JP2008176615A - File sharing network system, data structure used for file transer of file sharing network system, file transfer method for file sharing network system, file disclosure computer, cache computer, file acquisition computer and computer program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a new technical means for efficiently restoring a file. <P>SOLUTION: The file sharing network system comprises a metadata generation part 12 for generating a plurality of metadata m1, m2, m3 and m4 encoded so as to be erasable or correctable from a shared original file F, and a metadata decoding part 53 for acquiring the metadata for acquiring the original file, and decoding the metadata to restore the original file. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a file sharing network system and the like.

The file sharing network system is for sharing a file held by each computer on a network such as the Internet.
In a file sharing network system, a file released to the network by a user is a shared file. Another user searches for a shared file on the network and downloads the file to obtain a file published by another person via the network.

  As such a file sharing network system, P2P (peer-to-peer) file sharing network systems such as “Winny” and “Freenet” are known. Winny is disclosed in Non-Patent Document 1.

  In Winny, when a computer that needs a shared file (search node) obtains the shared file via the network, it not only obtains it from the computer that published the file (original node), but also duplicates the shared file (original file). It can also be acquired from the computer (cache node) that it has.

The cache file held by the cache node is obtained by dividing the shared file into cache blocks each having a predetermined size (for example, 64 Kbytes).
When the search node acquires the shared file, the file transfer efficiency can be improved by multiple downloading from the original node and the cache node.

In addition, since there is a cache node, a shared file can be acquired from the cache node even if the original node leaves the network. Therefore, if there is a cache node, there is a high possibility that the search node can obtain the necessary file, and the file sharing efficiency in the network can be increased to some extent.
Kaneko Isamu, “Winny's Technology”, 1st Edition, ASCII Co., Ltd., October 20, 2005, p. 73-77

For a shared file, the more cache nodes have the cache file, the more likely each search node can obtain the required file.
However, since the storage area (specifically, HDD) of the computer constituting the cache node is limited, there is a limit for many cache nodes to have the same cache file.

In view of this, it is conceivable that the cache node holds a part of a plurality of cache blocks (fragment blocks) constituting one cache file. In this case, a large number of cache nodes on the network hold the cache blocks constituting one cache file in a distributed manner.
One cache node has only a cache block that is a fragment block of the shared file, but the search node can obtain the entire shared file by collecting separate fragment blocks from a plurality of cache nodes. .

However, simply fragmenting the shared file remains highly likely that the search node cannot obtain the file.
For example, assume that one shared file is fragmented into five cache blocks, one cache node randomly holds two cache blocks, and five cache nodes hold cache blocks.

In this case, in the most ideal case, the original shared file can be restored by acquiring the cache block from the three nodes. However, in the worst case, there is a possibility that the original file cannot be restored even if the cache blocks of all five cache nodes are collected.
That is, in the above example, in order to restore one shared file, five types of cache blocks are required. However, even if 2 blocks × 5 nodes = 10 blocks are collected, 10 cache blocks are collected. If only four or less types of cache blocks are included, five types of cache blocks required for restoration cannot be obtained, and the original file cannot be restored.

  On the other hand, the more nodes have a cache block for a shared file, the more likely it is that the file can be obtained from the network. However, if many nodes have cache blocks, the number of cache blocks held in the node is unnecessarily large compared to the number of cache block types necessary for file restoration. In other words, many cache blocks of the same type are duplicated on the network, and file sharing efficiency is reduced.

  Accordingly, an object of the present invention is to provide a new technical means for efficiently restoring a file.

[File sharing network system]
The file sharing network system of the present invention includes a metadata generation unit that generates a plurality of metadata encoded so as to be erasure-correctable from a shared original file, and acquires the metadata for acquiring the original file. A metadata decrypting unit that decrypts and restores the original file.
According to the above configuration, the original file is encoded so that the loss can be corrected, so that the original file can be restored even if a part of the metadata is lost.

  Preferably, the metadata generation unit generates metadata by an erasure correction type rateless code. By using the erasure correction type rateless code, innumerable types of metadata can be generated, and the metadata can be diversified.

The metadata generation unit performs encoding with erasure correction codes on a plurality of blocks obtained by dividing the original file, and generates non-essential metadata for restoration. The metadata decoding unit includes: It is preferable to restore the original file by obtaining as many metadata as are necessary for the restoration as necessary for the restoration.
According to the above configuration, the metadata becomes non-essential data for restoration, and the metadata decoding unit can restore the original file if the number necessary for restoration can be acquired.

[data structure]
The metadata preferably has a metadata main body encoded so that the original file can be lost and corrected, and arithmetic expression information for encoding used to generate the metadata main body. In this case, the side receiving the metadata can grasp the calculation formula information for encoding from the metadata itself.

