JP2010009448A - Distributed information arrangement system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a distributed information arrangement system, can dynamically change the number of clusters each of which is a group of nodes caching the same data during operation of the system. <P>SOLUTION: Among node devices 1001 on a network provided with caches storing data, one cluster 1004-1 is formed by a plurality of node devices 1001-1 to 1001-3 storing the same data in the caches. The node device 1001-1 which serves as a representative node determines whether a new cluster creation condition is established, and transmits, when it is established, a creation instruction of new cluster together with a copy of the data to the other node device 1001-4 which has not cached the data yet. The node device 1001-4 stores the received copy of the data in the cache of its own node device, and creates a new cluster 1004-2 with the own node device as a representative node. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a distributed information arrangement system for distributing a cache of information on a communication network (hereinafter abbreviated as NW).

  In recent years, the use of information content distribution services on the Internet and peer-to-peer network services in which individual users transmit and exchange information contents have advanced. When distributing information content using a network, the processing capacity of the server device is limited. Therefore, in order to ensure processing performance, one or more caches are provided on the network, and the server device content is temporarily stored. To be held. An example of a technique for distributing content caches on an NW is described in Patent Document 1.

  In the technology described in Patent Document 1 (hereinafter referred to as related technology of the present invention), a server device is connected to a plurality of cache clusters via the Internet, and each cache cluster is connected to a client terminal via an access network. The Each cache cluster is composed of a plurality of caches. When receiving the delivery request from the client terminal, the server device selects an optimum cache from any one of the cache clusters in consideration of the load of each cache, and instructs data delivery. The cache instructed to deliver data delivers the data to the client terminal. The server device centrally manages the increase / decrease in the number of caches in each cache cluster according to the access frequency of the content.

JP 2003-122658 A

  In the related art of the present invention described above, the number of caches in each cache cluster is controlled to increase or decrease in accordance with the access frequency of the content, thereby optimizing the number of caches. However, the number of cache clusters itself is determined in advance by the administrator and is not changed during system operation (see paragraph 25 of Patent Document 1). For this reason, when the administrator overestimates the access demand to the contents and creates more cache clusters than the actual demand, a cache cluster with a low operation rate is generated and resources are wasted. Conversely, if the administrator underestimates the access demand to the content and creates fewer cache clusters than actual demand, the load on each cache cluster increases so that the increase in the number of caches in each cache cluster cannot be handled. This will cause a decrease in response and throughput. However, it is difficult to accurately estimate actual demand before system operation. Also, the access demand may change significantly during system operation.

  The present invention has been proposed in view of such circumstances, and an object thereof is to provide a distributed information arrangement system capable of dynamically changing the number of clusters during system operation.

  The first distributed information arrangement system of the present invention is a distributed information in which a plurality of node devices including a cache for storing data and means for reading the data from the cache and transmitting the data to a terminal device are connected via a network. A placement system comprising a cluster composed of a set of one or more node devices that store the same data in the cache, wherein the cluster includes any one of the one or more node devices included in the set One of them functions as a representative node, and the representative node has a means for determining whether or not a new cluster generation condition is satisfied, and has not yet cached the data when the new cluster generation condition is satisfied. Means for sending a new cluster generation instruction to the node device with a copy of the data, Star said node device receiving the instruction for generating a copy of said data stored in said cache of the node device comprises means for generating a new cluster to the representative node of the own node device.

  A first distributed information arrangement method according to the present invention is the same as a node device on a network including a cache for storing data and means for reading the data from the cache and transmitting the data to a terminal device. A step of forming one cluster by a set of one or more node devices to be stored and a node device functioning as a representative node among the one or more node devices constituting the cluster satisfy a new cluster generation condition. If the new cluster generation condition is satisfied, a new cluster generation instruction is transmitted with a copy of the data to the other node devices that have not yet cached the data. And the node device that has received the new cluster generation instruction sends a copy of the data to its own node device. Serial and stored in the cache, and generating a new cluster to the representative node of the own node device.

  The first node device of the present invention comprises a cache for storing data and means for reading the data from the cache and transmitting it to the terminal device, together with other node devices for storing the same data in the cache A node device that functions as a representative node of the cluster and determines whether or not a new cluster generation condition is satisfied, and when the new cluster generation condition is satisfied, And a means for transmitting a new cluster generation instruction with a copy of the data to the other node device that has not yet cached the data.

  A second node device according to the present invention is a node device comprising a cache for storing data and means for reading the data from the cache and transmitting the data to a terminal device, and is a representative of a cluster for which a new cluster generation condition has been established. When a new cluster generation instruction is received from a node, a copy of the data attached to the received new cluster generation instruction is stored in the cache of the own node apparatus, and the new cluster having the own node apparatus as a representative node Is provided.

  According to the present invention, the representative node of the cluster determines whether or not the new cluster generation condition is satisfied, and when the new cluster generation condition is satisfied, for other node devices that have not yet cached data. The node device that has sent the new cluster generation instruction with the copy of the data and received the new cluster generation instruction stores the copy of the data in the cache of the local node device, and the local node device is the representative node. Since a new cluster is generated, the number of clusters can be dynamically changed during system operation.

  Next, embodiments of the present invention will be described in detail with reference to the drawings.

  Referring to FIG. 19, the distributed information arrangement system according to the first exemplary embodiment of the present invention is the same as a plurality of node devices 1001-1 to 1001-9 among many node devices 1001 distributed on the IP network. The provision target data is stored, and the provision target data is provided from the plurality of node devices 1001-1 to 1001-9 to the user terminals 1002-1 to 1002-4. The provision target data corresponds to a certain type of file or database.

  In the system shown in FIG. 19, there are a total of nine node devices that cache the same data to be provided (for convenience of explanation, data D), of which node device 1001-1 is the original of data D The remaining eight node devices 1001-2 to 1001-9 store the copies. The node devices 1001-1 to 1001-9 each serve as a cache node for the data D. Further, the cache node 1001-1 storing the original of the data D is referred to as the original cache node of the data D in order to distinguish from the cache nodes 1001-2 to 1001-9 storing the copy.

  A total of three of the original cache node 1001-1 and the two cache nodes 1001-2 and 1001-3 form one cluster 1004-1 for the data D. Similarly, three cache nodes 1001-4 to 1001-6 form another cluster 1004-2 for data D, and three cache nodes 1001-7 to 1001-9 Another cluster 1004-3 is formed.

  In each cluster 1004-1 to 1004-3, one of the one or more cache nodes constituting the cluster functions as a representative node. In the cluster 1004-1 including the original cache node 1001-1, the original cache node 1001-1 is the representative node. In the clusters 1004-2 and 1004-3 that do not include the original cache node 1001-1, an arbitrary cache node is the representative node. In the state of FIG. 19, the nodes 1001-4 and 1001-7 are representative nodes.

  Intra-cluster overlay networks 1005-1 to 1005-3 are formed between the cache nodes constituting each cluster 1004-1 to 1004-3. Further, an inter-cluster overlay network 1006 is formed between the representative nodes 1001-1, 1001-4, and 1001-7 of each cluster 1004-1 to 1004-3. The overlay network is a virtual network formed in an upper layer using an IP network link connecting the node devices 1001. Between the cache nodes belonging to the same cluster 1004-1 to 1004-3, the global IP address, load information, and the like of the node devices that are members of the own cluster are exchanged through the intra-cluster overlay networks 1005-1 to 1005-3. . Between the clusters 1004-1 to 1004-3, the global IP address of the representative node of each cluster, the number of node devices serving as members of each cluster, and the like are exchanged through the intercluster overlay network 1006.

  FIG. 19 also shows several DNS (Domain Name Server) servers 1003-1 to 1003-6 connected to the IP network.

  The node device 1001, the user terminal 1002, and the DNS server 1003 can communicate with each other using an existing IP network.

  Next, a schematic operation of the distributed information arrangement system of the present embodiment configured as described above will be described.

  The cache nodes 1001-1 to 1001-9 caching the data D measure the load information of the own node, and load with other cache nodes belonging to the same cluster through the intra-cluster overlay networks 1005-1 to 1005-3. Information is exchanged.

  Each of the cache nodes 1001-1 to 1001-9 has a global IP address that is an identifier that can uniquely identify the node on the network, and a DNS server 1003 of the network to which the node belongs. A correspondence relationship with a URL (Uniform Resource Locator), which is a description method indicating the location of data D used when another node accesses, is registered.

  In FIG. 19, since the DNS server of the network to which the cache node 1001-1 belongs is the DNS server 1003-1, the cache node 1001-1 sends the global IP address of the own node and the URL of the data D to the DNS server 1003-1. Is registered. Similarly, the cache node 1001-2 is for the DNS server 1003-2, the cache node 1001-3 is for the DNS server 1003-3, and the cache node 1001-4 is for the DNS server 1003-4. 1001-5 and 1001-6 are for the DNS server 1003-5, and the cache nodes 1001-7 to 1001-9 are for the DNS server 1003-6. The correspondence is registered.

  As a result, in the DNS server 1003-5, two types of global IP addresses of the cache server 1001-5 and the cache server 1001-6 are registered for the URL of the data D. Also, in the DNS server 1003-6, three types of global IP addresses of cache servers 1001-7 to 1001-9 are registered for the URL of the data D. When a plurality of global IP addresses are registered for the same URL, each DNS server 1003 responds to one of the IP addresses in response to an inquiry about the IP address corresponding to the URL. At this time, preferably, the global IP address to be included in the response to the inquiry about the global IP address corresponding to the URL is determined so that the plurality of global IP addresses are used almost equally.

  Now, it is assumed that any user of the user terminals 1002-1 to 1002-4, for example, the user of the user terminal 1002-1 accesses the data D. When the user inputs the URL of the data D in the URL input text box of the Web browser operating on the user terminal 1002-1, the Web browser transmits an IP address search request to the nearest DNS server 1003-3. (Arrow 1007-11 in FIG. 19). Since the global IP address of the cache node 1001-3 is registered corresponding to the URL of the data D, the DNS server 1003-3 returns the global IP address of the cache node 1001-3 to the user terminal 1002-1 as a response. (Arrow 1007-12). The Web browser of the user terminal 1002-1 transmits a request for requesting acquisition of data D to the cache node 1001-3 using, for example, the HTTP protocol (arrow 1007-13) using the obtained global IP address.

  The cache node 1001-3 determines which cache node will process the request based on the load information of the own node and the other cache nodes 1001-1 and 1001-2 in the same cluster 1004-1. For example, if the cache node 1001-2 is determined as the request processing destination, the cache node 1001-3 transfers the request to the cache node 1001-2 via the intra-cluster overlay network 1005-1 (arrow 1007-14). . The cache node 1001-2 accepts the received request, reads the requested data from the cache of its own node, and transmits a response to the user terminal 1002-1 (arrow 1007-15).

  Next, it is assumed that the user of the user terminal 1002-2 accesses the data D. When the user inputs the URL of data D in the URL input text box of the Web browser operating on the user terminal 1002-2, the Web browser transmits an IP address search request to the nearest DNS server 1003-1. (Arrow 1007-21 in FIG. 19). The DNS server 1003-1 returns the global IP address of the cache node 1001-1 to the user terminal 1002-2 as a response because the global IP address of the cache node 1001-1 is registered corresponding to the URL of the data D. (Arrow 1007-22). The web browser of the user terminal 1002-2 uses the obtained global IP address to transmit a request for obtaining data D to the cache node 1001-1 using, for example, the HTTP protocol (arrow 1007-23).

  The cache node 1001-1 determines which cache node will process the request based on the load information of the own node and the other cache nodes 1001-2 and 1001-3 in the same cluster 1004-1. For example, if the own node 1001-1 is determined as the request processing destination, the cache node 1001-1 accepts the received request, reads the requested data from the own node cache, and responds to the user terminal 1002-2. Is transmitted (arrow 1007-24).

  Next, it is assumed that the user of the user terminal 1002-3 accesses the data D. When the user inputs the URL of data D in the URL input text box of the Web browser operating on the user terminal 1002-3, the Web browser transmits an IP address search request to the nearest DNS server 1003-6. (Arrow 100-31 in FIG. 19). In the DNS server 1003-6, since the global IP addresses of the plurality of cache nodes 1001-7 to 1001-9 are registered corresponding to the URL of the data D, one of them, for example, the global IP address of the cache node 1001-9 is registered. The address is returned to the user terminal 1002-3 as a response (arrow 1007-32). The Web browser of the user terminal 1002-3 transmits a request for requesting acquisition of data D to the cache node 1001-9 using, for example, the HTTP protocol (arrow 1007-33) using the obtained global IP address.

  The cache node 1001-9 determines which cache node will process the request based on the load information of the own node and the other cache nodes 1001-7 and 1001-8 in the same cluster 1004-3. For example, if the cache node 1001-8 is determined as the request processing destination, the cache node 1001-9 transfers the request to the cache node 1001-8 via the intra-cluster overlay network 1005-3 (arrow 1007-34). . The cache node 1001-8 accepts the received request, reads the requested data from the cache of its own node, and transmits a response to the user terminal 1002-3 (arrow 1007-35).

