of this invention, and its explanatory drawing. It is a file delivery request processing flow of the present invention. It is a file delivery request processing flow of the present invention. It is a file delivery request processing flow of the present invention. It is a file delivery processing flow in the root node of the present invention. It is a terminal node management processing flow of the present invention. It is the file delivery processing flow in the terminal node of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Root node apparatus 2 Network 3 Terminal node apparatus 11 Terminal node management means 12 File distribution means 13 Data management means 16 Large capacity file 31 Status report means 32 File distribution request means 33 File distribution means

Claims (6)

Data comprising a root node device that is a source of data, a plurality of terminal node devices to which the data is distributed, and a network that connects them, and the data is divided into a plurality of divided data and distributed. A data distribution method in a distribution device,
The root node device has distribution information including at least a node ID assigned to the terminal node device, a total number of the plurality of terminal node devices, and a total number of divided data obtained by dividing the data. Sent to each of the node devices,
Each of the plurality of terminal node devices determines division data that the terminal node device first requests distribution to the root node device based on the distribution information, and distributes the divided data to the root node device. Request,
In response to a request for distribution of the divided data, the root node device distributes the divided data to each of the plurality of terminal node devices,
Each of the plurality of terminal node devices determines which of the other terminal node devices to distribute the divided data other than the divided data distributed from the root node device based on the distribution information; Requesting the delivery of the divided data to the other terminal node device determined as described above. The data distribution method according to claim 1, wherein, when the root node device distributes the data, the distribution information is transmitted to each of the plurality of end node devices prior to the distribution. Each of the plurality of terminal node devices measures a distribution speed at which the divided data is distributed to each of the other terminal node devices, and determines the other terminal node device determined based on the distribution speed of the distribution speed. 2. The data distribution method according to claim 1, wherein a change is made to a fast end node device, and distribution of the divided data is requested to the changed end node device. The root node device transmits a delivery time, which is a time at which the end node device starts a request for delivery of the data, to each of the plurality of end node devices together with the delivery information;
2. The data distribution method according to claim 1, wherein each of the plurality of terminal node devices starts a request for distribution of the data to the root node device after the elapse of the distribution time. The root node device calculates a hash value for each of the divided data obtained by dividing the data, and transmits the calculated hash value to each of the plurality of terminal node devices as part of the distribution information,
Each of the plurality of terminal node devices calculates a hash value for the divided data received from the root node device, and compares the hash value with the hash value for the received divided data included in the distribution information. The data distribution method according to claim 1, wherein if the two match, the received divided data is managed as correct data. Data comprising a root node device that is a source of data, a plurality of terminal node devices to which the data is distributed, and a network that connects them, and the data is divided into a plurality of divided data and distributed. A distribution device,
The root node device is
Transmission information including at least a node ID assigned to the terminal node device, a total number of the plurality of terminal node devices, and a total number of divided data obtained by dividing the data is transmitted to each of the plurality of terminal node devices. Means to
Means for distributing the divided data to each of the plurality of terminal node devices in response to a request for distribution of the divided data from each of the plurality of terminal node devices;
The plurality of end node devices are each
The terminal node device first determines divided data to request distribution to the root node device based on the distribution information, and includes means for requesting the root node device to distribute the divided data. Data distribution system.

Images