  The metadata preferably includes location information on a network where the metadata exists. In this case, the metadata position can be acquired by referring to the metadata.

  The metadata preferably includes file information related to the original file that is the basis for generating the metadata. In this case, information about the original file can be acquired by referring to the metadata.

[File transfer method]
The file transfer method in the file sharing network system according to the present invention includes a step of generating metadata encoded so as to be erasure-correctable from a shared original file and transmitting it to the network, and a file acquisition computer from the network Obtaining the metadata and restoring the original file.

  According to the above method, the original file is encoded so that the loss can be corrected, and the original file can be restored even if a part of the metadata is lost.

  From another viewpoint, a file transfer method includes a step of generating metadata encoded so as to be erasure-correctable from a shared original file, and a step in which a plurality of cache computers on the network distribute and hold the metadata. And a step in which the file acquisition computer acquires the metadata from the cache computer and restores the original file, and the number of types of metadata held by a plurality of cache computers on the network restores the original file. The metadata is generated so as to be larger than the number of types of metadata necessary for the operation.

  According to the above method, since the number of types of metadata held by the cache computer is larger than the number of types of metadata necessary for restoring the original file, it is possible to easily obtain metadata for restoring the file.

  In the above method, it is preferable that the cache computer notifies other cache computers of the type of metadata held by the cache computer. In this case, each cache computer can grasp the types of metadata possessed by other cache computers, and can determine the cache method in consideration of the metadata possession status of surrounding nodes.

  In the above method, it is preferable that a different calculation expression is used each time metadata is generated from an original file. By doing so, many types of metadata are generated, which is advantageous for file restoration.

[File Public Computer]
The file publishing computer of the present invention includes a metadata generation unit that generates metadata encoded so as to be erasure-correctable from a shared original file, and a transmission unit that transmits the generated metadata to a network. Yes.
According to the above configuration, the original file is encoded so that the loss can be corrected, so that data can be transmitted to the network in a state where the original file can be restored even if a part of the metadata is lost. .

  The metadata generation unit includes an arithmetic expression generation unit that randomly determines an arithmetic expression for generating metadata, and an operation for generating metadata using the arithmetic expression randomly determined by the arithmetic expression determination unit. And a metadata operation unit that performs the above. By including an arithmetic expression generation unit that randomly determines an arithmetic expression for generating metadata, various types of metadata can be generated even from the same original file.

  The arithmetic expression generator preferably determines the number of arithmetic expressions to be generated according to the file size of the original file. In this case, the number of metadata corresponding to the file size of the original file can be generated.

  It is preferable that the metadata generation unit generates a plurality of metadata that is not essential for restoration by performing encoding using erasure correction codes on a plurality of blocks obtained by dividing the original file. A block obtained by dividing an original file is indispensable for file restoration. However, if the block is encoded with a loss correction code, metadata that is not essential for restoration can be generated.

  The metadata generation unit preferably generates metadata from the original file every time there is an original file transmission request or a metadata transmission request from another computer on the network. In this case, every time there is an original file transmission request or a metadata transmission request, metadata is generated.

  Preferably, the metadata generation unit generates different metadata from the original file every time there is an original file transmission request or a metadata transmission request from another computer on the network. In this case, every time there is an original file transmission request or a metadata transmission request, the types of metadata increase.

  The file publishing computer preferably includes means for transmitting information indicating the existence of the original file to be shared to other computers on the network. In this case, the other computer can grasp where the original file is.

[File Publishing Computer Program]
The file public computer program of the present invention is a computer program for causing a computer to function as the file public computer.

[Cache computer]
The cache computer of the present invention stores, as a cache block, metadata obtained by encoding an original file to be shared so that it can be lost and corrected, and transmits the metadata stored in the metadata storage unit to the network. And a transmission unit.
According to the above configuration, the cache computer can hold the metadata obtained by encoding the original file so that the original file can be corrected as a cache block, and other computers on the network can acquire the metadata as the cache block.

  The metadata storage unit preferably stores a smaller number of metadata than is necessary to restore the original file. In this case, a part of the metadata is stored in one cache computer for restoration, and it is not necessary to have a lot of metadata.

  The cache computer includes relay means for relaying one or a plurality of metadata transmitted from one computer on the network to another computer on the network, and the metadata storage unit receives for relay It is preferable to store a part or all of one or a plurality of metadata. In this case, the relayed metadata can be saved as a cache block.

  The cache computer selects metadata to be stored in the metadata storage unit from among one or a plurality of metadata received for relay, and among the metadata already stored in the metadata storage unit It is preferable to provide means for selecting metadata to be discarded. Providing means for selecting metadata to be stored and metadata to be discarded makes it possible to hold appropriate metadata.