  Next, it is assumed that the user of the user terminal 1002-4 accesses the data D. When the user inputs the URL of data D in the URL input text box of the Web browser operating on the user terminal 1002-4, the Web browser transmits an IP address search request to the nearest DNS server 1003-6. (Arrow 1007-41 in FIG. 19). In the DNS server 1003-6, since the global IP addresses of the plurality of cache nodes 1001-7 to 1001-9 are registered corresponding to the URL of the data D, any one of them, for example, the cache node not selected most recently The global IP address 1001-7 is selected and returned to the user terminal 1002-4 as a response (arrow 1007-42). The Web browser of the user terminal 1002-4 transmits a request for requesting acquisition of data D to the cache node 1001-7 using, for example, the HTTP protocol (arrow 1007-43) using the obtained global IP address.

  The cache node 1001-7 determines which cache node will process the request based on the load information of its own node and other cache nodes 1001-8 and 1001-9 in the same cluster 1004-3. For example, if the local cache node 1001-7 is determined as the request processing destination, the cache node 1001-7 accepts the request, reads the requested data from the cache of the local node, and sends a response to the user terminal 1002-4. Transmit (arrow 1007-44).

  As described above, in the distributed information arrangement system according to the present embodiment, when each cache node 1001-1 to 1001-9 of data D receives a request for data D from a user terminal, the cluster 1004- to which the own node belongs. 1 has a function of determining which cache node in 1-1004-3 should process the request based on the load information of each node. Therefore, the load related to access from the user terminal to the same provided data is determined for each cluster. Every 1004-1 to 1004-3 can be distributed among one or more cache nodes included therein.

  In addition, since each cache node is equipped with a function for determining a node for processing a request, it is possible to distribute the load of the processing itself for determining a node for processing a request.

  Further, each of the cache nodes 1001-1 to 1001-9 is different from each other in order to register the correspondence relationship between the global IP address of the own node and the URL of the data D with respect to the DNS server 1003 of the network to which the own node belongs. When a plurality of user terminals belonging to the network access the DNS server 1003 and acquire the global IP address of the cache node corresponding to the URL of the data D, the global IP addresses of different cache nodes are acquired. As a result, cache nodes from which a plurality of user terminals issue data D requests can be distributed.

  Further, in the DNS server 1003 in which the global IP addresses of a plurality of cache nodes are registered corresponding to the URL of the same data D, the global IP corresponding to the URL is used so that the plurality of global IP addresses are used almost evenly. If a global IP address to be included in a response to an address inquiry is determined, cache nodes from which a plurality of user terminals belonging to the same network issue data D requests can be distributed.

  In the system shown in FIG. 19, the number of cache nodes existing in each cluster 1004-1 to 1004-3 is three. However, the number of cache nodes in this cluster is not fixed, and increases or decreases according to the demand for data D. The increase / decrease of the number of cache nodes is performed for each cluster under the control of the representative node of the cluster.

  In the system shown in FIG. 19, the number of clusters related to data D is three. However, the number of clusters related to the data D is not fixed, and increases or decreases according to the demand for the data D. Specifically, in the initial state, there is only one cluster 1004-1 including only the original cache node 1001-1. When the demand for the data D increases, the original cache node 1001-1 that is the representative node generates new cache nodes 1001-2 and 1001-3 in the cluster 1004-1. When the demand further increases and a predetermined condition such as exceeding the maximum value of the number of cache nodes in one cluster is satisfied, the representative node of the cluster 1004-1 is another node that has not yet cached data D, For example, a new cluster generation instruction is transmitted to node 1001-4 with a copy of data D. The node 1001-4 stores a copy of the data D in the cache of the own node, and generates a new cluster 1004-2 with the own node as a representative node. Thereafter, when the new cluster generation condition is satisfied again, a new cluster 1004-3 is generated under the control of the representative node of the cluster 1004-1 or the cluster 1004-2.

  On the other hand, when the demand for data D declines, the number of cache nodes in each cluster 1004-1 to 1004-3 decreases accordingly. For example, when the number of cache nodes in the cluster 1004-3 becomes 0, the cluster 1004-3 disappears and becomes two, the cluster 1004-1 and the cluster 1004-2. Similarly, when the number of cache nodes in the cluster 1004-2 becomes 0, the cluster 1004-2 disappears and only the cluster 1004-1 remains.

  As described above, in the distributed information arrangement system according to the present embodiment, not only the number of cache nodes but also the number of clusters itself is increased or decreased according to the demand for data D.

  Next, embodiments of the present invention will be described in detail with reference to the drawings.

  This embodiment is a demand-dependent prosperity and decline for the distributed cache allocation problem of information existing on the NW, which determines the amount and location of the allocation autonomously in accordance with the demand for information, that is, the number of accesses. And a shared information arrangement system (hereinafter referred to as a unit distributed information arrangement system).

  For example, slime molds live as unicellular organisms in places where there is a lot of food due to a large amount of food, while they gather together to become multicellular organisms in places where there is little food and move in search of places where there is a lot of food. The egg medakas, Noso Blanchius and Synolebias, live in a savanna (grass) where the rainy season and the dry season alternate during the year. It grows and lays eggs in the soil, and when the dry season comes and the pond dries up, the parents die, and the eggs laid in the soil are dry enough to leave a little dampness until the next rainy season comes Wait patiently in the soil at the bottom of the pond. In this way, some creatures change their forms and adapt to the environment. In addition, organisms are known to maintain species prosperity by controlling the amount of individuals in a density-dependent manner, and this dynamics is generally described by logistic equations. Furthermore, there are helper genes in organisms, and it is also known to maintain the prosperity of species by taking altruistic social behavior at the individual level, which is selfish when viewed from the whole species.

  By having the characteristics as described above, organisms have created ecosystems that are robust against environmental changes. The present embodiment is realized by replacing the “species” in the above description with “information” in the NW, and realizes an adaptive function of these species on the communication NW. Therefore, the purpose of the technique is different from that of the conventional technique related to the distributed cache arrangement. In other words, the purpose of this embodiment is to determine the amount and quality of information cache allocation autonomously and distributedly by an altruistic and collaborative mechanism in accordance with the number of accesses to information on the communication NW. In addition to realizing a distributed cache arrangement of information, it is possible to realize efficient reproduction of information that has been once discarded and no longer accessed.

[Outline of this embodiment]
As a basic mechanism, “information necessity” is defined by a numerical value, and this is assigned to information as an information property. If there is an access to the information cache source, the “necessity of information” increases for every access, and decreases every fixed time. The frequency of access to information, in other words, “freshness of information” is determined by the decrease in the “necessity of information” at regular intervals. In the following, means and actions will be described for cases where the number of accesses to information increases and decreases. Hereinafter, the server that is the source of information and the server that caches the information are referred to as “cache nodes”.

(1) When the number of accesses to information increases Step 1:
The information and the “information necessity” of a certain value are arranged in the first information source cache node. Hereinafter, this cache node is referred to as an “original cache node”. At the same time, a lower-level cluster having itself as a representative cache node is generated, and the IP address of the cache node and “numerical value of information” are registered in the lower-level cluster map for the lower-level cluster. The lower level cluster map for the lower level cluster is distributedly owned by the cache nodes in each lower level cluster, and the cache node communicates with the lower level cluster NW by overlaying with multiple cache nodes registered in the lower level cluster map for the lower level cluster. I do.

  Furthermore, the original node registers its own global IP address to the DNS server of the NW to which it belongs and a URL that is a description method that indicates the location of cache information used when other nodes in the NW access.

Here, means for accessing general cache information on the Internet will be described. In normal Internet communication, a general node B that wants to access information accesses a server that caches information by either of the following methods (a) and (b).
(A) When the node B knows the URL of the cache node but does not know the IP address, the URL is designated and the DNS server is inquired about the IP address. If the DNS server knows the correspondence between the URL and the IP address, the DNS server transmits the IP address to the Node B using normal communication on the Internet. If the DNS server does not know the correspondence between the URL and the IP address, the DNS server makes an inquiry to another DNS server, and finally the DNS server that knows the correspondence between the URL and the IP address is a node. Tell B the IP address.
(B) If Node B knows the IP address, it can access the cache node directly.

  Hereinafter, the description will be continued on the assumption of the case (a).

Step 2:
Every time there is access to information, the “number of“ information necessity ”” of the cache node increases.

Step 3:
When access increases and the “number of“ information need ”” exceeds a certain threshold 1, the average value of the “number of“ information need ”” for a certain period of time is taken into account, taking into account fluctuations in the number. get. If the average value of “Numerical value of information” for a certain time exceeds a certain threshold value 1, the representative cache node caches information in a large-order cache node at a certain physical or logical NW distance. There is a random consultation with a normal Internet communication method, and the cache node with the consent is made to cache. Here, when a plurality of approvals are obtained, all of them, or a cache node having a large degree among them, is allowed to cache. The newly added cache node sends to the representative cache node in the lower cluster a “entry message” for recognizing that the newly added cache node belongs to the lower cluster by a communication method in the normal Internet. .

  The representative cache node that has made a request to the newly added cache node sends a message indicating that the “entry message” has been received to the newly added cache node by communication within the lower cluster, and at the same time is necessary for the lower cluster map. Request information. Upon receipt of the request, the newly added cache node sends request information regarding its own cache node to the representative cache node that has requested the cache node by communication within the lower cluster.

  The representative cache node that requested the additional cache node updates the lower cluster map based on the request information received from the new additional cache node, and the updated lower cluster map is assigned to the representative cache node by communication within the lower cluster. To other cache nodes in the subordinate cluster. Each cache node updates its own lower-level cluster map for the lower-level cluster, and sends a registration completion message to the cache node that has requested the cache node by communication within the lower-level cluster. Note that the addition of a new cache node to the lower cluster is performed after waiting for a certain time in consideration of the variation in the number of accesses. The access destination from the user at this time is a representative cache node.

Step 4:
A plurality of cache nodes including the representative cache node existing in the lower cluster perform mutual “survival” and ““ necessity of information ”by performing communication within the lower cluster at regular time intervals or at indefinite time intervals. Exchange information such as This updates and synchronizes the lower cluster map for the lower cluster of the cache node in the lower cluster. Here, in the lower cluster communication, when an event such as a new cache node is added or a failure occurs in the cache node or a link between the cache nodes, the event is triggered.

Step 5:
When at least one other cache node other than the representative cache node is added to the lower level cluster, they autonomously load balance. That is, access is sequentially transferred from a cache node having a smaller value according to “numerical value of“ information necessity ”” in the same lower cluster or a calculation result using the same in the lower cluster communication.

Step 6:
Each cluster has a cluster area identified by a circle whose radius is a certain physical distance or logical distance from the center of the representative cache node or cluster, and the maximum number of cache nodes that can be installed in this cluster area. Is predetermined. In other words, the maximum number of cache nodes that can be installed in the cluster area is the cache node density of the cluster. The cache node density may be expressed by “number of cache nodes / NW radius” or may be expressed by “number of cache nodes / number of nodes existing in NW radius”.

Step 7:
When it is detected that the representative cache node of the lower cluster “accesses from a distance exceeding the NW radius increases” or “approaching the maximum number of cache nodes that can be placed in the own lower cluster calculated in step 6 above”, Cannot cache the information in a cache node of a large degree that exists in a place exceeding the NW radius, or randomly consult with a normal Internet communication method, and obtain a consent from a normal Internet communication method Have them cache. Here, when a plurality of approvals are obtained, all of them, or a cache node having a large degree among them, is allowed to cache. In addition, the location exceeding the NW radius exists in the 360 degree direction around the own cluster, but if the location exceeds the NW radius is biased to the NW domain, priority is given to a cache node having a large degree near the NW domain. Select as a candidate. Note that this new cache node is added after waiting for a certain time in consideration of the variation in the number of accesses.

  The representative cache node that has requested the cache node creates an upper cluster composed only of the representative cache nodes including the own cache node, and creates an upper cluster map for the upper cluster.

  The newly added cache node newly forms a subordinate cluster having itself as a representative cache node, and creates a subordinate cluster map for the subordinate cluster. Here, the overlay NW in the upper cluster by the representative cache nodes is constructed in the form of the overlay NW on the NW in the lower cluster.

  Also, the newly added cache node has the representative cache node that received the cache node request recognize that the newly added cache node belongs to the upper cluster consisting of only the representative cache node. An “entry message” is sent by a communication method in a normal Internet.

  The representative cache node that has made a request to the cache node sends a message indicating that the “entry message” has been received to the newly added representative cache node by communication within the upper cluster, and at the same time, information necessary for the upper cluster map. Request. Upon receiving the request, the newly added cache node sends request information regarding its own cache node to the representative cache node that has requested the cache node by communication within the upper cluster.

