JP6712744B2 - Network system, cache method, cache program, management device, management method and management program - Google Patents

Description

The present invention relates to a network system, a cache method, a cache program, a management device, a management method, and a management program.

Heretofore, various techniques have been proposed for intermediately caching files transmitted between computers connected to a network. By downloading the cache file instead of the original file, it is possible to avoid concentration of access to the original file and to shorten the time until the download is completed on the receiving side.

For example, when the management server obtains a request for content, it searches the original node and the cache overlay node for a node that holds the content, and selects the node with the smallest number of hops with the user as the content delivery source. A technique of selecting a node has been proposed (Patent Document 1). In this technology, the overlay node determines a route between the content distribution source node and the distribution requesting user, distributes the content according to the determined route, and if a cache overlay node exists on the content distribution route, Cache the content.

Also, a technique for caching moving images in communication between mobile phone terminals has been proposed. For example, a technique of grouping moving images by genre, grouping users by preference, and determining moving images to be cached in a user's mobile phone according to favorite genre (Non-Patent Document 1), user's SNS (Social Network Service) ) A technique for determining content to be cached based on the above behavior and liquidity pattern has also been proposed (Non-Patent Document 2).

Also, a technique for determining which server holds the content has been proposed (for example, Non-Patent Document 3). A technique that uses a genetic algorithm to determine the distributed arrangement of caches that minimizes the communication cost has also been proposed (for example, Non-Patent Document 4).

JP, 2012-114649, A

Xiangyang Zhang, Ying Wang, Ruijin Sun, Dong Wang, “Clustered Device-to-Device Caching Based on File Preferences” IEEE PIMRC 2016 Xiaofei Wang, Min Chen, Zhu Han, Dapeng Oliver Wu, Ted “Taekyoung” Kwon, “TOSS: Traffic Offloading by Social Network Service-Based Opportunistic Sharing in Mobile Social Networks” IEEE INFOCOM 2014 Bruce M. Maggs, Ramesh K. Sitaraman, “Algorithmic Nuggets in Content Delivery” ACM SIGCOMM Computer Communication Review Volume 45 pp.52-66 Zhe Li and, Gwendal Simon, “In a Telco-CDN, Pushing Content Makes Sense”, IEEE Transactions on Network and Service Management

Conventionally, when files are cached by being distributed to multiple nodes that make up a network system, it generally takes a long time to determine the file arrangement that can avoid the concentration of access and reduce the total communication volume. There was a problem.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a network system having a simple configuration that avoids concentration of access to files and reduces the total communication volume.

A network system according to the present invention includes a plurality of node devices that temporarily hold files to be transmitted and received. A node device is one of a plurality of groups for identifying a file to be held by the node device, and belongs to a group different from the adjacent node device in the network topology, and the file has zero or more groups. Associated with the indicated information. Further, when the node device receives a file associated with the information indicating the group to which the node device belongs, the node device stores the file in a cache holding unit that temporarily holds the file as a cache file and a file requested by another node device. When the corresponding cache file is held, the cache transmission unit includes a cache transmission unit that transmits the cache file to another node device.

By assigning different groups to adjacent node devices and associating the groups to be cached with the files, the files can be appropriately distributed and cached on the network. Further, in other words, it is possible to provide a sub-optimal network system with a simple configuration, which can avoid the concentration of access to the original file and reduce the total communication volume.

Further, there may be four or more groups. For example, when the nodes on the network topology where the links do not intersect are divided into groups, it can be considered in the same manner as the problem that the areas on the plane are colored in different colors. Therefore, according to at least four groups, the nodes can be grouped so that the same group is not adjacent.

Further, the access management apparatus may further include an access management apparatus that calculates the frequency of access to the file, and the access management apparatus may change the information indicating the group associated with the file based on the calculated frequency. By doing so, it is possible to change the group of nodes that should cache the file, following changes in the access frequency tendency. The access management device may be a device that delivers the original file.

Further, the access management device notifies the node device of the deletion of the information indicating the group assigned to the file to delete the file, and the node device deletes the cache file based on the notification. You may make it further have a part. When changing the association of a group with a file as described above, the cache deletion in the node device may also be controlled.

In addition, the cache holding unit is based on a white list for specifying files that are allowed to be cached, a black list for specifying files that are not allowed to be cached, or a flag indicating whether or not to allow the files to be cached. It may be determined whether or not to retain the received file. By doing so, it is possible to explicitly define a file whose cache is prohibited, and to prevent leakage of confidential information, for example.

Further, the file may be linked from a document in a predetermined hypertext, and information indicating a group to be assigned to the file may be assigned based on the number of links passing from the reference document. By doing so, even if there is no information such as the access frequency, the file to be cached can be preferentially defined.

The file may be divided into a plurality of partial files by the secret sharing method, and each partial file may be associated with information indicating a different group. By doing so, it is possible to prevent information leakage.

Further, the cache method according to the present invention is executed by a network system including a plurality of node devices that temporarily hold files to be transmitted and received. Further, the node device is one of a plurality of groups for specifying the file to be held by the node device, belongs to a group different from the adjacent node device in the network topology, the file is zero or more. Associated with information indicating a group. Then, when the node device receives the file associated with the information indicating the group to which the node device belongs, the step of temporarily holding the file as a cache file and the file requested by another node device are supported. When the cache file is held, the step of transmitting the cache file to another node device is executed.

Further, the management device according to the present invention is one of a plurality of groups for specifying a file to be held by itself, and a group different from an adjacent node device is preferentially assigned in the network topology. When a file associated with the information indicating the group to which it belongs is received, it is connected to a network system including a plurality of node devices that temporarily hold the file as a cache file. Then, the management device counts the acquisition requests for the file from the node device, determines a frequency calculation unit that calculates the frequency, and a group of zero or more assigned to the file based on the frequency calculated by the frequency calculation unit, and determines the file A setting changing unit that transmits information indicating a combination with a group to the node device.

In this way, it is possible to set the group of nodes that should cache the file according to the frequency of acquisition requests for the file. The management device may be a device that delivers the original file.

Further, the setting changing unit may allocate more groups to files that have a higher frequency of receiving an acquisition request. By doing so, it becomes possible to cause more node devices to cache popular files that are frequently requested.

Further, the management method according to the present invention is one of a plurality of groups for identifying a file to be held by itself, and a group different from an adjacent node device is preferentially assigned in the network topology. When a file associated with the information indicating the group to which it belongs is received, the management device connected to the network system including a plurality of node devices temporarily holding the file as a cache file executes it. Specifically, the number of acquisition requests for the file from the node device is counted, the step of calculating the frequency, the group of zero or more to be assigned to the file based on the frequency calculated by the frequency calculation unit is determined, and the file and the group And a step of transmitting information indicating the combination to the node device.

The contents described in the means for solving the problems can be combined as much as possible without departing from the problems and technical ideas of the present invention. The contents of the means for solving the problems can be provided as a device such as a computer or a system including a plurality of devices, a method executed by the computer, or a program executed by the computer. A recording medium holding the program may be provided.