  The cache computer preferably includes means for acquiring the type of metadata held by another cache computer on the network. In this case, it is possible to determine how to hold a cache block in consideration of the type of metadata held by another cache computer.

  The cache computer preferably includes means for notifying other cache computers on the network of the type of metadata held by the cache computer. In this case, the type of metadata held by itself can be notified to another cache.

  The cache computer stores metadata to be stored in the metadata storage unit such that at least one metadata different from the metadata held by other adjacent cache computers is included in its metadata storage unit. It is preferable to select. In this case, metadata different from those of other adjacent cache computers can be held, and diversity of metadata on the network can be obtained.

  The metadata transmitted by the transmission unit to the network is preferably metadata of a type requested by another computer among the metadata stored in the metadata storage unit. In this case, only metadata requested by other computers can be efficiently transmitted.

  The cache computer receives unnecessary metadata information indicating a type of metadata that is not required to be transmitted from another computer on the network, and transfers metadata that does not correspond to the metadata type indicated by the unnecessary metadata information to the network. It is preferable to transmit. In this case, it is possible to prevent the metadata already owned by another computer from being duplicated.

[Cache computer program]
The cache computer program of the present invention is a computer program for causing a computer to function as the cache computer.

[File acquisition computer]
The file acquisition computer according to the present invention includes a metadata reception unit that acquires metadata obtained by encoding an original file to be shared so that it can be erasure corrected from a network, and decoding that decrypts the acquired metadata and restores the original file. And a section. According to the above configuration, since the metadata is encoded so that the original file can be lost and corrected, the original file can be restored even if a part of the metadata is lost.

  The file acquisition computer preferably comprises means for searching for metadata present on the network. In this case, the file acquisition computer can grasp the existence of metadata on the network.

  It is preferable that the metadata receiving unit obtains the number of types of metadata more than the number of types of metadata necessary for restoring the original file from the network. In this case, the original file can be reliably restored.

  Preferably, the file acquisition computer acquires information necessary for calculating the number of types of metadata necessary for restoring the original file from the network. In this case, the file acquisition computer can easily grasp the number of types of necessary metadata.

  If the original file restoration fails even after decrypting using the acquired metadata, the file acquisition computer additionally acquires another type of metadata from the network that is different from the already acquired metadata. Is preferred. In this case, even if the original file restoration fails, the original file can be restored.

  The file acquisition computer preferably acquires the metadata preferentially from the cache computer that holds the metadata as a cache block, rather than the file public computer having the original file that has not been converted to metadata. In this case, it is possible to avoid an increase in load due to concentration of access to the file public computer.

  The file acquisition computer preferably acquires the metadata preferentially from other computers connected by a smaller number of logical paths. In this case, metadata can be efficiently acquired from a nearby computer.

  The file acquisition unit preferably includes a metadata generation unit that regenerates metadata from the original file restored by the decryption unit, and publishes the generated metadata to the network. In this case, since the computer that acquired the file also discloses the metadata, the metadata on the network increases.

  It is preferable that the metadata generation unit can generate metadata having a different type from the metadata used for restoring the original file. In this case, since different metadata is generated, the diversity of metadata in the network can be obtained.

[File acquisition computer program]
The file acquisition computer program of the present invention is a computer program for causing a computer to function as the file acquisition computer.

  According to the present invention, a file can be efficiently restored.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
FIG. 1 shows an example of a P2P (Peer to Peer) file sharing network system. Hereinafter, the file sharing network system is referred to as “file sharing network” or simply “network”.
Since this file sharing network is a P2P type, it does not have a server that centrally manages file transfer, but a server may exist if necessary.
There are a plurality of nodes (peers) Ni (i: natural number) on the network. The node Ni is configured by a computer (node computer) participating in the network. A node establishes a logical path P between other nodes and constitutes a P2P network. Note that communication between node computers is performed using the Internet, for example.

  The file transfer is performed along the logical path P. For example, the file transfer from the node N1 to N3 in FIG. 1 is performed via the intermediate node N2. Note that the node Ni may leave the network at the discretion of the computer user.

[Node computer configuration]
The node is configured by a computer 101 in which the file sharing program 100 is installed. More specifically, as shown in FIG. 2, the computer 101 has a storage device 104 including a processing device 103 and an HDD, and the file sharing program 100 is stored in the storage device 104 so that the computer can execute it. Has been.

[Node functions]
The node can function as any one of an original node, a relay node, a search node, and a cache node. That is, the file sharing program (node computer program) is configured to allow the computer 101 to execute any function of the four nodes.