  The representative cache node that made the cache node request updates the upper cluster map based on the request information received from the new representative cache node, and the other representative cache nodes in the upper cluster configured only by the representative cache node Sent by upper cluster communication. Each cache node updates its own upper cluster map for the upper cluster, and sends a registration completion message to the representative cache node that has requested the cache node by communication within the upper cluster.

  In principle, the lower-level clusters communicate with each other by performing overlay communication between the cache nodes, which are representative cache nodes, such as the number of cache nodes in each lower-level cluster at regular time intervals or at indefinite time intervals. Check survival and situation.

(2) When the number of accesses to information decreases Step 1:
When access to a certain cache node decreases and the value of “Necessary information” falls below a certain threshold, the numerical value of “Necessity of information” for a certain time from that point is taken into account when the numerical value is considered. ”Is obtained. If the average value of “Numerical value of information” for a certain time falls below a certain threshold, as long as there are other cache nodes in the lower cluster to which the own cache node belongs, other cache nodes in those lower clusters A withdrawal message is sent to all through lower cluster communication. The other cache node in the lower cluster that has received the withdrawal message erases information other than the IP address of the withdrawal cache node from the lower cluster map in the lower cluster of its own cache node, and a confirmation message that “the withdrawal message has arrived” Is transmitted to the withdrawal cache node by communication within the lower cluster. The withdrawal cache node receives the confirmation message by communication within the lower cluster from all other cache nodes in the lower cluster, and then deletes the cache data held by the own cache node, and the URL of the information and its own A deletion request is issued to the DNS server that has registered the IP address, and the subordinate cluster is exited.

  Here, if the representative cache node leaves, it will be as follows. When the access to the representative cache node decreases and the “value of“ information need ”” falls below a certain threshold, the “value of“ information need ”” for a certain time from that point is taken into account. ”Is obtained. If the average value of the “Need for Information” for a certain time falls below a certain threshold, it is the most “Need for Information” among the other cache nodes in the lower cluster to which the own cache node belongs. The representative cache node that communicates with other lower level clusters is transferred to the small cache node, and all other cache nodes in the lower level cluster to which the own cache node belongs are communicated with the lower level cluster and other lower level nodes. The representative cache node of the cluster is notified by “intra-cluster communication” that “another cache node takes over the representative cache node and leaves itself”. All other cache nodes in the lower level cluster that have received this message delete the information of the representative cache node from the lower level cluster map for lower level clusters, and set the “cache node that inherited the representative cache node” as the representative cache node. Is set in the lower cluster map for the lower cluster. In addition, for the "original representative cache node" and the "cache node that inherited the representative cache node", a message confirming that "new representative cache node has been registered in the lower cluster map for lower cluster" through lower cluster communication Tell as.

  Similarly, the representative cache node of another lower cluster deletes the information of the representative cache node from the upper cluster map for the upper cluster NW, and uses the new representative cache node as the representative cache node of the lower cluster, and the upper cluster map for the upper cluster NW. Set to. Further, the representative cache node of another lower cluster transmits a confirmation message to the “cache node that inherited the representative cache node” and the “original representative cache node” by communication within the upper cluster.

  After receiving confirmation messages from all the cache nodes in the lower level cluster to which the own cache node belongs and from all the representative cache nodes in other lower level clusters via lower level cluster communication and higher level cluster communication, The “representative cache node” deletes the cache data held by its own cache node, issues a deletion request to the DNS server that registered the URL of the information and its own IP address, and exits from the lower cluster.

Step 2:
The cache node that finally remains in a subordinate cluster is inevitably the representative cache node of the subordinate cluster. However, even if the representative cache node falls below a certain threshold, the numerical value changes. Considering this, the average value of the “numerical value of“ information need ”” for a certain period of time from that point is acquired. If the average value of the “value of information” for a certain time falls below a certain threshold, the representative cache node sends a withdrawal message to all the representative cache nodes in other lower clusters by communication within the upper cluster. . At the same time, the withdrawal representative cache node looks at the upper cluster map for the upper cluster NW of its own cache node, and among all the representative cache nodes of the other lower clusters, the “number of information necessity” per cache node is The lower cluster map for the lower cluster to which the own cache node belongs is transmitted by communication within the upper cluster to the representative cache node having the smallest cache node density and the largest value.

  All the representative cache nodes in the other lower-level clusters have sent a message confirming that the “representative cache node is leaving” to the representative cache node that has sent the withdrawal message to the “representative cache node” via upper cluster communication. Later, information other than the IP address related to the representative cache node from the destination of the periodic message is deleted from the upper cluster map for the upper cluster NW. Here, the representative cache node that has received the lower cluster map for the lower cluster from the withdrawal representative cache node transmits a confirmation message “having received the lower cluster map” to the withdrawal representative cache node by communication within the upper cluster.

  The withdrawal representative cache node receives a message confirming that “representative cache node is removed” from all the representative cache nodes of the other lower clusters through communication within the upper cluster, and the lower representative cluster for the lower cluster from the withdrawal representative cache node. After receiving a message confirming that “the lower cluster map has been received” from the representative cache node that received the map by communication within the upper cluster, the cache data held by the own cache node is deleted, and the URL of the information and its own URL are deleted. A deletion request is issued to the DNS server that has registered the IP address, and the host cluster NW is exited.

Step 3:
In the subordinate cluster that finally remains other than the subordinate cluster including the original cache node, even the finally remaining cache node (hereinafter referred to as the terminal cache node) falls below a certain threshold value Then, taking the fluctuation of the numerical value, the average value of “the numerical value of“ the degree of necessity of information ”” for a certain time from that time point is acquired. When the average value of the “value of information” for a certain time falls below a certain threshold, the terminal cache node holds that “I am the cache node of the final data” for the original cache node. "Cache data" and "Lower level cluster map for lower level clusters of other lower level clusters that have left in the past" may be left as they are on the own cache node. 2) Some other form (for example, as a link destination) Etc.), or 3) “desired information” that the user wants to be deleted, and is communicated by intra-cluster communication.

Step 4:
Based on the “desired information” from the terminal cache node, the original cache node can place the information 1) as it is, or 2) have it placed in some other form (for example, as a link destination), Or, 3) Consider whether or not to delete, and communicate the result to the weekend cache node by communication within the upper cluster. Here, the reason why the terminal cache node continues to cache information is that when the data is needed again, it is estimated that there are many cases from the periphery of the cache node that has been accessed to the end.

Step 5:
If you want the information to be deleted, I would like the other cache nodes in the last remaining lower cluster to leave the information 1) as is 2) Some other form (for example, as a link destination) I tell you that I want you to put it in the normal Internet communication means.

Step 6:
The terminal cache node determines the information management form according to the answer from the original cache node. That is, the information is either 1) placed as it is, 2) placed in some other form (for example, as a link destination), or 3) deleted.

Step 7:
Received a message from the original cache node 1) I want you to leave it as it is, 2) I want you to put it in some other form (for example, as a link destination), etc. The other cache nodes that were in the lower level cluster may place information on the original cache node 1) as it is in the local cache node 2) in some other form (for example, as a link destination) Or 3) “desired information” that the user does not want to place is transmitted by a normal Internet communication means.

Step 8:
The original cache node that has received the “desired information” by the normal Internet communication means may place the information in the cache node as it is. 2) Some other form (for example, as a link destination) Etc.), the cache node and the upper cluster are configured in the priority order of 1) and 2), and the information to be cached is transmitted by communication within the upper cluster.

Step 9:
If the original cache node that received the “desired information” has only the answer that it does not want to place 3) in it, the information is sent to the other cache nodes in the subordinate cluster deleted second from the end. I want you to leave it as it is, 2) I want you to put it in some other form (for example, as a link destination, etc.) and tell it by the usual Internet communication means.

Step 10:
Thereafter, steps 7 to 9 are repeated.

Step 11:
If there is no cache node that caches other than the original cache node, the original cache node keeps the information stored in its own cache node.

(3) Explanation of operation when access is increased again In steps 7 to 11 of “when the number of accesses to information decreases”, depending on which cache node holds information to be finally cached Only the first step is different.

(3-1) When only the original cache node holds information to be cached in this case In this case, step 3 of “(1) When the number of accesses to information increases” is the first step. Become.

(3-2) When the original cache node and the terminal cache node hold information to be cached finally In this case, among the original cache node and the terminal cache node, in the cache node that has received access, “( Step 3 of “1) When the number of accesses to information increases” is the first step. When access is increased at the terminal cache node, and it becomes necessary to add a cache node exceeding the threshold 1 shown in “(1) When the number of accesses to information increases”, “(1) In the case where the number of accesses increases, after waiting for a certain period of time in accordance with the procedure shown in Step 3, the terminal cache node remains in the lower cluster map in the lower cluster held by itself in the lower cluster. Whether or not the cache node that has been a cache node in the past can cache the information is examined by a normal Internet communication means. If consent is not obtained from either of them, a normal Internet communication means determines whether information can be randomly cached by a cache node having a large degree among cache nodes having a certain number of hops. Consult with. If consent is not obtained from a cache node within a certain number of hops, the terminal cache node increments the number of search hops by one and searches for a cache node that can obtain consent.

  If consent is received from any cache node in the past by a cache node in the lower cluster or a search based on the number of hops, the description will be given after step 3 of “(1) When the number of accesses to information increases”. The cache nodes are increased according to the procedure.

  As described above, the amount of information cache allocation and quality are determined autonomously and distributed by an altruistic and cooperative mechanism in accordance with the number of accesses to information on the communication NW. In addition to realizing a distributed cache arrangement, it is possible to efficiently reproduce information that has been once discarded and no longer accessed.

[Description of configuration]
Next, the configuration of the present embodiment will be described in detail.

  Referring to FIG. 1, an example of a change in “numerical value of“ information necessity ”” in the distributed information arrangement system of the present embodiment is shown. Each time there is an access, the “numerical value of information” increases by a certain amount. The “numerical value of“ information requirement ”” decreases by a certain amount every time a certain time elapses.

  Referring to FIG. 2, an example of how a new cache node is added in the distributed information arrangement system of this embodiment is shown. Referring to FIG. 3, there is shown an example of a flowchart for adding a new cache node in the distributed information arrangement system of this embodiment. Referring to FIG. 4, an example of a lower cluster map for a lower cluster held by each cache node in the distributed information arrangement system of the present embodiment is shown.

  As shown in FIG. 4, the contents of the lower-level cluster map for the lower-level cluster include the DNS registration name, global IP address, “information necessity” numerical value, and “information necessity” numerical value of the cache node. The primary differential value and the secondary differential value of the numerical value of “Necessity of information”. The first-order differential value of the numerical value of “information necessity” is an increase / decrease amount of the numerical value of “information necessity” per unit time. The second derivative of the value of “information necessity” is the rate of increase / decrease of the value of “information need” per unit time.

  Here, for overlay communication in the lower cluster NW2, a local IP address that can be used only in the lower cluster NW2 is assigned to each cache node in the lower cluster NW2, and the lower cluster NW2 using this local IP address is assigned. The overlay communication may be performed. In this case, the contents of the lower cluster map for the lower cluster may include a local IP address added locally in the lower cluster 2 other than that shown in FIG.

  The contents of the lower cluster map for the lower cluster may be obtained by deleting the second derivative value of the “information necessity” value from the contents of FIG. In addition, the contents of the lower cluster map for the lower cluster may include a variable representing a state of change of the numerical value of “information necessity” with time other than that shown in FIG.

  The original node is a description indicating the global IP address, which is an identifier that can uniquely identify the DNS server of the NW to which the original node belongs, and the location of cache information used when other nodes in the NW access A URL that is a method is registered (one-dot chain line from the original node 1 in FIG. 2a or 2b).

  Information and a certain value (initial value) “necessity of information” are arranged in the original cache node 1. At the same time, the lower cluster 2 having itself as a representative cache node is generated, and the IP address of the original cache node 1 and the “numerical value of information” are registered in the lower cluster map 4-2 for the lower cluster. The lower cluster map 4-2 for the lower cluster is distributedly owned by the cache nodes in each lower cluster, and the cache node is composed of a plurality of cache nodes and overlays registered in the lower cluster map for the lower cluster. Perform internal communications. As shown in FIG. 1, every time information is accessed, the “numerical value of“ information necessity ”” increases.

  In FIG. 2, for the purpose of easy understanding, the access is represented by ant 3, and the increment of “numerical value of“ information need ”” due to the access is represented by the package A carried by ant 3. When access increases and the “value of“ information need ”” exceeds a certain threshold Θ1, the average value of “number of“ information need ”” for a certain time from that point is taken into account get. When the average value of the “value of information” for a certain time exceeds a certain threshold Θ1, the representative cache node caches information in a large-order cache node at a certain physical or logical NW distance. A random consultation is performed to ask the cache node 6 that has received consent to cache. Here, when a plurality of approvals are obtained, all of them, or a cache node having a large degree among them, caches them. The number of hops is used as “a certain physical or logical NW distance”. Here, a certain RTT value may be used as the “certain physical or logical NW distance”.