With a simple configuration, it is possible to provide a network system that can avoid the concentration of access to files and reduce the total communication volume.

It is a figure showing an example of composition of a network system concerning an embodiment. It is a figure for demonstrating the information which shows the group attached to a file. It is a block diagram showing an example of composition of a node. It is a block diagram showing an example of composition of a distribution server. It is a processing flow figure which shows an example of the file delivery process which a network system performs. It is a figure which shows the example of a full mesh type network. It is a figure showing an example of a bus type network. It is a figure which shows the Example in a network system of a company. It is a processing flow figure which shows an example of cache object update processing. It is a figure which shows an example of the tag which shows the number of access to a file in a predetermined period, and the group allocated in association with the said file. 3 is an example of information stored in the node 11a in FIG. 1 and indicating a direction in which a nearby node exists. It is an example of information stored in the node 11b of FIG. 1 and indicating a direction in which a nearby node exists. FIG. 6 is a diagram for explaining an example in which a relay node has a white list indicating permission or prohibition of cache. It is a figure which shows the example which memorize|stores the tag which shows the group which should cache the said file with respect to a file, and the information for specifying the user who is permitted access to the said file, linking. It is a figure for demonstrating the procedure of determining cache arrangement|positioning using a genetic algorithm. It is a graph which shows the amount of communications with an external server. 6 is a graph showing files and the number of files cached on the network. It is a graph which shows the amount of communication in a network.

Hereinafter, embodiments for carrying out the present invention will be described with reference to the drawings. The configuration of the embodiment is an example, and the present invention is not limited to the configuration shown in the embodiment.

<Network configuration>
FIG. 1 is a diagram showing an example of a configuration of a network system (also simply referred to as “network”) 1 according to the embodiment. The network 1 according to the embodiment is formed by a plurality of computers that are communicably connected to each other via, for example, an electric communication line, and various types of electronic data (also referred to as “file”) stored in the computers are stored between the computers. Can be sent and received at. The mesh type network shown in FIG. 1 is a model representing an example of a backbone provided in Japan.

Each of the hatched perfect circles shown in FIG. 1 represents a computer (also referred to as “node” or “node device”) connected to the network 1. It should be noted that some of the nodes are denoted by reference numerals 11a to 11e for convenience. Further, each node is connected to one or more adjacent nodes by a transmission line (also referred to as “link”) such as an optical fiber cable shown by a straight line in FIG. A node that is directly connected to a certain node by a link without passing through another node is called an adjacent node.

The network 1 also includes a distribution server 2 that is a computer that stores original files and distributes the files in response to a request from a node, for example. The distributed file is transmitted from the distribution server 2 to the requesting node via a node on the route. Further, in this embodiment, each node also functions as a cache server. The cache server temporarily stores (caches) the file to be transmitted and received as a cache file, and transmits the cache file in response to a request for the same file.

Further, in this embodiment, a plurality of nodes distributes and caches the same file. Then, the node requesting the download of the file receives the file from, for example, the node having the smallest number of hops from its own node among the nodes that cache the file. Specifically, a group for specifying the file that each node should hold is defined. Each node is assumed to belong to one of a plurality of predetermined groups. In the example of FIG. 1, the hatching pattern applied to each node represents the group to which the node belongs. On the other hand, each file has information indicating zero or more groups that should cache the file. Then, each node caches the file when transmitting/receiving the file having the information indicating the group to which the node belongs.

In the example of FIG. 1, the nodes are classified into four groups. Here, according to the four-color theorem, it can be seen that in a mesh type network in which links do not intersect as shown in FIG. 1, nodes can be colored so that the same color does not adjoin if there are at least four colors. When the nodes are divided into N groups so that the adjacent nodes belong to different groups, the groups can be determined by solving the coloring problem with N colors. Specifically, by assigning colors with a long inter-color distance in order from a node with a high degree, it is possible to color the same colors so that they are not adjacent to each other. That is, the colors existing near the nodes are excluded in order from the node with the most connected links, and the remaining colors are added. When a new node is connected to the existing network, the color may be different from that of the adjacent node, or the entire network may be repainted. The number of groups is preferably four or more, but depending on the network, there may be cases in which the number of groups is less than four and can be painted separately. Hereinafter, the group according to the present embodiment will also be referred to as “color”.

FIG. 2 is a diagram for explaining information indicating a group attached to a file. The file 3 in FIG. 2 is various electronic data generated, read, written, stored, etc. by a computer, and is a moving image, a document, an application execution file, or the like. Further, the file 3 is associated with information indicating zero or more of the above groups. In the example of FIG. 2, one file 3 is associated with a combination of 0 to 4 groups represented by a hatching pattern as shown in the balloon. Specifically, in each node, for example, a tag represented by a bit string having the number of digits corresponding to the number of groups is held in association with the file identification information, so that each node specifies a file to be cached. be able to. Also, such a list of files may be stored in a database table in each node. Further, such association is managed by a management device such as the distribution server 2 in a centralized manner, and for example, a list of files to be cached is broadcast to a node on the network and held in a storage device of each node. Alternatively, it may be stored in a predetermined management device on the network so that each node can read it. Note that when there are four groups, 2 ⁴ types of tags can be generated with a 4-digit bit string. Also, the more frequently accessed files may be cached by more groups. Further, it is not necessary to associate a group with all files, and a file with a tag of "0000" indicates a file that is not cached in any group.

When the node shown in FIG. 1 receives a file having information indicating a group to which the node belongs, the node caches the file and outputs the information cached by itself in response to a request from another node. To do.

<Node configuration>
FIG. 3 is a block diagram showing an example of the configuration of a node. The node 4 is a general computer and includes a communication I/F 41, a storage device 42, an input/output device 43, a processor 44, and a bus 45.

The communication I/F 41 is, for example, a network card, and communicates with another computer via the link shown in FIG. 1 based on a predetermined protocol. For example, the file 3 is transmitted and received between the nodes.

The storage device 42 is a main storage device such as a RAM (Random Access Memory) or a ROM (Read Only Memory) and an auxiliary storage device (secondary storage) such as a HDD (Hard-disk Drive), an SSD (Solid State Drive), or a flash memory. Device). The main storage device temporarily stores programs and data read by the processor and reserves a work area for the processor. The auxiliary storage device stores a program executed by the processor, a cache of transmitted and received files, information indicating a group to which the node 4 belongs, and the like.

The input/output device 43 is, for example, an input device such as a keyboard and a mouse, an output device such as a monitor, and a touch panel. For example, the user requests the distribution server to transmit a file via the input/output device 43 or reads out and displays the received file.