  The newly added cache node 6 registers the URL of the information and its own IP address in the DNS server of the NW domain to which it belongs. As a result, (a) if the NW domain to which it belongs is different from the original cache node (or representative cache node), the URL of the information and its own IP address will be sent to a DNS server different from the original cache node (or representative cache node). It is registered (one-dot chain line from the cache server 6 in FIG. 2a). On the other hand, if the NW domain to which (b) own belongs is the same as the original cache node (or representative cache node), the URL of the information and its own IP address are registered in the same DNS server (from the cache server 6 in FIG. 2b). The one-dot chain line that comes out).

  The newly added cache node 6 sends an “entry message” to the representative cache node 1 in the lower cluster 2 for recognizing that the cache node 6 belongs to the lower cluster 2 by a communication means on a normal Internet. send. The representative cache node 1 which has made a request to the cache node 6 sends a message indicating that the “entry message” has been received to the newly added cache node 6 by communication within the lower cluster 2, and at the same time, the lower cluster map 4 -2 requests the necessary information. The newly added cache node 6 that has received the request sends request information regarding its own cache node 6 to the representative cache node 1 that has requested the cache node 6 by communication within the lower cluster 2.

  The representative cache node 1 that made the cache node request updates the lower cluster map 4-2 based on the request information received from the new representative cache node 6, and uses the updated lower cluster map 4-2 as the lower cluster 2. The data is sent to another cache node in the lower cluster to which the representative cache node 1 belongs by internal communication. Each cache node updates its own lower level cluster map for the lower level cluster, and sends a registration completion message to the representative cache node that has requested the cache node by communication within the lower level cluster.

  Referring to FIG. 5, an example of communication between cache nodes in the same lower cluster in the distributed information arrangement system of this embodiment is shown. The plurality of cache nodes 6 and 8 including the representative cache node 1 existing in the lower cluster 2 perform communication within the lower cluster at regular time intervals or at indefinite time intervals, thereby allowing each other's “survival” and “ Exchange information such as “the value of information”. As a result, the lower cluster map for the lower cluster 2 of all the cache nodes in the lower cluster 2 is updated and synchronized. Here, in the lower cluster communication, when an event such as a new cache node is added or a failure occurs in the cache node or a link between the cache nodes, the event is triggered.

  Referring to FIGS. 6A and 6B, one example of how access loads are distributed in the same lower cluster in the distributed information arrangement system of this embodiment is shown. Referring to FIG. 7A and FIG. 7B, one example of a flowchart of access load distribution in the distributed information arrangement system of this embodiment is shown.

  When at least one other cache node other than the representative cache node is added to the lower level cluster, they autonomously load balance. In other words, the “number of information necessity” in the lower cluster 2 by the lower cluster communication indicates that the cache nodes 1, 6 and 8 hold the lower cluster map for the lower cluster 2 having the same contents. , The access is transferred in order from the cache node having the smallest “number of“ information necessity ””.

  Referring to FIG. 8, in the distributed information arrangement system according to the present embodiment, by increasing access from a cache node having a certain NW distance or more, a cache node having a higher degree than the NW distance is converted into a cache node and a new subordinate there. An example of how cluster generation is performed is shown. Referring to FIG. 9, in the distributed information arrangement system according to the present embodiment, by increasing access from a cache node having a certain NW distance or more, a cache node having a higher degree than the NW distance is converted into a cache node and a new subordinate there. An example of a flowchart when cluster generation is performed is shown. Referring to FIG. 10, an example of the upper cluster map 18 of the upper cluster NW composed of only representative cache nodes in the distributed information arrangement system of the present embodiment is shown.

  The newly added cache node 11 newly forms a lower cluster 15 having itself as a representative cache node, and creates a lower cluster map 4-15 for the lower cluster 15. Also, the newly added cache node 11 registers the URL of the information and its own IP address in the DNS server 34 in the NW domain to which it belongs. Here, the DNS servers registered by the cache node 11 are all the DNS servers 31, 32, 33 in which the cache nodes in the cluster 2 to which the original cache node belongs respectively register the URL of the information and its own IP address. Is different. Thereafter, the load distribution operation in the representative node of the added lower cluster and the cache node added in the lower cluster will be described with reference to FIGS. 2 (a), 2 (b), 3 (a), and 3 (3). Performed as shown in b). By these methods, autonomous distributed load distribution is realized. Here, the upper inter-cluster overlay NW 17 by the representative cache nodes is constructed on the upper inter-cluster overlay NW of the lower cluster 2 and the lower cluster 15.

  Here, for communication in the upper cluster in the upper cluster NW17, a local IP address that can be used only in the upper cluster NW17 is assigned to the representative cache node in the upper cluster NW17, and this local IP address is used. Communication within the upper cluster in the upper cluster NW 17 may be performed.

  Here, as shown in FIG. 10, the contents of the upper cluster map for the upper cluster include the DNS registration name of the representative cache node, the global IP address, the number of cache nodes in each cluster, and the “necessity of information”. And the first derivative value of the total value of the numerical values of “Necessity of information”. Here, the contents of the upper cluster map for the upper cluster 17 may include a local IP address added locally in the upper cluster 17 other than that shown in FIG. Further, the contents of the upper cluster map for the upper cluster 17 may include a variable representing a state of change of a numerical value of “Necessity of information” with time other than that shown in FIG.

  Each subordinate cluster has a cluster area specified by a circle having an NW radius 10 of a certain physical distance or logical distance from the representative cache node or the center of the cluster, and cache nodes that can be installed in this cluster area. The maximum number is predetermined.

  When the representative cache node 1 of the lower cluster 2 detects that “the number of accesses from a distance exceeding the NW radius 10 increases” or “approaching the calculated maximum number of cache nodes that can be placed in the own lower cluster 2”, the NW A cache that is located in a place where the radius exceeds 10 and that is cached by a large-order cache node or is randomly consulted by a communication means on a normal Internet, and has been approved by a communication means on a normal Internet Have node 11 cache. Here, when a plurality of approvals are obtained, all of them, or a cache node having a large degree among them, is allowed to cache. The addition of the new cache node 11 is performed after waiting for a certain time in consideration of the fluctuation of the number of accesses.

  The representative cache node 1 that has requested the cache node creates an upper cluster NW17 including only the representative cache node 1 including its own cache node 1, and creates an upper cluster map 18-2 for the upper cluster NW17.

  The newly added cache node 11 newly forms a lower cluster 15 having itself as a representative cache node, and creates a lower cluster map 4-15 for the lower cluster 15. Here, the overlay NW 17 in the upper cluster formed by the representative cache nodes is constructed on the overlay NW of the lower cluster 2 and the lower cluster 15. Further, the newly added cache node 11 indicates that the cache node 11 belongs to the upper cluster 17 with respect to the upper cluster NW 17 composed only of the representative cache node, with respect to the representative cache node 1 that has received the cache node request. An “entry message” for recognition is sent by a communication means on a normal Internet. The representative cache node 1 which has made a request to the cache node 11 sends a message indicating that the “entry message” has been received to the newly added cache node 11 by communication within the upper cluster 17, and at the same time, the upper cluster map 18 Request information required for The newly added cache node 11 that has received the request sends request information related to its own cache node 11 to the representative cache node 1 that has made a request to the cache node 11 by communication within the upper cluster 17.

  The representative cache node 1 that made the cache node request updates the upper cluster map 18-11 based on the request information received from the new representative cache node 11, and uses the updated upper cluster map 18-11 as the upper cluster 17. By internal communication, the data is sent to another representative cache node in the lower cluster to which the representative cache node belongs. Each representative cache node updates its own upper cluster map for the upper cluster, and sends a registration completion message to the representative cache node 1 that has requested the cache node by communication within the upper cluster.

  Referring to FIG. 11, the upper cluster level NW17 (upper layer), the lower cluster level NWs 2, 15, 16 (middle layer), and nodes, which are composed of only representative cache nodes in the distributed information arrangement system of this embodiment. An example of the level NW (lowermost layer) relationship is shown.

  In principle, the lower-level clusters are the representative cache nodes, and the cache nodes 1, 11, and 14 send macro information such as the number of cache nodes in the lower-level clusters 2, 15, and 16 at regular time intervals or at indefinite time intervals. Are exchanged by performing overlay communication on the upper cluster NW17 to check the existence and status of each other, and each of the lower clusters 2, 15, 16 held by the representative cache nodes 1, 11, 14 Update macro information.

  Referring to FIG. 12, an example of a state of withdrawal of a cache node in the distributed information arrangement system of this embodiment is shown. Referring to FIG. 13, there is shown an example of a flowchart of withdrawal of a cache node in the distributed information arrangement system of this embodiment. Referring to FIG. 14, there is shown an example of a flowchart of withdrawal of a representative cache node in the distributed information arrangement system of this embodiment.

  Take the lower cluster 15 of FIG. 12 as an example. When the access to the cache node 19 decreases and the “numerical value of“ information ”” falls below a certain threshold value 2, the fluctuation of the numerical value is taken into consideration, and “ Get the average value of "number". If the average value of “the value of“ information necessity ”” after a certain period of time falls below a certain threshold value 2, as long as there are other cache nodes in the lower cluster 15 to which the own cache node 19 belongs, A “withdrawal message” is sent to all other cache nodes by communication within the lower cluster. The other cache node in the lower cluster 15 that has received the withdrawal message deletes information other than the IP address of the withdrawal cache node 19 from the lower cluster map for lower cluster 15 held by its own cache node, A confirmation message “A message has arrived” is transmitted to the withdrawal cache node 19 by communication within the lower cluster. The withdrawal cache node 19 receives the confirmation message from all the other cache nodes in the lower cluster 15, and then deletes the cache data held by the own cache node 19 and the URL of the information and its own IP address. Is deleted from the lower level cluster 15.

  Here, if the representative cache node 11 leaves, it will be as follows. When the access to the representative cache node 11 decreases and the “numerical value of“ information ”” falls below a certain threshold value 2, the ““ necessity of information ”” for a certain period of time from that point is considered in consideration of the fluctuation of the numerical value. The average value of "number". When the average value of the “numerical value of information” for a certain time falls below a certain threshold value 2, among the other cache nodes in the lower cluster 15 to which the own cache node 11 belongs, “ The representative cache node that communicates with the other lower-level clusters 2 and 16 is transferred to the cache node 13 with a small “number”, and all the other cache nodes in the lower-level cluster 15 to which the own cache node 11 belongs and the other lower-level clusters 2 and 16 are notified to the representative cache nodes 1 and 14 by the communication in the lower cluster and the communication in the upper cluster, respectively, that “the other cache node 13 takes over the representative cache node and leaves itself 11”. Receiving the withdrawal message from the representative cache node 11, all the other cache nodes in the lower cluster 15 delete the information of the representative cache node 11 from the lower cluster map for the lower cluster 15 except the IP address, The “cache node 13 that has inherited the representative cache node” is set as the representative cache node in the lower cluster map for the lower cluster 15. Further, a message confirming that “new representative cache node 13 has been registered in lower cluster map 4 for lower cluster 15” is sent to “original representative cache node 11” and “cache node 13 inheriting representative cache node”. As shown in FIG.

  Similarly, each of the representative cache nodes 1 and 14 of the other lower clusters 2 and 16 deletes the information of the representative cache node 11 from the upper cluster maps 18-2 and 18-16 for the upper cluster NW17 held by each of them. Then, the new representative cache node 13 is set as the representative cache node of the lower cluster 15 in the upper cluster maps 18-2 and 18-16 for the upper cluster NW17. Further, the representative cache nodes 1 and 14 of each of the other lower clusters 2 and 16 send confirmation messages to the “cache node 13 that inherited the representative cache node” and the “original representative cache node 11” by communication within the upper cluster. Send.

  Confirmation messages from all the cache nodes in the lower cluster 15 to which the own cache node 11 belongs and from all the representative cache nodes 1 and 14 in the other lower clusters 2 and 16 through lower cluster communication and upper cluster communication, respectively. After receiving this, the “original representative cache node 11” deletes the cache data held by the own cache node 11 and issues a delete request to the DNS server that registered the URL of the information and its own IP address. Exit from cluster 15.

  Referring to FIG. 15, an example of the state of withdrawal of the lower cluster in the distributed information arrangement system of the present embodiment is shown. Referring to FIG. 16, there is shown an example of a flowchart of withdrawal of a lower cluster in the distributed information arrangement system of this embodiment.

  A description will be given by taking the lower cluster 16 as an example. The cache node 14 finally remaining in the lower cluster 16 is inevitably a representative cache node of the lower cluster. However, even if the representative cache node 14 falls below a certain threshold value, the numerical value of Taking into account fluctuations, an average value of “numerical value of“ information necessity ”” for a certain period of time from that point is acquired. When the average value of the “numerical value of information” for a certain time falls below a certain threshold, the representative cache node 14 sends a withdrawal message to the representative cache nodes 1 and 11 of all other lower clusters 2 and 15. Is sent by intra-cluster communication. At the same time, the withdrawal representative cache node 14 looks at the lower cluster map 18-16 for the upper cluster NW17 held by the own cache node 14, and among the representative cache nodes 1 and 11 of all the other lower clusters 2 and 15, The representative cache node 11 having the smallest “number of“ information necessity ”” per cache node and the largest cache node density in the lower cluster is assigned to the lower cluster 16 to which the own cache node 14 belongs. The lower cluster map 4-16 is transmitted by upper cluster communication.

  The representative cache nodes 1 and 11 of all the other lower clusters 2 and 15 send a message confirming that “representative cache node 14 is leaving” to the representative cache node 14 that has transmitted the withdrawal message “representative cache node 14”. After transmitting to the destination by communication within the upper cluster, information other than the IP address related to the representative cache node 14 is deleted from each of the lower cluster maps 18-2 and 18-15 for the upper cluster NW17. Here, the representative cache node 11 that has received the lower cluster map 4-16 for the lower cluster 16 from the withdrawal representative cache node 14 uses the confirmation message “The lower cluster map 4-16 has been received” as a confirmation message. 14 is transmitted by upper cluster communication.

  The withdrawal representative cache node 14 receives a message confirming that “the representative cache node 14 is leaving” from the representative cache nodes 1 and 11 of all the other lower-level clusters 2 and 15, and the lower-level cluster 16 from the withdrawal representative cache node 14. After receiving the message confirming that “the lower cluster map 4-16 has been received” from the representative cache node 11 that has received the internal lower cluster map 4-16, the cache data held by the own cache node 14 is deleted. Then, a deletion request is issued to the DNS server in which the URL of the information and its own IP address are registered, and the host cluster NW 17 is exited.

  Referring to FIG. 17, there is shown an example of the state of withdrawal of the cache node remaining at the end in the distributed information arrangement system of this embodiment. Referring to FIG. 18, there is shown an example of a flowchart of withdrawal of the cache node remaining at the end in the distributed information arrangement system of the present embodiment.

  In the lower cluster 15 that finally remains other than the lower cluster including the original cache node, even if the finally remaining cache node 11 (terminal cache node) falls below a certain threshold value, Taking into account the fluctuation of the numerical value, an average value of “the numerical value of“ the degree of necessity of information ”” for a certain time from that point is acquired. When the average value of the “number of“ information necessity ”” for a certain time falls below a certain threshold value, “the node 11 is the cache node of the final data” and the terminal cache node holds for the original cache node 1 “Cache data” and “lower cluster map for lower cluster of other lower cluster that has left in the past” may be placed in the own cache node 11 as it is, or 2) some other form ( For example, it may be placed as a link destination or the like, or 3) “desired information” that the user wants to be deleted is communicated by intra-cluster communication.

  Based on the “desired information” from the terminal cache node 11, the original cache node 1 receives the information 1) as it is, or 2) places it in some other form (for example, as a link destination). Or 3) to determine whether to delete it, and to notify the terminal cache node 11 of the result by communication within the upper cluster. Here, the reason why the terminal cache node 11 continues to cache the information is that when the data is needed again, it is estimated that there are many cases from the periphery of the cache node that has been accessed to the end.

  When the terminal cache node 11 deletes the information cached by the terminal cache node 11, the original cache node 1 sends the information to the other cache nodes 13 and 19 in the last remaining lower cluster 15. 1) I want you to leave it as it is, 2) I want you to let me put it in some other form (for example, as a link destination, etc.).

  The terminal cache node 11 determines the information management form according to the answer from the original cache node 1. That is, the information is either 1) placed as it is, 2) placed in some other form (for example, as a link destination), or 3) deleted.

  Received a message from the original cache node 1 by 1) I want you to leave it as it is, 2) I want you to put it in some other form (for example, as a link destination, etc.) The remaining cache nodes 13 and 19 in the remaining lower level cluster 15 may leave information on the original cache node 1 as it is for the original cache node 1 and 2) some other form (for example, link) "Preferred information" indicating that the user may place it at the destination, or 3) that he or she does not want to place it, is transmitted by means of ordinary Internet communication means.

  The original cache node 1 that has received the “desired information” may place the information in the cache node 1 as it is, or 2) in some other form (for example, as a link destination). If there is a good answer, 1) and 2) priorities are used to form a higher-level cluster with the cache node, and the information may be left in 1) its own cache node. 2) Some other form (for example, The information to be cached is transmitted by communication within the upper cluster to the cache node that has made a reply that it may be placed as a link destination.

  If the original cache node 1 that has received the “desired information” has only a reply that it does not want to be placed 3), the information is sent to all the cache nodes in the subordinate cluster 16 that is deleted second from the end. 1) I want you to leave it as it is, and 2) I want you to put it in some other form (for example, as a link destination, etc.), and tell it by the communication method on the normal Internet.

  Received a message from the original cache node 1 that the information should be placed 1) as it is, 2) the message that you want to place it in some other form (for example, as a link destination), etc. All the cache nodes 14, 20, and 21 in 16 may store information on the original cache node 1 as it is in 1) its own cache node, 2) some other form (for example, as a link destination) Etc.), or 3) “desired information” that the user does not want to place is transmitted by a normal communication method on the Internet.

  Thereafter, the procedure from paragraph 0125 to paragraph 0128 is repeated.

  If no cache node finally caches other than the original cache node 1, the original cache node 1 keeps the information stored in its own cache node 1.

  As described above, the amount of information cache allocation and quality are determined autonomously and distributed by an altruistic and cooperative mechanism in accordance with the number of accesses to information on the communication NW. In addition to realizing a distributed cache arrangement, it is possible to efficiently reproduce information that has been once discarded and no longer accessed.

[Description of operation]
Next, the operation of the distributed information arrangement system of this embodiment will be described.

  Referring to FIG. 1, an example of a change in “numerical value of“ information necessity ”” in the distributed information arrangement system of the present embodiment is shown. Each time there is an access, the “numerical value of information” increases by a certain amount. The “numerical value of“ information requirement ”” decreases by a certain amount every time a certain time elapses. Note that the “numerical value of“ information necessity ”” that decreases with time may decrease non-linearly. In addition, the “numerical value of information” that increases with each access varies from cache node to cache node depending on the number of interfaces that can receive cache node access and the input buffer size of the interface. May be. Further, the “numerical value of“ information requirement ”” that increases each time access is made may differ for each lower cluster according to the maximum number of cache nodes that can be placed in the own lower cluster. In addition, the “numerical value of information” that decreases over time differs for each cache node depending on the number of interfaces that can receive cache node access and the input buffer size of the interface. Also good. Further, the “numerical value of“ information necessity ”” that decreases with the passage of time may be different for each lower cluster according to the maximum number of cache nodes that can be placed in the own lower cluster.

(1) Description of operation related to increase in cache node Referring to FIGS. 2 (a) and 2 (b), FIGS. 3 (a) and 3 (b), and FIG. 4, cache nodes in the distributed information arrangement system of this embodiment The operation related to the increase will be described.

  Information and a certain value (initial value) “necessity of information” are arranged in the original cache node 1 (S301 in FIG. 3a, S321 in FIG. 3b). The original cache node also has a global IP address that is an identifier that can uniquely identify itself on the DNS server of the NW to which the original cache node belongs, and the location of the cache information that is used when other nodes in the NW have access. Is registered (a one-dot chain line extending from the original cache node 1 in FIG. 2a or 2b). At the same time, the lower cluster 2 having itself as a representative cache node is generated, and the IP address of the original cache node 1 and the “numeric value of“ information necessity ”” are registered in the lower cluster map 4-2 for the lower cluster. . The lower level cluster map 4-2 for the lower level cluster is distributedly owned by the cache nodes in each lower level cluster, and the cache node includes a plurality of cache nodes registered in the lower level cluster map for the lower level cluster and the lower level cluster. Communication within the lower cluster NW2 is performed by overlay communication.

  As shown in FIG. 1, every time information is accessed, the “numerical value of“ information necessity ”” increases.

  In FIGS. 2 (a) and (b), for the purpose of easy understanding, the access is represented by ant 3, and the ant 3 carries the increment of the “value of“ information necessity ”” due to the access. It is represented by luggage A. When access increases and the “value of“ information need ”” exceeds a certain threshold Θ1, the average value of “number of“ information need ”” for a certain time from that point is taken into account Obtain (S302, S303 in FIG. 3a, S322, S323 in FIG. 3b). If the average value of the “value of information” for a certain time exceeds a certain threshold Θ1, the representative cache node is within a certain number of hops, and the cache node with a high degree is cached or random. The cache node 6 having received the consent is allowed to cache (S304 to S307 in FIG. 3a and S324 to S327 in FIG. 3b).

  Here, the threshold value Θ1 is within the range of 1 to 100% of the maximum value of the input buffer size of the interface that can receive the access of the cache node, and the threshold value Θ2 when the cache node described later leaves the lower cluster. Set to a large percentage. Further, the threshold Θ1 may be different for each cache node according to the number of interfaces that can receive access of the cache node, the input buffer size of the interface, and the like. The threshold Θ1 may be different for each lower cluster according to the maximum number of cache nodes that can be placed in the own lower cluster.

  When consent is obtained from a plurality of cache nodes by a communication method in the ordinary Internet, all of them or cache nodes having a higher degree among them are cached (S308 in FIG. 3a, in FIG. 3b). S328). If consent is not obtained from a cache node within a certain number of hops, the original cache node increments the number of search hops by +1 and searches for a cache node that can obtain consent (No in S306 in FIG. 3a, in FIG. 3b). No in S326).

  The newly added cache node 6 registers the URL of its own information and its own IP address in the DNS server of the NW domain to which the node itself belongs (S309 in FIG. 3a, S329 in FIG. 3b, cache server 6 in FIG. 2a). 1 dash line coming out from the cache server 6 in FIG. 2b).

  The newly added cache node 6 sends an “entry message” to the representative cache node 1 in the lower cluster 2 for recognizing that the cache node 6 belongs to the lower cluster 2 by a communication means on a normal Internet. send. The representative cache node 1 which has made a request to the cache node 6 sends a message indicating that the “entry message” has been received to the newly added cache node 6 by communication within the lower cluster 2, and at the same time, the lower cluster map 4 -2 is requested (S310 in FIG. 3a, S330 in FIG. 3b). The newly added cache node 6 that has received the request sends request information regarding its own cache node 6 to the representative cache node 1 that has requested the cache node 6 by communication within the lower cluster 2 (S311 in FIG. 3a, FIG. 3b). S331).

  The representative cache node 1 that made the cache node request updates the lower cluster map 4-2 based on the request information received from the new representative cache node 6, and uses the updated lower cluster map 4-2 as the lower cluster 2. By internal communication, it is sent to another cache node in the lower cluster to which the representative cache node belongs (S312 in FIG. 3a, S332 in FIG. 3b). Each cache node updates its own lower level cluster map for the lower level cluster, and sends a registration completion message to the representative cache node that has requested the cache node by communication within the lower level cluster (S313 in FIG. 3a, FIG. 3b). S333).

  Referring to FIG. 5, an example of communication between cache nodes in the same lower cluster in the distributed information arrangement system of this embodiment is shown. The plurality of cache nodes 6 and 8 including the representative cache node 1 existing in the lower cluster 2 perform mutual “survival” and “[” by performing the communication within the lower cluster at regular time intervals or at indefinite time intervals. Exchange information such as “the value of information”. As a result, the lower cluster map for the lower cluster 2 of all the cache nodes in the lower cluster 2 is updated and synchronized. Here, in the lower cluster communication, when an event such as a new cache node is added or a failure occurs in the cache node or a link between the cache nodes, the event is triggered.

  Here, in order to prevent a loop of the same information, the overlay communication in the lower cluster NW is performed by setting the representative cache node 1 as a route reflector defined in RFC 2796 of IETF (The Internet Engineering Task Force). Communication may be used.

  Further, in order to prevent a loop of the same information here, OSPF (Open Shortest Path Algorithm) defined in RFC 2328 of IETF may be used as a protocol for lower-layer intra-cluster communication.

  In order to prevent a loop of the same information here, IS-IS (Intermediate System-Intermediate System) defined in IETF RFC 1142 or RFC 1195 may be used as a protocol for some overlay communication in the lower cluster NW.

  Furthermore, in order to synchronize the lower cluster map for the lower cluster 2 of all the cache nodes, the information exchanged by the lower cluster communication in the lower cluster 2 may include version information in some format. good.

(2) Description of operation regarding access load distribution in lower cluster Referring to FIG. 6 (a), FIG. 6 (b), FIG. 7 (a) and FIG. An operation related to load distribution of access in a lower cluster in the system will be described.

  When at least one other cache node other than the representative cache node is added to the lower level cluster, they autonomously load balance. In other words, the “numerical value of“ information ”” in the lower cluster 2 by the overlay communication in the lower cluster indicates that the cache nodes 1, 6, 8 have the same lower cluster 2 in the lower cluster 2 for the same internal use. Therefore, the access is transferred in order from the cache node having the smallest “numerical value of“ information necessity ””.