The processor 44 is an arithmetic processing unit such as a CPU (Central Processing Unit), and executes each application according to the present embodiment by executing an application (also referred to as a program or software). In the example of FIG. 3, functional blocks are shown in the processor 44. The processor 44 functions as, for example, each processing unit of the file request unit 441, the transmission relay unit 442, the cache holding unit 443, the cache transmission unit 444, the cache deletion unit 445, and the security control unit 446. The file request unit 441 requests a file delivery server connected via a network to send a file, for example, based on a user operation. When the node 4 is located on the path between another node and the distribution server, the transmission relay unit 442 relays a file transmission request from the other node to the distribution server or a file to be transmitted to another node. The cache holding unit 443 determines whether or not to hold the received file based on the information indicating the group associated with the file, and causes the storage device 42 to temporarily hold the received file. In response to a file request from another node to the distribution server, the cache transmission unit 444 transmits the cached file to another node device when the file is cached in the storage device 42 of the node 4. The cache deleting unit 445 deletes the cache stored in the storage device 42 based on a predetermined rule. Further, the security control unit 446 performs a predetermined concealment when holding the cache.

The above components are connected via the bus 45.

<Structure of distribution server>
FIG. 4 is a block diagram showing an example of the configuration of the distribution server. The distribution server 2 is also a general computer, and includes a communication I/F 21, a storage device 22, an input/output device 23, a processor 24, and a bus 25.

The communication I/F 21 is, for example, a network card, and communicates with another computer via the link shown in FIG. 1 based on a predetermined protocol. For example, the file 3 is transmitted/received to/from an adjacent node.

The storage device 22 is a main storage device such as RAM or ROM and an auxiliary storage device such as HDD, SSD, flash memory, or the like. The main storage device temporarily stores programs and data read by the processor and reserves a work area for the processor. The auxiliary storage device stores programs executed by the processor, original files to be distributed, and the like. Further, information indicating a group that caches the file may be stored in association with each file. Furthermore, the transmission request for the file may be counted, and the setting of the group for caching each file may be changed based on the frequency in a predetermined period.

The input/output device 23 is, for example, an input device such as a keyboard and a mouse, an output device such as a monitor, and a touch panel.

The processor 24 is an arithmetic processing unit such as a CPU and executes each application by executing an application. In the example of FIG. 4, functional blocks are shown in the processor 24. The processor 24 functions as, for example, each processing unit of the file distribution unit 241, the frequency calculation unit 242, the setting change unit 243, and the security control unit 244. The file distribution unit 241 reads out the file for which the transmission request is received from the storage device 22 and transmits it to the requesting node. The frequency calculation unit 242 counts the number of accesses for each file and calculates the access frequency within a predetermined period. The setting changing unit 243 changes the information indicating the group that caches each file according to the access frequency. When the setting is changed, the setting changing unit 243 may transmit the information instructing the node on the network to delete the specific cache file. Further, the security control unit 244 performs a predetermined concealment on the file to be transmitted.

The above components are connected via the bus 25.

<File distribution process>
FIG. 5 is a process flow diagram showing an example of a file distribution process performed by the network system. The request source node of FIG. 5 corresponds to any of the nodes shown in FIG. 1 and includes the functional unit shown in FIG. The distribution server of FIG. 5 corresponds to the distribution server 2 shown in FIG. The relay node in FIG. 5 corresponds to the node existing on the route between the request source node and the distribution server 2 in FIG. That is, it may pass through two or more relay nodes or may not pass through relay nodes. In FIG. 5, one relay node is shown for convenience.

First, the file request unit 441 of the request source node requests transmission of the file held by the distribution server (FIG. 5: S1). On the other hand, the transmission relay unit 442 of the relay node receives the request (S2) and determines whether or not the cache file corresponding to the storage device 42 of the own node is held (S3). When the cache is held (S3: Yes), the cache transmission unit 444 of the relay node transmits the cache file to the request source node (S4). When the relay node does not hold the cache (S3: No), the transmission relay unit 442 of the relay node relays the request to the adjacent relay node (not shown) or the distribution server (S5). When there are a plurality of relay nodes, similar processing is repeated. Then, the file distribution unit 241 of the distribution server 2 receives the transmission request for the file (S6), reads the corresponding file from the storage device 22 and transmits it (S7). On the other hand, the transmission relay unit 442 of the relay node relays the file to the request source node (S8). Then, after S8, or after the relay node transmits the cache in S4, the file request unit 441 of the request source node receives the file requested in S1 (S9). When transmitting the file to the request source node in S8, the cache holding unit 443 of the relay node determines whether the file is a target to be cached in the own node (S10). In this step, it is determined whether or not the file is associated with the information indicating the group to which the own node belongs. The association between the file and the information indicating the group can be stored in a table or file of a database held by the storage device 42 of the own node or a predetermined device on the network. Then, when it is determined that the relayed file is the target to be cached (S10: Yes), the cache holding unit 443 of the relay node temporarily holds the file in the storage device 42. On the other hand, when it is determined that the relayed file is not the target to be cached, the relay node does not hold the file. The network system 1 repeats the file delivery processing as shown in FIG. 5 in response to a file transmission request from the request source node.

<Effect>
According to the present embodiment, by a simple method of grouping nodes so that nodes belonging to the same group are not adjacent to each other, the nodes that cache the same file are distributed on the network, and some nodes are close to each other. It is possible to realize a sub-optimal cache arrangement in which cache servers belonging to each group exist at the specified positions. Further, as shown in FIG. 2, information indicating a plurality of groups may be associated with one file. If the files that are frequently accessed are cached by the nodes belonging to many groups, the cache hit rate on the network can be improved.

Although FIG. 1 exemplifies a mesh type network showing a domestic backbone, the network according to the embodiment can be applied to networks of various scales and topologies. For example, it may be applied to a network system of an organization such as a company in which a network is formed for each of a plurality of bases and a common file server is provided on a so-called cloud. By caching the files in the intranet of each site, it can be said that generally, the time required to acquire the files can be shortened. Further, when a charge is incurred according to the amount of communication with the cloud, the cost can be reduced by using the cache.

6 and 7 are diagrams showing other examples of network topologies. The topology of the network according to the present embodiment is not particularly limited, and can be applied to so-called ring type, star type, tree type, line type, etc. in addition to the mesh type. However, as shown in FIG. 6, in a full-mesh type network (FIG. 6: left diagram), by deleting an intersecting arbitrary link (FIG. 6: right diagram), nodes belonging to the same group are adjacent to each other. You may make it possible to classify into four groups so that it may not be done. Also, by increasing the number of groups, nodes belonging to the same group may not be adjacent to each other.

Further, FIG. 7 is an example of a bus type network. For a bus type network (Fig. 7: upper diagram), for example, by adding a definition of a new group (Fig. 7: colorless) and treating the switch as a node belonging to the new group (Fig. 7: lower diagram), the same The nodes belonging to the group may not be adjacent to each other. Here, the layer 2 switch, the layer 3 switch, the router, etc. are referred to as “switch”. The switch of FIG. 7 transfers files, but the auxiliary storage device has a capacity of zero. Further, a computer connected to one switch may be defined as a node having a distance of zero, and the groups may be allocated so as to be distributed in a plurality of nodes having a distance of zero.