  FIG. 6A shows a case where a plurality of cache nodes in the lower cluster 2 are registered in individual DNS. In this case, the access source nodes 3a, 3b, and 3c access the DNS servers 31, 32, and 33 of the NW domain to which they belong to know the IP address of the cache node of the cached information (FIG. 7a). S701). Here, for simplicity, it is assumed that the DNS servers 31, 32, and 33 know the IP address of the cache node.

  The DNS servers 31, 32, and 33 teach the IP addresses of the cache nodes 1, 6, and 8 to the access source nodes 3a, 3b, and 3c, respectively (S702 in FIG. 7a). As a result, the access source nodes 3a, 3b, and 3c respectively access the cache nodes 1, 6, and 8 (S703 in FIG. 7a).

  The cache nodes 1, 6, and 8 that have received the access look in the cluster map 4-2 of the lower cluster 2 that each holds, and access in order from the cache node with the smallest “number of“ information need ”” Is transferred (S704 in FIG. 7A).

  By these methods, autonomous distributed load distribution is realized.

  FIG. 6B shows a case where a plurality of cache nodes in the lower cluster 2 are registered in the same DNS. In this case, the access source nodes 3a, 3b, and 3c each access the DNS server 31 in order to know the IP address of the cache node of the cached information (S711 in FIG. 7b). Here, for simplicity, it is assumed that the DNS server 31 knows the IP address of the cache node.

  The DNS server 31 selects one of the cache nodes 1, 6, and 8 for each of the access source nodes 3 a, 3 b, and 3 c according to the method implemented by itself, and the IP address of the cache node (S712 in FIG. 7b). As a result, the access source nodes 3a, 3b, and 3c each access the cache node selected by the DNS server (S713 in FIG. 7b).

  Each cache node that has received an access looks at the cluster map 4-2 of the lower cluster 2 that it holds, and transfers the access in order from the cache node with the smallest “numerical value of“ information necessity ”” (see FIG. 7b S714).

  By these methods, autonomous distributed load distribution is realized.

  Here, the order of transferring the access is “the average value of the access time interval” and “the“ necessity of information ”in the“ current “information necessity” numerical value ”of all the cache nodes in the cluster map 4. By calculating the value obtained by adding the product of “the first derivative of the numerical value of”, the predicted value of the future “number of“ information need ”” of each cache node is calculated, and the predicted value is the smallest The order may be from the cache node.

  In addition, the order of transferring access is determined by adding “average value of access time interval” and “information necessity” to the “current value of information need” of all cache nodes in the cluster map 4. Calculate the value obtained by adding the product of the first derivative of the numerical value and the product of the square of the average value of the access time interval and the second derivative of the numerical value of “Necessary information” Thus, the predicted value of the future “numerical value of information” of each cache node may be calculated, and the cache node having the smallest predicted value may be calculated in order.

  Further, the order of transferring the access is increased to the threshold Θ1 for the cache node having the smallest “value of“ information need ””, and then the “number of“ value of information ”” is the highest at that point. You may repeat increasing to threshold value (theta) 1 with respect to few cache nodes.

  Also, the order of transferring the access is to repeat the transfer of the access from the cache node having the smallest “number of“ information necessity ”” in the order of the “number of“ necessity of information ””. Also good.

  Further, the order of transferring the access may be the order in which the number of interfaces that can receive the access of the cache node in the lower cluster 2 is small.

  Further, the order of transferring the access may be the order in which the input buffer size of the interface capable of receiving the access of the cache node in the lower cluster 2 is small.

  Further, the order of transferring access may be different for each lower cluster.

(3) Explanation of operation concerning increase of cluster With reference to FIG. 8, FIG. 9, and FIG. 10, the operation concerning the increase of the cluster in the distributed information arrangement system of the present embodiment will be explained.

  Each subordinate cluster 2 can be installed in a cluster area specified by a circle having an NW radius 10 of a certain physical distance or logical distance from the center of the representative cache node 1 or cluster 2 and the cluster area. The maximum number of cache nodes is predetermined.

  Here, the NW radius 10 of the access source centered on the own cache node 1 or the lower cluster 2 is set based on the number of interfaces that can receive access to all or a part of the cache nodes in the lower cluster 2. May be.

  The NW radius 10 of the access source centered on the own cache node 1 or the lower cluster 2 is set based on the input buffer size of the interface that can receive access of all or a part of the cache nodes in the lower cluster 2. You may do it.

  Also, the NW radius 10 of the access source centering on the own cache node 1 or the lower cluster 2 can receive the access and the number of interfaces that can receive access to all or a part of the cache nodes in the lower cluster 2. It may be set based on both the input buffer size of a simple interface.

  In this way, the representative cache node 1 of the lower cluster 2 is “accessed from a distance exceeding the NW radius 10” or “approaching the calculated maximum number of cache nodes that can be placed in the own lower cluster 2”. Knowing that the information can be cached in a large-order cache node that exists in a place where the NW radius exceeds 10, or is randomly consulted by a normal Internet communication method, and the cache node 11 that has obtained consent is cached. (S901 to S904 in FIG. 9). Here, when a plurality of approvals are obtained, all of them or cache nodes having a large degree among them are cached (S905 in FIG. 9).

  The representative cache node 1 that has requested the cache node creates an upper cluster NW17 composed only of the representative cache node 1 including its own cache node 1, and creates an upper cluster map 18-2 for the upper cluster NW17 ( S906 in FIG. 9).

  The newly added cache node 11 newly forms a lower cluster 15 having itself as a representative cache node, and creates a lower cluster map 4-15 for the lower cluster 15 (S907 in FIG. 9).

  Further, the newly added cache node 11 registers the URL of the information and its own IP address in the DNS server 34 in the NW domain to which it belongs (S907 in FIG. 9). Here, the DNS servers registered by the cache node 11 are all the DNS servers 31, 32, 33 in which the cache nodes in the cluster 2 to which the original cache node belongs respectively register the URL of the information and its own IP address. Is different. Thereafter, the load distribution operation in the representative node of the added lower cluster and the cache node added in the lower cluster will be described with reference to FIGS. 2 (a), 2 (b), 3 (a), and 3 (3). Performed as shown in b). By these methods, autonomous distributed load distribution is realized (S907 in FIG. 9).

  Here, the upper inter-cluster overlay NW 17 by the representative cache nodes is constructed on the intra-cluster overlay NW of the lower cluster 2 and the lower cluster 15 (S907 in FIG. 9).

  Further, the newly added cache node 11 indicates that the cache node 11 belongs to the upper cluster 17 with respect to the upper cluster NW 17 composed only of the representative cache node, with respect to the representative cache node 1 that has received the cache node request. An “entry message” for recognition is sent by a communication means on a normal Internet (S907 in FIG. 9).

  The representative cache node 1 which has made a request to the cache node 11 sends a message indicating that the “entry message” has been received to the newly added cache node 11 by communication within the upper cluster 17, and at the same time, the upper cluster map 18 (S908 in FIG. 9). The newly added cache node 11 that has received the request sends request information related to its own cache node 11 to the representative cache node 1 that has made a request to the cache node 11 by communication within the upper cluster 17.

  The representative cache node 1 that made the cache node request updates the upper cluster map 18-11 based on the request information received from the new representative cache node 11, and uses the updated upper cluster map 18-11 as the upper cluster 17. By internal communication, the data is sent to another representative cache node in the lower cluster to which the representative cache node belongs. Each representative cache node updates its own upper cluster map for the upper cluster, and sends a registration completion message to the representative cache node 1 that has requested the cache node by communication within the upper cluster (S908 in FIG. 9). .

  Referring to FIG. 11, the upper cluster level NW17 (upper layer), the lower cluster level NWs 2, 15, 16 (middle layer), and nodes, which are composed of only representative cache nodes in the distributed information arrangement system of this embodiment. An example of the level NW (lowermost layer) relationship is shown.

  In principle, the lower-level clusters are the representative cache nodes, and the cache nodes 1, 11, and 14 send macro information such as the number of cache nodes in the lower-level clusters 2, 15, and 16 at regular time intervals or at indefinite time intervals. Are exchanged by performing overlay communication on the upper cluster NW17 to check the existence and status of each other, and each of the lower clusters 2, 15, 16 held by the representative cache nodes 1, 11, 14 The macro information is updated (S909 in FIG. 9).

  Here, in order to prevent a loop of the same information, some overlay communication may be communication after setting any one of the representative cache nodes 1, 11, and 14 as a route reflector defined in RFC 2796 of IETF. In order to prevent the same information loop, any overlay communication in the lower cluster NW may use OSPF defined in RFC 2328 of the IETF as a protocol. In order to prevent a loop of the same information here, any overlay communication in the lower cluster NW may use IS-IS defined in RFC 1142 or RFC 1195 of IETF as a protocol.

  Further, in order to synchronize the lower cluster map for the upper cluster 17 of all the representative cache nodes, the information exchanged by some overlay communication in the upper cluster 17 may include version information in some format. good.

(4) Description of Operation Regarding Reduction of Cache Node With reference to FIG. 12, FIG. 13, and FIG. 14, an operation regarding reduction of the cache node in the distributed information arrangement system of the present embodiment will be described.

  Take the lower cluster 15 of FIG. 12 as an example. When the access to the cache node 19 decreases and the “numerical value of“ information ”” falls below a certain threshold Θ2, taking into account the fluctuation of the numerical value, Average values of “numerical values” are acquired (S1301 and S1302 in FIG. 13). If the average value of “numerical value of information” after a certain period of time falls below a certain threshold Θ2, as long as there is another cache node in the lower cluster 15 to which the own cache node 19 belongs, A “withdrawal message” is sent to all other cache nodes by communication within the lower cluster 15 (S1303 and S1304 in FIG. 13).

  Here, the threshold Θ2 is within the range of 0 to 99% of the maximum value of the input buffer size of the interface that can receive the cache node access, and the threshold Θ1 when a cache node described later is added to the lower cluster. Set to a smaller percentage. The threshold Θ2 may be different for each cache node according to the number of interfaces that can receive the cache node access, the input buffer size of the interface, and the like. The threshold Θ2 may be different for each lower cluster according to the maximum number of cache nodes that can be placed in the own lower cluster.

  The other cache nodes in the lower cluster 15 that have received the withdrawal message delete information other than the IP address of the withdrawal cache node 19 from the lower cluster map for the lower cluster 15 held by the own cache node, The confirmation message “has arrived” is transmitted to the withdrawal cache node 19 by communication within the lower cluster (S1305 and S1306 in FIG. 13). The withdrawal cache node 19 receives the confirmation message from all the other cache nodes in the lower cluster 15, and then deletes the cache data held by the own cache node 19 and the URL of the information and its own IP address. Is deleted from the lower level cluster 15 (S1307 in FIG. 13).

  Here, if the representative cache node 11 leaves, it will be as follows. When the access to the representative cache node 11 decreases and the “numerical value of“ information ”” falls below a certain threshold Θ2, the “necessity of information” for a certain time from that point in time is considered in consideration of the fluctuation of the numerical value. Average value ”is acquired (S1401, S1402 in FIG. 14). When the average value of the “numerical value of“ information ”” for a certain time falls below a certain threshold Θ2, among the other cache nodes in the lower cluster 15 to which the own cache node 11 belongs, “the“ necessity of information ” The representative cache node that communicates with the other lower clusters 2 and 16 is transferred to the cache node 13 having a small “number” and all other cache nodes in the lower cluster 15 to which the own cache node 11 belongs and other lower nodes All of the representative cache nodes 1 and 14 of the clusters 2 and 16 are informed by the communication in the lower cluster and the communication in the upper cluster, respectively, that “the other cache node 13 takes over the representative cache node and leaves itself 11”. (S1403, S1404 in FIG. 14).

  All other cache nodes in the lower cluster 15 that have received the withdrawal message from the representative cache node 11 delete the information of the representative cache node 11 from the lower cluster map for the lower cluster 15 except for the IP address, The “cache node 13 that has inherited the representative cache node” is set as the representative cache node in the lower cluster map for the lower cluster 15. Further, for “the original representative cache node 11” and “the cache node 13 that has inherited the representative cache node”, “new representative cache node 13 is changed to lower cluster map 4-15 for lower cluster 15 by communication within the lower cluster”. "Registered" is transmitted as a confirmation message (S1405 in FIG. 14).

  Similarly, each of the representative cache nodes 1 and 14 of the other lower clusters 2 and 16 deletes the information of the representative cache node 11 from the upper cluster maps 18-2 and 18-16 for the upper cluster NW17 held by each of them. Then, the new representative cache node 13 is set as the representative cache node of the lower cluster 15 in the upper cluster maps 18-2 and 18-16 for the upper cluster NW17. Further, the representative cache nodes 1 and 14 of each of the other lower clusters 2 and 16 transmit confirmation messages to the “cache node 13 that has inherited the representative cache node” and the “original representative cache node 11” by communication within the upper cluster. (S1406 in FIG. 14).

  Confirmation messages from all the cache nodes in the lower cluster 15 to which the own cache node 11 belongs and all the representative cache nodes 1 and 14 in the other lower clusters 2 and 16 are transmitted by communication in the lower cluster and communication in the upper cluster. After receiving each, the “original representative cache node 11” deletes the cache data held by the own cache node 11 and issues a deletion request to the DNS server that registered the URL of the information and its own IP address. Exit from the cluster 15 (S1407 in FIG. 14).

(5) Description of operation related to cluster reduction With reference to FIG. 15 and FIG. 16, an operation related to cluster reduction in the distributed information arrangement system of this embodiment will be described.

  A description will be given by taking the lower cluster 16 as an example. The cache node 14 finally remaining in the lower cluster 16 is inevitably a representative cache node of the lower cluster. However, even if the representative cache node 14 falls below a certain threshold value, the numerical value of In consideration of the fluctuation, an average value of “numerical value of“ information necessity ”” for a certain time from that time is acquired (S1601 and S1602 in FIG. 16). When the average value of the “numerical value of information” for a certain time falls below a certain threshold, the representative cache node 14 sends a withdrawal message to the representative cache nodes 1 and 11 of all other lower clusters 2 and 15. (S1603, S1604 in FIG. 16). At the same time, the withdrawal representative cache node 14 looks at the upper cluster map 18-16 for the upper cluster NW17 held by the own cache node 14, and among the representative cache nodes 1 and 11 of all the other lower clusters 2 and 15, The representative cache node 11 having the smallest “number of“ information necessity ”” per cache node and the largest cache node density in the lower cluster is assigned to the lower cluster 16 to which the own cache node 14 belongs. The lower cluster map 4-16 is transmitted by upper cluster communication (S1606 in FIG. 16).

  The representative cache nodes 1 and 11 of all the other lower clusters 2 and 15 send a message confirming that “representative cache node 14 is leaving” to the representative cache node 14 that has transmitted the withdrawal message “representative cache node 14”. After transmission to the destination by communication within the upper cluster, information other than the IP address related to the representative cache node 14 is deleted from each of the upper cluster maps 18-2 and 18-15 for the upper cluster NW17 (S1605 in FIG. 16). Here, the representative cache node 11 that has received the lower cluster map 4-16 for the lower cluster 16 from the withdrawal representative cache node 14 uses the confirmation message “The lower cluster map 4-16 has been received” as a confirmation message. 14 is transmitted by communication within the upper cluster (S1607 in FIG. 16).

  The withdrawal representative cache node 14 receives a message confirming that “the representative cache node 14 is leaving” from the representative cache nodes 1 and 11 of all the other lower-level clusters 2 and 15, and the lower-level cluster 16 from the withdrawal representative cache node 14. After receiving a message confirming that “the lower cluster map 4-16 has been received” from the representative cache node 11 that has received the lower cluster map 4-16 for internal use, the cache of the own cache node 14 is retained. The cache data to be deleted is deleted, a deletion request is issued to the DNS server in which the URL of the information and its own IP address are registered, and the process leaves the upper cluster NW 17 (S1608 in FIG. 16).

(6) Description of Operation Related to Withdrawal of End Cache Node With reference to FIGS. 17 and 18, an operation related to withdrawal of the end cache node will be described.

  In the finally remaining lower cluster 15, even if the finally remaining cache node 11 (terminal cache node) falls below a certain threshold value of “the“ necessity of information ””, the change of the numerical value is taken into consideration. The average value of “numerical value of“ information necessity ”” for a certain time from the time is acquired (S1801 in FIG. 18). When the average value of the “number of“ information necessity ”” for a certain time falls below a certain threshold value, “the node 11 is the cache node of the final data” and the terminal cache node holds for the original cache node 1 “Cache data” and “lower cluster map for lower cluster of other lower cluster that has left in the past” may be placed in the own cache node 11 as it is, or 2) some other form ( For example, it may be placed as a link destination or the like, or 3) “desired information” that the user wants to be deleted is transmitted by intra-cluster communication (S1802 in FIG. 18).

  Based on the “desired information” from the terminal cache node 11, the original cache node 1 receives the information 1) as it is, or 2) places it in some other form (for example, as a link destination). Or 3) to determine whether or not to delete, and to communicate the result to the terminal cache node 11 by communication within the upper cluster (S1803 in FIG. 18). Here, the reason why the terminal cache node 11 continues to cache the information is that when the data is needed again, it is estimated that there are many cases from the periphery of the cache node that has been accessed to the end.

  When the terminal cache node 11 deletes the information cached by the terminal cache node 11, the original cache node 1 sends the information to the other cache nodes 13 and 19 in the last remaining lower cluster 15. 1) I want you to leave it as it is, and 2) I want you to put it in some other form (for example, as a link destination, etc.) and tell it by the communication means on the normal Internet (S1805 in FIG. 18).

  The terminal cache node 11 determines the information management form according to the answer from the original cache node 1. That is, the information is either 1) placed as it is, 2) placed in some other form (for example, as a link destination), or 3) deleted (S1804 in FIG. 18).

  Received a message from the original cache node 1 that 1) want the information to be left as it is, 2) have the message be placed in some other form (for example, as a link destination), and was in the last remaining lower cluster 15 Each of the other cache nodes 13 and 19 can store information on the original cache node 1 in the form of 1) its own cache node, or 2) in some other form (for example, as a link destination). Alternatively, 3) “desired information” that the user does not want to place is transmitted by a normal communication method on the Internet (S1806 in FIG. 18).

  The original cache node 1 that has received the “desired information” may place the information in the cache node 1 as it is, or 2) in some other form (for example, as a link destination). If there is a good answer, 1) and 2) priorities are formed to form a higher cluster with the cache node, and the information may be left as it is in 1) its own cache node. 2) Some other form ( For example, the information to be cached is transmitted by communication within the upper cluster to the cache node that has made a reply that it may be placed as a link destination (S1807 in FIG. 18).

  If the original cache node 1 that has received the “desired information” has only the answer that 3) it does not want to be placed in it, the information is sent to all the cache nodes in the subordinate cluster 16 that is deleted second from the end. 1) I want you to leave it as it is, 2) I want you to let me put it in some other form (for example, as a link destination, etc.), and tell it by the communication method on the normal Internet.

  Received a message from the original cache node 1 that the information should be placed 1) as it is, 2) the message that you want to place it in some other form (for example, as a link destination), etc. All the cache nodes 14, 20, and 21 in 16 may store information on the original cache node 1 as it is in 1) its own cache node, 2) some other form (for example, as a link destination) Etc.), or 3) “desired information” that the user does not want to place is transmitted by a normal communication method on the Internet (S1808 in FIG. 18).

  Thereafter, the procedure from paragraph number 0199 to paragraph number 0202 is repeated.

  If there is no cache node that caches other than the original cache node 1 in the end, the original cache node 1 keeps the information stored in its own cache node 1 (S1809 in FIG. 18).

  Here, the information stored in the original cache node 1 may be the cache information and the registered name and IP address on the DNS server of one or all representative nodes registered in the past of the upper cluster. The information stored by the original cache node 1 may include cache information and registered names and IP addresses on the DNS server of one or all cache nodes registered in the past in a plurality of lower-level clusters. The information stored in the original cache node 1 includes cache information and one or all representative nodes registered in the past of the upper cluster and one or all cache nodes registered in the past of a plurality of lower clusters. The registered name and IP address on the DNS server may be included.

(7) Explanation of operation when access increases again In the above-mentioned “(6) Explanation of operation concerning withdrawal of terminal cache node”, the first depends on which cache node holds information to be finally cached. Only the steps are different.

(7-1) In the case where only the original cache node holds information to be cached finally In this case, “(1) Description of operation concerning increase of cache nodes” is the first step.

(7-2) When the original cache node and the terminal cache node hold information to be cached finally

  This will be described with reference to FIG. In this case, among the original cache node 1 and the terminal cache node 11, “(1) Description of operation related to increase in cache nodes” is the first step in the cache node that has been accessed. When access is increased in the terminal cache node 11 and it becomes necessary to add a cache node exceeding the threshold Θ1 described in “(1) Explanation of operation regarding increase in cache node”, “(1) Cache node The terminal cache node 11 remains in the lower-level cluster map 4-15 in the lower-level cluster 15 held by the terminal cache node 11 after waiting for a certain time in accordance with the procedure described in “Explanation of the operation related to increase”. The cache nodes 13 and 19 that have been cache nodes in the past are inquired by the communication method on the normal Internet whether or not the information is cached. If consent is not obtained from either of them, it is determined whether or not information can be randomly cached by a cache node having a higher degree among cache nodes having a certain number of hops. Consult with the method. If consent is not obtained from a cache node within a certain number of hops, the terminal cache node increments the number of search hops by one and searches for a cache node that can obtain consent.

  When consent is received from a cache node 13 or 19 in the lower-level cluster 15 or any of the cache nodes by a communication method on a normal Internet by a search based on the number of hops, “(1) Operation related to increase of cache nodes” The cache nodes are increased in accordance with the procedure described in “Description”.

  According to the present embodiment, the following effects can be obtained.

  The first effect is that the distributed cache arrangement can be performed efficiently. The reason for this is that a lower-level clustering of cache nodes using a representative cache node, the introduction of an NW distance centered on the representative cache node or lower-level cluster, and an NW when access from a destination farther than the NW distance increases by a certain amount or more This is because another subordinate cluster is formed around the cache node located farther away than the distance.

  The second effect is that load distribution of access between cache nodes can be performed autonomously and efficiently. The reason is that a unique overlay NW is formed in the lower level cluster including the representative cache node, and a mechanism in which all the cache nodes belonging to the lower level cluster can grasp the load status of each cache node is used. is there.

  A third effect is that information that has been reduced in access over a period of time and that has been hardly or not accessed at all can be efficiently increased when access is increased again. The reason is that a mechanism for reducing cache nodes and lower-level clusters accompanying a decrease in access amount and a mechanism for efficiently storing final information are provided.

  The fourth effect is that each lower cluster performs load distribution in an autonomous distributed manner, and each lower cluster is arranged at a different distance on the network, so that information from destructive actions on cached information can be obtained. It is possible to protect, that is, the robustness against information attack acts is improved.

  Next, an embodiment of a node device that is a component of the distributed information arrangement system of the present invention will be described.

  Referring to FIG. 20, the node device according to the present exemplary embodiment includes a processing device 1100, a storage device 1200, and a communication device 1300.

  The storage device 1200 includes computer main memory and auxiliary memory, and stores cache information 1210 that is information to be provided to the user terminal, the lower cluster map 1220 shown in FIG. 4, and the upper cluster map 1230 shown in FIG. Used for.

  The communication device 1300 includes a network interface card or the like, and is used as the Internet communication unit 1310 and the overlay communication unit 1320.

  The processing device 1100 is constituted by a central processing unit of a computer, and includes a load distribution unit 1110, a service providing unit 1120, a load information update unit 1130, a cache node increase unit 1140, a cache node decrease unit 1150, a cluster increase unit 1160, and a cluster decrease unit. 1170, terminal means 1180, and cluster map synchronization means 1190. Each of these means can be realized by a program executed on the processing apparatus 1100. The program is recorded on a computer-readable recording medium such as a magnetic disk, and is read by the computer, thereby realizing the above-described units 1110 to 1190 on the computer.

  When the load distribution unit 1110 receives an access request for the cache information 1210 from the user terminal, the load distribution unit 1110 determines which node of the cluster to which the own node belongs the access request is received by each node described in the lower cluster map 1220. It has a function of making a determination based on load information (in the case of the present embodiment, a numerical value of “information necessity”). The load distribution unit 1110 transfers the access request to the service providing unit 1120 of its own node when it is determined to be received by its own node, and when it is determined that it is received by another node, Forward the access request to the node. The load balancer 1110 has a function of registering the correspondence between the global IP address of the own node and the URL of the cache information 1210 in the DNS server of the NW domain to which the own node belongs.

  The service providing unit 1120 has a function of transmitting the cache information 1210 to the requesting user terminal in accordance with the access request notified from the load distribution unit 1110.

  Each time the service providing unit 1120 accesses the cache information 1210, the load information updating unit 1130 increases the value of “information necessity” of the own node in the lower cluster map 1220, as described in FIG. Accordingly, it has a function of updating the primary and secondary differential values. Further, the load information update unit 1130 decreases the value of “information necessity” of the own node in the lower cluster map 1220 every predetermined time, and updates the primary and secondary differential values accordingly. It has a function.

  As described in FIGS. 2 and 3, the cache node increasing unit 1140 has a function of adding a new cache node to the cluster to which the own node belongs. The cache node increasing unit 1140 includes a master unit 1141 that is a processing unit that is executed by a representative node of the cluster and a slave unit 1142 that is a processing unit that is executed by a node other than the representative node. It consists of.