FIG. 8 is a diagram showing an example of distributing and caching files in employees' computers in a network system of a company having four bases. The bases in Tokyo, Nagoya, Osaka and Hakata are connected via a VPN (Virtual Private Network) constructed on the Internet. In addition, it is assumed that the distribution server 2 that provides the cloud service stores the content used in the company and downloads it from the distribution server 2 as appropriate. In the network system of FIG. 8, a group may be assigned such that adjacent nodes belong to different groups as a node that temporarily caches files transmitted/received by the VPN server at each base. Further, at each base, a plurality of nodes having a distance of zero are connected via a switch. In the example of FIG. 8, the nodes are assigned so that the groups are distributed even at the respective bases. When accessing a desired file, the node connected to the local network of the base downloads from the node in the base when cached in the node in the base. If it is not cached in the node in the base, it is downloaded from another base or the distribution server 2. By doing so, it is possible to suppress an increase in the amount of communication with the distribution server 2 on the cloud and reduce the time required to download the file.

<Cache target update process>
Further, in the above-mentioned network system, the priority of the cache for the file may be determined, and the information indicating the group associated with the file (list of files to be cached) may be dynamically changed. The cache priority is determined based on the access frequency to each file calculated by the frequency calculation unit 242 of the distribution server 2 or another management device.

FIG. 9 is a process flow diagram showing an example of a cache target update process for changing a cache target file. For example, the network system 1 performs the process shown in FIG. 9 in addition to the file distribution process shown in FIG.

First, the relay node counts the number of times a cache file is transmitted in response to a request for each file, and transmits the number of times to the distribution server 2 at a predetermined timing (FIG. 9: S21). On the other hand, the distribution server 2 receives the file transmission count from each relay node and stores the access count for each file (S22). In this step, an access log for counting the number of accesses may be received. Further, in the process flow diagram shown in FIG. 4, for example, when the distribution server transmits a file in S7, an access log to the file is further written in the storage device 22 (S23).

Then, the distribution server 2 calculates the number of times each file has been accessed (that is, the frequency of access) in a predetermined period at a predetermined timing (not shown), and the higher the frequency of the file, the higher the cache priority. Have many groups cache. That is, information indicating more groups is associated with the file. Specifically, first, the frequency calculation unit 242 determines whether it is time to update the cache target (S24). For example, when a predetermined operation time arrives, or when a condition such as a case where a predetermined period has elapsed from the previous processing is satisfied, it is determined that the cache target is a timing. If it is determined that it is not the time to update the cache target, the distribution server 2 does not perform any processing. On the other hand, when it is determined that it is time to update the cache target (S24: Yes), the frequency calculation unit 242 calculates the access frequency for each file and determines the cache target (S25).

FIG. 10 is a diagram showing an example of tags indicating the number of accesses to a file in a predetermined period and a group assigned in association with the file. The distribution server 2 tabulates the access log for the file transmitted by the distribution server 2 in response to the request and the access log indicating that the relay node has transmitted the cache for a predetermined period, and for each file as shown in FIG. Generate statistical data. The table shown in FIG. 10 has items of “number of accesses”, “rank”, “popularity”, “file name”, and “tag”. In the “access count” field, the number of accesses to the file in a predetermined period, and a value representing the frequency is recorded. In S25 of FIG. 9, the number of accesses in a predetermined period is calculated. In the “rank” field, a numerical value indicating the order of files arranged based on high frequency is recorded. In the "popularity" field, information indicating a category classified based on high frequency or rank is recorded. In the “file name” field, a name, which is an example of identification information of each file, is recorded. In the "tag" field, information indicating a group that caches the file is recorded in the tag format described above.

The setting changing unit 243 also determines a group in which each file is cached according to the access frequency (FIG. 9: S25). In this step, a cache group is assigned to each file. First, the degree of popularity shown in FIG. 10 is determined based on the number of accesses in a predetermined period. For example, when the nodes are divided into four groups, the degree of popularity is classified into 5 levels. That is, popular "high" files are cached in all four groups, popular "middle and high" files are cached in three groups, and popular "medium" files are cached in two groups. Files are cached in one group, and popular "low" files are not cached in any group. In this way, the number of groups of nodes that cache the file is determined according to the popular category, and then the combination of groups is cyclically allocated to the files to which the same number of groups are allocated. For example, the tags determined by cyclically assigning the combinations of the three groups are recorded in the “tag” field of the records of the 11th rank and on in FIG. In S25, the tag associated with the file is updated based on the access frequency to each file in the most recent predetermined period.

Then, the setting changing unit 243 distributes a list indicating a group in which each file is a cache target to each node (FIG. 9: S26). In this step, for example, the information shown in the columns of “file name” and “tag” in FIG. 10 is broadcast to each node. When the relay node receives the list, the relay node stores the list in the storage device 42 and thereafter determines the file to be cached by the own node based on the list (S27).

In addition, in this embodiment, each node may store in advance the direction in which a neighboring node to which each group is assigned exists. FIG. 11A is an example of information stored in the node 11a in FIG. 1 and indicating the direction in which a nearby node exists. 11B is an example of information stored in the node 11b in FIG. 1 and indicating the direction in which a nearby node exists. The tables of FIGS. 11A and 11B include a “group” column and a “destination node” column. Identification information indicating a group assigned to a node and a file is registered in the “group” field. Here, the horizontal hatching in FIG. 1 is called a group A, the thick solid line hatching from the upper right to the lower left is a group B, the vertical broken line hatching is a group C, and the thin solid line hatching from the upper left to the lower right is called a group D. I shall. Further, in the field of “destination node”, identification information indicating a node which is a destination of a request for accessing a node which can be reached with the minimum hop number among nodes belonging to each group is registered. Specifically, identification information such as the MAC address, IP address, and port number of the device is registered according to the configuration of the network system and the adopted protocol. The node closest to the node 11a in FIG. 1 and belonging to the group A is the node 11d, which can be accessed via the node 11b. Therefore, in FIG. 11A, the node 11b is registered in association with the group A. Further, the node closest to the node 11a in FIG. 1 and belonging to the group B is the node 11c, and the node 11c is registered in association with the group B in FIG. 11A. The node belonging to the group C closest to the node 11a in FIG. 1 is the node 11b, and the node 11b is registered in association with the group C in FIG. 11A. For the group D to which the own node belongs, no other node is registered in the field of the destination node.

The node 11a can know that a certain file is cached by the node belonging to any group based on the list received in S27 of FIG. Then, based on the information shown in FIG. 11A, the file can be requested to the closest node to which the group to which the desired file belongs is assigned. Similarly, the node 11b holds information as shown in FIG. 11B. For example, when receiving a request from the node 11a to acquire a file belonging to the group A, the node 11b refers to the record in which "A" is registered in the group field in the information shown in FIG. It can be seen that the closest node belonging to A is node 11d. Further, the nearest nodes belonging to the group B based on the node 11b are the nodes 11c and 11e each having a hop number of 2, and the information shown in FIG. 11B shows that the nodes 11a and 11d are the destination nodes. Two are registered. In such a case, one of a plurality of destination nodes may be registered as a main system and normally used, and the other may be used when a failure such as network disconnection occurs. Further, a round robin method in which a plurality of destination nodes are sequentially used may be adopted to distribute the load on the nodes and communication lines.