  As described with reference to FIGS. 12, 13, and 14, the cache node reducing unit 1150 has a function of reducing cache nodes from the cluster. This cache node reduction means 1150 is executed by the master unit 1151 which is a processing unit executed by the node to be reduced and the nodes other than the node to be reduced among the processes described in FIG. 13 and FIG. It is comprised with the slave 1152 which is a process part.

  The cluster increasing unit 1160 has a function of adding a new cluster as described with reference to FIGS. The cluster increasing means 1160 includes a master unit 1161 that is a processing unit executed by a representative node that is to generate a new cluster and a non-representative node that is to generate a new cluster. And a slave unit 1162 which is a processing unit executed by the nodes.

  As described with reference to FIGS. 15 and 16, the cluster reduction unit 1170 has a function of reducing clusters. This cluster reduction means 1170 is executed by the master unit 1171 which is a processing unit executed by the representative node of the cluster to be reduced and the nodes other than the representative node of the cluster to be reduced among the processes described in FIG. And a slave unit 1172 which is a processing unit.

  The terminal unit 1180 has a function of managing processing related to the terminal cache node as described with reference to FIGS. 17 and 18. The termination means 1180 includes a master unit 1181 that is a processing unit that is executed by the original cache node and a slave unit 1182 that is a processing unit that is executed by other nodes among the processes described in FIG. Is done. The cluster map synchronization unit 1190 has a function of managing the process of matching the contents of the cluster map between cache nodes. The cluster map synchronization means 1190 is a processing unit that executes the upper cluster unit 1191 that is a processing unit that executes the synchronization processing of the upper cluster map shown in FIG. 10 and the synchronization processing of the lower cluster map that is shown in FIG. A lower cluster unit 1192 is included.

  As mentioned above, although embodiment and the Example of this invention were described, this invention is not limited to the above example, Various other addition changes are possible. For example, in the above example, the cache node of each cluster registers the correspondence between the global IP address of the own node and the URL of the information in the DNS server of the NW domain to which the own node belongs, but the representative node of the cluster to which the own node belongs. The correspondence relationship between the global IP address and the information URL may be registered in the DNS server of the NW domain to which the own node belongs. In this case, an access request from the user terminal to the cache information of the own cluster is issued to the representative node of the own cluster, and the representative node determines which node in the own cluster will process the access request. .

  Examples of utilization of the present invention include Web services, peer-to-peer services, data storage, etc. that perform caching on a communication NW.

It is a figure which shows one example of the mode of a change of the numerical value of the necessity level of the information in the distributed information arrangement | positioning system of one Example of this invention. It is a figure which shows an example of the mode of the new cache node addition in the distributed information arrangement | positioning system of one Example of this invention. It is a figure which shows an example of the mode of the new cache node addition in the distributed information arrangement | positioning system of one Example of this invention. It is a figure which shows one example of the flowchart of a new cache node addition in the distributed information arrangement | positioning system of one Example of this invention. It is a figure which shows one example of the flowchart of a new cache node addition in the distributed information arrangement | positioning system of one Example of this invention. It is a figure which shows an example of the low-order cluster map for low-order clusters which each cache node hold | maintains in the distributed information arrangement | positioning system of one Example of this invention. It is a figure which shows one example of the mode of communication between the cache nodes in the same low-order cluster in the distributed information arrangement | positioning system of one Example of this invention. It is a figure which shows an example of the mode of the load distribution of the access in the same low-order cluster in the distributed information arrangement | positioning system of one Example of this invention. It is a figure which shows an example of the mode of the load distribution of the access in the same low-order cluster in the distributed information arrangement | positioning system of one Example of this invention. It is a figure which shows one example of the flowchart of the load distribution of the access in the distributed information arrangement | positioning system of one Example of this invention. It is a figure which shows one example of the flowchart of the load distribution of the access in the distributed information arrangement | positioning system of one Example of this invention. In the distributed information arrangement system according to an embodiment of the present invention, an increase in accesses from cache nodes having a certain NW distance or more increases the number of cache nodes having a higher order than the NW distance, and generation of a new lower cluster therein. It is a figure which shows one example of a mode performed. In the distributed information arrangement system according to an embodiment of the present invention, an increase in accesses from cache nodes having a certain NW distance or more increases the number of cache nodes having a higher order than the NW distance, and generation of a new lower cluster therein. It is a figure which shows an example of the flowchart at the time of being performed. It is a figure which shows an example of the upper cluster map of the upper cluster NW comprised only by a representative cache node in the distributed information arrangement system of one Example of this invention. In a distributed information arrangement system according to an embodiment of the present invention, an upper cluster level NW (upper layer), a lower intra-cluster level NW (middle layer), and a node level NW (lowermost layer) configured only by representative cache nodes It is a figure which shows one example of the relationship of). It is a figure which shows one example of the mode of withdrawal of the cache node in the distributed information arrangement system of one Example of this invention. It is a figure which shows one example of the flowchart of withdrawal of the cache node in the distributed information arrangement | positioning system of one Example of this invention. It is a figure which shows one example of the flowchart of withdrawal of the representative cache node in the distributed information arrangement system of one Example of this invention. It is a figure which shows one example of the mode of withdrawal of the low-order cluster in the distributed information arrangement | positioning system of one Example of this invention. It is a figure which shows one example of the flowchart of withdrawal of the low-order cluster in the distributed information arrangement | positioning system of one Example of this invention. It is a figure which shows one example of the mode of withdrawal of the cache node which remained at the end in the distributed information arrangement | positioning system of one Example of this invention. It is a figure which shows one example of the flowchart of withdrawal of the cache node which remained at the end in the distributed information arrangement | positioning system of one Example of this invention. It is a block diagram of the distributed information arrangement system of one embodiment of this invention. It is a block diagram of one Example of the node apparatus of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Original cache node and representative cache node 2 of lower cluster 2 ... Lower clusters 3, 3a, 3b, 3c including original cache nodes ... Ants 4, 4-2, 4-15, 4-16 representing access Map 5 ... A certain distance 6 from the original cache node 1 ... A cache node 8 in the lower cluster 2 ... A cache node 10 in the lower cluster 2 ... A certain physical or logical distance 10a from the original cache node 1 or the lower cluster 2 ... A certain distance or logical distance 10 b from the representative cache node 11 or the lower cluster 15... A certain distance or logical distance 11 from the representative cache node 11 or the lower cluster 16... A newly added cache node and a representative cache node of the lower cluster 15. 13 Cache node 14 in lower cluster 15 ... newly added cache node and representative cache node 15 in lower cluster 16 ... lower cluster 16 created by newly added cache node 11 ... lower level created by newly added cache node 14 Cluster 17: upper cluster composed of only representative cache nodes of each lower cluster and its NW
18 ... Cluster map for upper cluster 19 ... Cache node 20 in lower cluster 15 ... Cache node 21 in lower cluster 16 ... Cache nodes 31, 32, 33, 34, 35, 36, 37 in lower cluster 16 38, 39 ... DNS server A ... Increase in numerical value of necessity of information by access, baggage B carried by ant 3 ... general node (node not cached)

Claims (19)

A distributed information arrangement system in which a plurality of node devices including a cache for storing data and a means for reading the data from the cache and transmitting the data to a terminal device are connected via a network,
A cluster comprising a set of one or more node devices that store the same data in the cache;
In the cluster, any one of the one or more node devices included in the set functions as a representative node,
The representative node is configured to determine whether or not a new cluster generation condition is satisfied, and to the other node device that has not yet cached the data when the new cluster generation condition is satisfied. A means for sending a new cluster generation instruction with a copy of the data is provided,
The node device that has received the new cluster generation instruction includes means for storing a copy of the data in the cache of the own node device and generating a new cluster having the own node device as a representative node. Distributed information placement system.   The means for determining whether or not the new cluster generation condition is satisfied collects and monitors access information from the terminal device to each node device in its own cluster, and exceeds a predetermined network radius. 2. The distributed information arrangement system according to claim 1, wherein the new cluster generation condition is determined to be satisfied when a situation occurs in which access from a distance increases. 3.   The means for determining whether or not the new cluster generation condition is satisfied is that the new cluster generation condition is satisfied when the number of node devices belonging to the own cluster reaches a predetermined maximum number. The distributed information arrangement system according to claim 1, wherein the determination is performed.   The representative node collects and monitors the load information of each of the node devices in its own cluster, and when the load becomes high, copies the data to other node devices that have not yet cached the data. The distributed information arrangement system according to any one of claims 1 to 3, wherein a process of transmitting and adding the node device to a new member of the own cluster is performed.   The representative node collects and monitors load information of each of the node devices in the own cluster, and deletes the data cache for several constituent members in the own cluster when the load becomes low. The distributed information arrangement system according to any one of claims 1 to 3, wherein an instruction is transmitted to perform processing for removing the node device from a member of the own cluster.   The representative node of the newly generated cluster will extinguish its own cluster when the node device belonging to its own cluster is only its own node and the access frequency from the terminal device to the own node decreases. 6. The distributed information arrangement system according to claim 1, wherein processing is performed.   In the cluster that finally remains other than the cluster that includes the node device that holds the original data, if the access frequency from the terminal device decreases even in the terminal node that is the node device that finally remains, When the end node and the node device that owns the original of the data negotiate the maintenance and destruction of the cache data in the end node, and if both agree, the end node caches the data The distributed information arrangement system according to claim 6, wherein the distributed information arrangement system is maintained.   7. The cluster according to claim 1, wherein the cluster has a function of distributing a load related to access from the terminal device to the same provided data among the one or more node devices included in the own cluster. The distributed information arrangement system according to claim 1. Among node devices on a network including a cache for storing data and means for reading the data from the cache and transmitting the data to a terminal device, one set of one or more node devices that store the same data in the cache. Forming two clusters;
A node device functioning as a representative node among the one or more node devices constituting the cluster determines whether or not a new cluster generation condition is satisfied, and when the new cluster generation condition is satisfied, the data Sending a new cluster generation instruction with a copy of the data to the other node devices that have not yet cached,
The node device that has received the new cluster generation instruction stores a copy of the data in the cache of the own node device, and generates a new cluster having the own node device as a representative node. Distributed information arrangement method.   When the representative node collects and monitors access information from the terminal device to each of the node devices in its own cluster, and a situation occurs in which access from a distance exceeding a predetermined network radius occurs. 10. The distributed information arrangement method according to claim 9, wherein it is determined that the new cluster generation condition is satisfied.   The said representative node determines with the said new cluster production | generation conditions having been satisfied, when the number of the said node apparatuses which belong to a self-cluster reaches the predetermined maximum number. Distributed information placement method. A cache for storing data, and means for reading the data from the cache and transmitting the data to a terminal device, and forming a cluster together with other node devices for storing the same data in the cache, A node device that functions as a representative node of
Means for determining whether or not a new cluster generation condition is satisfied; and when the new cluster generation condition is satisfied, a copy of the data is attached to another node device that has not yet cached the data. And a means for transmitting a new cluster generation instruction.   The means for determining whether or not the new cluster generation condition is satisfied collects and monitors access information from the terminal device to the node device in its own cluster, and exceeds a predetermined network radius. 13. The node device according to claim 12, wherein when a situation occurs in which access from the network increases, it is determined that the new cluster generation condition is satisfied.   The means for determining whether or not the new cluster generation condition is satisfied is that the new cluster generation condition is satisfied when the number of node devices belonging to the own cluster reaches a predetermined maximum number. The node device according to claim 12, wherein the node device is determined.   A node device comprising a cache for storing data and means for reading the data from the cache and transmitting the data to a terminal device, wherein a new cluster generation instruction has been received from a representative node of a cluster for which a new cluster generation condition has been satisfied A copy of the data attached to the received new cluster generation instruction is stored in the cache of the own node device, and a means for generating a new cluster having the own node device as a representative node is provided. Node device to perform. A cache for storing data, and means for reading the data from the cache and transmitting the data to a terminal device, and forming a cluster together with other node devices for storing the same data in the cache, A computer constituting a node device functioning as a representative node of
Means for determining whether or not a new cluster generation condition is satisfied; and when the new cluster generation condition is satisfied, a copy of the data is attached to another node device that has not yet cached the data. To function as a means for transmitting a new cluster generation instruction.   The means for determining whether or not the new cluster generation condition is satisfied collects and monitors access information from the terminal device to each node device in its own cluster, and exceeds a predetermined network radius. The program according to claim 16, wherein it is determined that the new cluster generation condition is satisfied when a situation occurs in which access from a distance increases.   The means for determining whether or not the new cluster generation condition is satisfied is that the new cluster generation condition is satisfied when the number of node devices belonging to the own cluster reaches a predetermined maximum number. The program according to claim 16, wherein the program is determined.   A computer constituting a node device comprising a cache for storing data and a means for reading the data from the cache and transmitting the data to the terminal device is instructed to generate a new cluster from a representative node of the cluster that satisfies the new cluster generation condition. When received, a copy of the data attached to the received new cluster generation instruction is stored in the cache of the own node device, and functions as a means for generating a new cluster with the own node device as a representative node. Program.

Images