The group assigned to each node is basically unchanged unless a new node is added between nodes or an existing node between nodes is deleted. Therefore, basically, it is not necessary to change the information shown in FIGS. 12A and 12B.

<Effect>
According to the cache target update process, a file newly distributed from the distribution server can be cached according to the number of subsequent accesses, and even if the tendency of the access frequency to the existing file changes, The settings can be changed so that the cache hit rate in the network can be maintained and improved by following the changes. The access frequency may be calculated and the list of files to be cached may be updated by a management device on a network different from the distribution server 2. The access log to the file stored in each relay node may be held in a predetermined device on the network so that the distribution server 2 or the management device can read the access log in each relay node.

<Modification 1>
In the network system described above, information indicating whether or not the cache of the file is permitted may be further set. The information indicating whether or not to permit the cache may be held as metadata in each file, for example. Further, each node may hold information indicating permission/prohibition of cache by a black list or a white list. The black list or the white list may be described by, for example, a URL (Uniform Resource Locator) filter, or may be described by a list having identification information such as a file name or a hierarchical structure indicating a file storage location. Good.

In such a case, in the process flow diagram shown in FIG. 4, for example, the relay node further determines whether the file relayed in S10 is permitted to be cached, and if the cache is not permitted, the cache is not held. To do so.

FIG. 12 is a diagram for explaining an example in which a relay node has a white list indicating permission/prohibition of cache. In FIG. 12, the distribution server 2 holds files such as “A.pdf” to “D.pdf”. In addition, the relay node has a white list indicating that caches such as “A.pdf” to “C.pdf” are permitted in the regular expression. In such a case, “A.pdf” is transmitted from the distribution server 2 to the terminal 1 in response to the request from the terminal 1 connected to the relay node, and the distribution server 2 receives the request from the terminal 2 in the terminal 2. It is assumed that “D.pdf” is transmitted to. In such a case, even if the information indicating the group to which the relay node belongs is associated with both “A.pdf” and “D.pdf”, the relay node is permitted to cache in the whitelist. "A.pdf" is cached, and "D.pdf" that is not allowed to be cached in the white list is not cached.

According to the first modification, since the cache can be explicitly prohibited, it is possible to prevent the leakage of information when the file is confidential such as employee's pay slip. Further, by holding the blacklist or the whitelist in the node, it is possible to set the permission/prohibition of the cache independently of the degree of popularity of the file shown in FIG. That is, for example, the distribution server 2 re-obtains the order of files based on the access frequency or sets the file tag to "0000" (even if the setting indicating whether or not to allow a cache is changed for a certain file. It is not necessary to change the setting so that nodes belonging to any group will not be cached. Also, by holding a blacklist or whitelist in a node, different settings can be made for each node. For example, a certain confidential file may be set to permit cache to nodes forming a local network or private cloud inside the company and to reject cache to nodes forming a public cloud outside the company. ..

<Modification 2>
It may be possible to predict a file that is easily accessed and determine whether or not to cache the file. For example, if a file is linked from a document in hypertext format, predict that it will be easier to access as the number of links that go through the reference document, such as the top page of a website, is smaller. Good. That is, it is determined that a file that arrives from the top page with a few clicks is likely to be accessed and is cached. Specifically, the web crawler circulates the website and counts the links that go through to reach each file, and creates a list of files in which the number of the links that go through is less than a predetermined threshold value. By doing so, it is possible to automatically extract a file that is predicted to be easily accessed and generate a setting for caching the file.

The setting for caching the file is broadcast from, for example, the distribution server and stored in each node. In this case, the file may be distributed and cached in advance, or may be cached when transmission/reception of the corresponding file first occurs in each node.

According to the second modification, it is possible to determine whether or not to cache the file based on a predetermined criterion without calculating the access frequency for each file. Further, for example, it is possible to judge whether or not a file, which is newly distributed from the distribution server and whose access frequency is unknown, should be cached based on a predetermined standard.

<Modification 3>
Further, the file held by the distribution server may be updated at any timing. In this modification, the relay node confirms or notifies whether or not the cached file needs to be reacquired.

For example, even when the relay node caches the file requested by the request source node, the relay node may check whether or not the distribution server has updated each time. Specifically, when it is determined that the cache is held in S3 of the process flow diagram shown in FIG. 4 (S3: Yes), it is further determined whether the file held by the distribution server has been updated (not shown). No). Whether or not the file is updated may be determined based on whether or not the time stamp is updated for the file on the distribution server having the same identification information as the cached file such as the file name. Alternatively, the determination may be made based on the checksum calculated from the file by a predetermined algorithm. If the distribution server has not updated the file, the process proceeds to S4 of FIG. 4 to send the cache file to the requesting node. On the other hand, when the file is updated in the distribution server, the process proceeds to S5 of FIG. 4 and the file is acquired from the distribution server. By doing so, even when the file is updated in the distribution server, the latest file can be promptly transmitted to the requesting node.

Alternatively, the relay node may set a predetermined expiration date for the cache file, and when the expiration date is exceeded, the relay server may confirm whether or not the cache file is updated. For example, when the cache is held in S11 of the processing flow chart shown in FIG. 4, the date and time at that point or the date and time indicating the expiration date after a predetermined period from that point is associated and held in the storage device 42. After that, when it is determined in S3 that the cache is held (S3: Yes), it is further determined whether or not the predetermined period has elapsed from the time when the cache is performed, and if it is exceeded, the file held by the distribution server is updated. It is judged whether it is not done (not shown). Subsequent processing is the same as in the case of checking every time. With this configuration, it is possible to check the presence/absence of the update at an appropriate timing while suppressing an increase in communication that occurs to confirm the update of the file.

In addition, the distribution server may generate a difference of a list indicating files to be cached again when the original file is added, updated (changed), deleted, etc., and broadcast it to each node (Fig. (Not shown). When the notification of the update or the like is received in this way, the relay node transitions to S5 in S3 of FIG. 4 regardless of the presence or absence of the cache, and acquires the updated file from the distribution server. In this way, the latest file is promptly transmitted to the requesting node even if the file is updated in the distribution server while suppressing the increase in communication that occurs to confirm the file update. be able to.

<Modification 4>
Further, the cache may be deleted in the relay node at an appropriate timing. For example, in the second modification, the list of files to be cached is updated based on the access frequency to the files, and a command to delete the cache from files that are not in the list is broadcast to each node. On the other hand, the cache deletion unit 445 of the relay node deletes the cached file from the storage device 42 according to the instruction.

Also, the cache file to be deleted may be determined by each relay node. For example, the LFU (Least Frequently Used) method may be adopted, and the cache deleting unit 445 may preferentially delete the cache file having a low access frequency.

According to the modified example 4, the storage area of the relay node required for the file cache can be appropriately released.

<Modification 5>
Further, the relay node or the distribution server 2 may include a security control unit that conceals or encrypts the cache file. The security control unit 446 may store, for example, information indicating a user who is permitted to access a file in association with each other, and authenticate the user when the request source node requests transmission of the file. ..

FIG. 13 shows an example in which a tag indicating a group in which the file should be cached and information (ACL (Access Control List)) for identifying a user who is authorized to access the file are stored in association with each other. FIG. In the example of FIG. 13, information indicating the department to which the user belongs is registered in the ACL field. Then, the request source node transmits a request including the user authentication information to the relay node in S1 of FIG. On the other hand, when the relay node determines in S3 of FIG. 4 that the cache file is held (S3: Yes), the relay node further determines whether to permit transmission of the requested file based on the department to which the user indicated by the authentication information belongs. To do. Then, when the access is authorized to the user, the process proceeds to S4 and the cache file is transmitted to the request source node.

Further, the communication relating to the transmission/reception of the file may be encrypted by using, for example, SSL (Secure Socket Layer)/TLS (Transport Layer Security). By doing so, it is possible to prevent the leakage of information due to interception of communication data and spoofing.

Further, the security control unit 244 of the distribution server 2 may divide the file by using the secret sharing method and cache the divided files by the nodes belonging to different groups. For example, the file is divided into the number of groups to which the node belongs, and in the list as shown in FIG. 10, different tags are associated with each of the divided files. Information leakage can be prevented also by such a method.

Further, the security control unit 446 of the relay node may encrypt the file and cache the file, calculate the hash value of the cache file, and store the hash value in the storage device 42 in association with the cache file. Further, the security control unit 446 compares the hash value calculated from the held cache data with the hash value held by the storage device 42, and if different, detects that the cache file has been tampered with. It is also possible to judge and delete the cache file. By doing so, the integrity of the cache file to be held is guaranteed. Further, by encrypting, only the user who has the proper authority can read the file. Therefore, the confidentiality of the cache file can be secured.

<Modification 6>
The network system according to the above-described embodiments and modifications can be applied to a content delivery network (CDN) for delivering digital content such as moving images. That is, the above-mentioned files are moving image data, audio data, VR (Virtual Reality) contents, 3D contents, data used for games, and other digital contents, and the node 4 is a cache server forming a CDN, etc. Good.

The above-mentioned adjacent nodes on the network topology are not intended to include no device other than wiring between the nodes, and the adjacent nodes among a plurality of nodes forming a predetermined network such as a CDN are not included. Point to. The node 2 also functions as a so-called cache server and transmits the content file requested by the user's terminal. The CDN is not limited to the one published on the Internet, but may be formed on a local network. At this time, the protocol used in the network is not particularly limited. The node 4 or the distribution server 2 or the above-mentioned network system may be connected to the network via a protocol conversion device such as a gateway.

For example, when HTTPS is used, the cache file may be represented by the following URI (Uniform Resource Identifier).
<protocol>://<cache server URL>:<cache server port number>/<encoded access destination URL>
Here, it is assumed that the URL of the file distributed by HTTPS from the content server of a certain CDN is as follows.
https://cdn.example.com/video.mp4
When this is URL-encoded (percent-encoded), the following character string is generated.
https%3A%2F%2Fcdn.example.com%2Fvideo.mp4

It should be noted that the host name and IP address of the cache server may be associated one-to-one, or because there are multiple cache servers, they may be associated one-to-many. In the case of one-to-one correspondence, for example, the requesting user terminal specifies the URL of a nearby cache server (for example, the cache server with the smallest number of hops) among the cache servers that make up the CDN, and accesses the cache server. To do. In the case of one-to-many correspondence, the user terminal requests a DNS (Domain Name System) server for name resolution for the URL of the cache server. For example, suppose the access destination address is as follows.
https://cache.example.net
When the above-mentioned character string is set in this, the following request URL is generated.
https://cache.example.net/https%3A%2F%2Fcdn.example.com%2Fvideo.mp4
Here, when the host name and the IP address of the cache server are associated with each other on a one-to-one basis, the DNS server returns the IP address associated with the cache server. In the case of one-to-many correspondence, the DNS server specifies the area of the connection source based on, for example, the IP address of the user terminal, and returns the IP address of the cache server close to the area. After that, the user terminal tries to connect to the IP address returned by the DNS server. That is, a client-side application such as an internet browser provided in the user terminal connects to cache.example.net by HTTPS and performs encrypted communication between the browser and the cache server. At this time, the server certificate used is cache.example.net. It should be noted that the plurality of cache servers have a certificate of cache.example.net.

On the other hand, the cache server decodes the request URL and generates the original URL (that is, https://cdn.example.com/video.mp4). When the cache server does not hold the cache file, the cache server acquires the file from another device based on the group assigned to the file and the groups assigned to the cache server itself and the neighboring cache servers. When the file is obtained from the content server, an HTTPS session is established with cdn.example.com based on the decoded URL and a request to send video.mp4 is made. At this time, the server certificate used is cdn.example.com. When the cache server receives a requested file from another device, the cache server transfers it to the request source application of the user terminal. According to the above processing, even when passing through the cache server, it is possible to avoid the mismatch of the certificates and to send the file distributed by HTTPS to the user terminal. The correspondence relationship between the host name of the cache server and the IP address, the naming convention of the request URL to the content server for the cache server, etc. are examples, and the present invention is not limited to the above examples.

<Modification 7>
Further, the process of assigning a group to the node 4 is not limited to that shown in the above-described embodiment. For example, if there are multiple groups with the same distance between groups, the groups are selected so that the distance between the groups increases in order from the node with the highest degree. May be assigned. In this way, a group different from that of the adjacent node 4 can be preferentially assigned to a certain node 4, and the groups can be balanced so as to be included in the entire network as evenly as possible.

Also, a plurality of groups may be assigned to one node 4. In particular, when the storage capacities (that is, cache capacities) of the plurality of nodes 4 are different, a plurality of groups can be allocated to the node 4 having a large capacity to efficiently use resources. Also in this case, a group different from that of the adjacent node 4 is preferentially assigned to a certain node 4.

<Modification 8>
In the processing of totaling the number of accesses shown in S21 and S22 of FIG. 9, the distribution server 2 or a management device (not shown) receives the number of accesses from each node 4 according to a predetermined route on the network. Good. For example, a tree-structured network whose root is the distribution server 2 or a management device (not shown) is defined. For example, in the network 1 of FIG. 1, a tree-structured network including the distribution server 2 as a root and all nodes 4 and connected by existing links is separately defined. Then, the number of accesses for each file is transmitted from the leaf toward the root at a predetermined timing. In this case, the number of accesses for each file may be aggregated at the node that is the confluence of the branches. By doing so, it is possible to reduce the amount of communication required for counting and the load of counting in the distribution server 2 or another management device.

The matters described in the above-described embodiment and the modified examples 1 to 8 can be implemented in an appropriate combination.

<Comparative example>
FIG. 14 is a diagram for explaining a procedure for determining cache placement using a genetic algorithm. When the caches are held in a distributed manner in the network as described above, a genetic algorithm can be used as a method for determining the cache placement. Specifically, the following procedure is repeated to search for the optimum cache placement.

(1) Determination of Initial Value First, a fixed number of files, which are randomly determined to cache files in each server, are generated. In this step, for example, 300 cache arrangements are randomly generated. In the example of FIG. 14, three relay nodes that are connected to the distribution server and one request source node that is connected to each relay node are schematically shown. In determining the initial value, files to be cached are randomly determined as shown in cache allocations 1, 2, 3,.... The numbers in the squares shown in the cache arrangements 1, 2, 3,... In FIG. 14 indicate identification information of files to be cached. In addition, three relay nodes show an example in which five files are cached.

また、発生する総通信量Ｔは、以下の式（２）によって求めることができる。

なお、ｓ_ｋはファイルｋのサイズ、ｒ_ｊｋはノードｊがファイルｋを保持しているとき１、それ以外のとき０、ｄ（ｉ，ｊ）はノードｉからノードｊまでのホップ数、ｐ_ｋはファイルｋを要求する頻度である。図１４において、例えばキャッシュ配置１、２、３・・・の右に示す通り、キャッシュ配置１、２について算出された通信量が１００、９０であったものとする。(2) Calculation of communication amount Next, the communication amount is calculated for each cache arrangement. The amount of communication is calculated from the topology, cache arrangement, and access frequency. The communication amount t _ik when the request source node i acquires a file from the node j that holds the closest file k can be _calculated by the following equation (1).

Further, the total communication traffic T to be generated can be obtained by the following equation (2).

Incidentally, _{s k} is the number of hops size of the file _k, 1 when _{r jk} is the node j holds the file k, at other times 0, d (i, j) from node i to node j, p _k is the frequency of requesting the file k. In FIG. 14, for example, as shown on the right of the cache arrangements 1, 2, 3,..., It is assumed that the communication amounts calculated for the cache arrangements 1 and 2 are 100 and 90.

(3) Selection Then, a predetermined number of cache arrangements are selected based on the communication volume. Specifically, the cache placement is selected based on the probability proportional to the reciprocal of the communication amount. That is, the cache arrangement is extracted according to the rule that the smaller the communication amount, the easier the selection. In the example of FIG. 11, the magnitudes of the probabilities that the cache arrangements 1 and 2 are selected are relatively 1/100 and 1/90. In the example of FIG. 14, it is assumed that cache arrangements 1 and 2 are selected.

(4) Crossover Next, a new cache arrangement is generated from the selected cache arrangement. Specifically, first, as shown in the crossover 1 of FIG. 14, in the cache arrangement 1 and the cache arrangement 2, a cache file common to each relay node is extracted. Then, as shown by crossover 2, a file is randomly selected from files that are not common to the original cache arrangements 1 and 2.

(5) Mutation After that, a part of the generated cache arrangement is randomly changed. In this step, all cache files are subject to random change with a predetermined probability such as 0.1%.

(6) Selection Furthermore, from a plurality of cache arrangements generated by repeating steps (3) to (5), a predetermined number with a small communication amount is selected. In this step, for example, 300 types of cache arrangements are selected.

After that, the procedure is returned to the procedure (2) and the processing is repeated. As described above, it is possible to converge to a cache arrangement having a smaller communication amount and obtain a sub-optimal cache arrangement.

FIG. 15 is a graph comparing the amount of communication with an external server with respect to the network according to the present invention, the network according to the comparative example, and the network that performs caching by the LFU without cooperation. The vertical axis of the graph represents the amount of communication with the external server relative to 1 when the cache is not performed. As the network topology, the mesh type network shown in FIG. 1 is adopted. The external server is the distribution server 2 in FIG. Further, in the calculation of the communication amount, each node can hold 10% of the entire file.

FIG. 16 is a graph showing files cached on the network and the number thereof. The horizontal axis of FIG. 16 represents the serial numbers of files sorted in descending order of access frequency, and the vertical axis represents the number of nodes that cache each file. In the network according to the present invention, the setting of tags attached to files may be determined according to the type of files cached on a node and the number of nodes that cache the same file in the cache arrangement required in the above-described comparative example. .. The solid line in FIG. 16 represents the relationship between the file in the sub-optimal cache arrangement obtained by the genetic algorithm (GA) and the number of files cached on the network. Also, the dotted circles represent an example of setting the number of files to be cached, which is determined according to the number of files in the GA in the network formed by the nodes classified into four types of groups according to the present invention. The thick and short broken line represents an example in which files to be cached by the nodes classified into eight types of groups according to the present invention are determined based on the number of files obtained by GA. A thin long broken line represents an example in which files cached by the nodes classified into 16 types of groups according to the present invention are determined according to the number of files obtained by GA. In the present embodiment, the number of nodes on the network that cache the file can be adjusted by changing the group assigned to each file, which is represented by the tag described above. Further, by increasing the number of types of groups, it is possible to improve the accuracy of approximation for the sub-optimal arrangement. As described above, the number of files associated with each category indicating the degree of popularity shown in FIG. 10 approximates a sub-optimal cache arrangement in almost the same network, which is obtained in advance by using a predetermined algorithm. At least one of the group to be assigned to the file and the group to be assigned to the node on the network may be determined so as to have the above value.

The non-cooperative LFU shown on the horizontal axis of the graph of FIG. 15 shows an example of the communication amount when each node independently deletes a cache file with low access frequency. The 4 colors, 8 colors, and 16 colors indicate the communication amounts when the nodes are classified into 4 groups, 8 groups, and 16 groups and the cache arrangement according to the present invention is performed. GA indicates the traffic when the cache placement is determined by the genetic algorithm according to the comparative example. As shown in FIG. 15, according to the cache arrangement according to the present invention, the effective cache capacity increases as the number of groups increases, and in the example of 16 colors, the communication volume with the file distribution server is 95 for the uncoordinated LFU. The amount of communication has been reduced by 3%. On the other hand, although the GA has a lower communication volume, the 16-color example has a communication volume almost equal to that of the suboptimal cache arrangement by GA.

FIG. 17 is a graph showing the amount of communication inside the network for the network according to the present invention, the network for the comparative example, and the network for caching by the LFU without cooperation. The vertical axis represents the amount of communication in the network when the case where the cache is not performed is 1. The horizontal axis is the same as in the example of FIG. For example, in a 16-color network, the amount of communication is reduced by 51% compared to the LFU without coordination. Further, it can be said that the network according to the present invention and the suboptimal cache arrangement by GA have almost the same communication volume.

A general computer requires 10 hours or more of calculation to determine the cache placement by the GA, whereas the cache placement according to the present invention is completed in about several seconds. That is, according to the present invention, the cache arrangement of files can be determined by cyclically tagging the files according to the access frequency of the files. Further, according to the present invention, the setting of grouping for nodes can also be determined as a coloring problem with N colors. The setting of grouping for nodes does not need to be recalculated because there is little change in the configuration in the network composed of cache servers. Further, in the case where each PC is made to function as a cache server in a network composed of PCs (Personal Computers), it is possible to determine a group by assigning a group that does not exist in the vicinity to a PC newly connected to the network. As described above, according to the present invention, it is possible to easily determine a cache placement that is comparable to the suboptimal placement.

<Other>
The present invention includes a computer program that executes the above processing. Furthermore, a computer-readable recording medium in which the program is recorded also belongs to the category of the present invention. With respect to the recording medium in which the program is recorded, the above-described processing can be performed by causing a computer to read and execute the program in the recording medium.

Here, the computer-readable recording medium refers to a recording medium that can store information such as data and programs by electrical, magnetic, optical, mechanical, or chemical action and can be read by a computer. Among such recording media, those removable from the computer include flexible disks, magneto-optical disks, optical disks, magnetic tapes, memory cards and the like. Further, a hard disk drive, a ROM, and the like are examples of the recording medium fixed to the computer.

1 network system 2 distribution server 241: file distribution unit 242: frequency calculation unit 243: setting change unit 244: security control unit 3 file 4 node 441: file request unit 442: transmission relay unit 443: cache holding unit 444: cache transmission unit 445: cache deletion unit 446: security control unit

Claims (14)

A network system including a plurality of node devices for temporarily holding files to be transmitted and received,
The plurality of node devices is one of a plurality of groups for specifying a file to be held by the node device from the node device having a high degree , and exists in the vicinity of the node device in the network topology. The remaining groups that are excluded in order are assigned,
The file is associated with each of the plurality of groups, and is associated with information indicating zero or more of the groups , which is represented by a bit string including a bit indicating that the corresponding group should be retained ,
The node device is
When receiving the file associated with the information indicating the group to which the node device belongs, a cache holding unit that temporarily holds the file as a cache file,
When holding the cache file corresponding to the file requested by another node device, a cache transmission unit that transmits the cache file to the other node device,
Network system including. The network system according to claim 1, wherein there are four or more groups. Further comprising an access management device that calculates the frequency of access to the file,
The network system according to claim 1, wherein the access management device changes information indicating the group associated with the file based on the calculated frequency. The access management device, for a file from which the information indicating the group to be assigned to the file is deleted, notifies the node device to delete the file,
The network system according to claim 3, wherein the node device further includes a cache deletion unit that deletes the cache file based on the notification. The cache holding unit is based on a white list for specifying files that are allowed to be cached, a black list for specifying files that are not allowed to be cached, or a flag indicating whether or not to allow the files to be cached by the file, The network system according to claim 1, wherein it is determined whether to retain the received file. The file is linked from a document in a predetermined hypertext, and information indicating the group to be assigned to the file is assigned based on the number of links passing from a reference document. The network system according to one item. The network system according to any one of claims 1 to 6, wherein the file is divided into a plurality of partial files by a secret sharing method, and each partial file is associated with information indicating a different group. The file is digital content,
The network system according to any one of claims 1 to 7, wherein the node device is a cache server forming a content distribution network. A method of caching a file, which is executed by a network system including a plurality of node devices that temporarily hold files to be transmitted and received,
The plurality of node devices is one of a plurality of groups for specifying a file to be held by the node device from the node device having a high degree , and exists in the vicinity of the node device in the network topology. The remaining groups that are excluded in order are assigned,
The file is associated with each of the plurality of groups, and is associated with information indicating zero or more of the groups , which is represented by a bit string including a bit indicating that the corresponding group should be retained ,
The node device is
When the file associated with the information indicating the group to which the node device belongs is received, a step of temporarily holding the file as a cache file,
When holding the cache file corresponding to the file requested by another node device, transmitting the cache file to the other node device,
How to cache files to run. In a network system including a plurality of node devices for temporarily holding files to be transmitted and received, a file cache program executed by the node device,
The plurality of node devices is one of a plurality of groups for specifying a file to be held by the node device from the node device having a high degree , and exists in the vicinity of the node device in the network topology. The remaining groups are assigned in turn, and the remaining groups are assigned.
The file is associated with each of the plurality of groups, and is associated with information indicating zero or more of the groups , which is represented by a bit string including a bit indicating that the corresponding group should be retained ,
In the node device,
When the file associated with the information indicating the group to which the node device belongs is received, a step of temporarily holding the file as a cache file,
When holding the cache file corresponding to the file requested by another node device, transmitting the cache file to the other node device,
A file caching program that runs One of a plurality of groups for identifying a file to be held by itself, in the network topology, from the node device having a higher order among the plurality of node devices, the groups existing in the vicinity of the node device are sequentially ordered. When the remaining excluded group is assigned and the file associated with the information indicating the group to which it belongs is received, it is connected to a network system including a plurality of node devices that temporarily retain the file as a cache file. Management device,
A frequency calculation unit that counts acquisition requests for the file from the node device and calculates a frequency,
Based on the frequency calculated by the frequency calculation unit, assigns to the file , determines zero or more groups that should hold the file, and indicates the file and the determined group, and corresponds to each of the plurality of groups. A setting changing unit for transmitting information indicating a combination with information represented by a bit string including bits to the node device;
A management device including. The management device according to claim 11, wherein the setting changing unit allocates more groups to files having a higher frequency of receiving an acquisition request. One of a plurality of groups for identifying a file to be held by itself, in the network topology, from the node device having a higher order among the plurality of node devices, the groups existing in the vicinity of the node device are sequentially ordered. When the remaining excluded group is assigned and the file associated with the information indicating the group to which it belongs is received, it is connected to a network system including a plurality of node devices that temporarily retain the file as a cache file. Management device
Counting acquisition requests for the file from the node device, and calculating a frequency,
Assigns to the file based on the frequency calculated by the frequency calculation unit, determines zero or more groups that should hold the file, and indicates the file and the determined group, corresponding to each of the plurality of groups Transmitting information indicating a combination with information represented by a bit string including a bit to the node device,
Management method to perform. One of a plurality of groups for identifying a file to be held by itself, in the network topology, from the node device having a higher order among the plurality of node devices, the groups existing in the vicinity of the node device are sequentially ordered. When the remaining excluded group is assigned and the file associated with the information indicating the group to which it belongs is received, it is connected to a network system including a plurality of node devices that temporarily retain the file as a cache file. Management device,
Counting acquisition requests for the file from the node device, and calculating a frequency,
Assigns to the file based on the frequency calculated by the frequency calculation unit, determines zero or more groups that should hold the file, and indicates the file and the determined group, corresponding to each of the plurality of groups Transmitting information indicating a combination with information represented by a bit string including a bit to the node device,
A management program that executes.

Images