Here, the original node is a node that publishes (shares on a network) a certain file (first).
The relay node is a node that relays file transfer between nodes.
The search node is a node that needs (looks for) a file.
The cache node is a node that holds a cache block for a certain file. In this embodiment, any node (original node, relay node, search node) involved in the transfer of a file has a possibility of functioning as a cache node.

[Configuration of original node (file public computer)]
FIG. 3 shows functional blocks when the node functions as an original node that publishes a file. The original node is an original file storage unit 11 for storing an original file to be disclosed for file sharing, and encodes metadata (also referred to as “metablock” or “metadata block”) from the original file. A metadata generation unit 12 to generate and a metadata transmission unit 13 to transmit metadata to the network are provided. The original node also includes a control unit 20 that controls the units 11, 12, and 13, and a P2P network communication processing unit 21 that performs processing necessary for communication over the P2P network.
The original file storage unit 11 can store one or a plurality of files to be disclosed.

  The metadata generation unit 12 includes an arithmetic expression generation unit (arithmetic expression determination means) 12a that generates an arithmetic expression for generating metadata, and an arithmetic operation for generating metadata using the generated arithmetic expression. And a metadata calculation unit 12b.

[Configuration of relay node (relay computer; relay means)]
FIG. 4 shows functional blocks when the node functions as a relay node that relays file transfer. The relay node receives a metadata receiving unit 31 that receives metadata sent from another node, a metadata storage unit 32 that stores the received metadata, and transmits the metadata to another node for transfer. A metadata transmission unit 33 is provided. In addition, the relay node includes a control unit 20 that controls each unit 31, 32, and 33 and a P2P network communication processing unit 21 that performs processing necessary for P2P network communication.

[Configuration of cache node (cache computer)]
FIG. 5 shows functional blocks when the node functions as a cache node.
The cache node includes a metadata receiving unit 41 that receives metadata transmitted from other nodes, a metadata storage unit 42 that stores received metadata, and a metadata transmission unit that transmits metadata to other nodes. 43. The relay node also includes a control unit 20 that controls the units 41, 42, and 43, and a P2P network communication processing unit 21 that performs processing necessary for P2P network communication.

[Configuration of search node (file acquisition computer)]
FIG. 6 shows functional blocks when the node functions as a search node.
The search node includes a metadata receiving unit 51 for acquiring metadata from other nodes, a metadata storage unit 52 for storing the received metadata, and metadata for decrypting the acquired metadata and restoring the original file. The decryption unit 53 includes an original file storage unit 54 that stores the restored original file.
In addition, the metadata decoding unit 53 includes an arithmetic expression generating unit (arithmetic expression determining means) 53a that generates an arithmetic expression used for decoding, and a metadata decoding that performs an operation for decoding metadata using the generated arithmetic expression. And an arithmetic unit 53b.

The search node further includes a metadata generation unit 55 that generates metadata from the restored original file, a metadata storage unit 56 that stores the generated metadata as a cache block, and a metadata transmission unit 57 that transmits the metadata to the network. It has.
In addition, the metadata generation unit 55 performs an operation for generating metadata using an operation expression generation unit (operation expression determination means) 55a that generates an operation expression for generating metadata, and the generated operation expression. And a metadata calculation unit 55b to be performed.

3 to 6, for convenience of explanation, different reference numerals may be given to functions that are common between the drawings, but functional blocks having common names are not particularly described. As long as it has a common function.
In the present embodiment, the above functions in the original node, relay node, cache node, and search node are realized by a computer program.

[About metadata]
The metadata generation unit 12 included in the original node or the like converts the original file into metadata using an erasure correction type rateless code. The erasure correction code is a code that can correct the erasure even if a part of the data is lost. Further, the erasure correction type rateless code is characterized in that it can generate infinitely encoded data for one original data. Examples of the erasure correction type rateless code include Digital Fountain's Raptor and LT code.

In the erasure correction type rateless code, an original file (original data) is divided into blocks of a certain size (for example, 64 Kbytes), and metadata (metablock) obtained by performing calculation between a plurality of blocks is generated.
Specifically, as illustrated in FIG. 7, the metadata generation unit 12 divides the published original file F into three original blocks b1 to b3. Since the original block has a fixed size, the number of divisions is determined by the size of the original file F.
It should be noted that the original file before division may be encrypted, or the original block after division may be encrypted.

In addition, the arithmetic expression generation unit 12a of the metadata generation unit 12 has a library for obtaining an arithmetic expression for generating metadata, and the number of calculations necessary for generating the required number of metadata. A formula (mainly a formula using XOR (exclusive OR) of bit operations) is randomly generated. In FIG. 7, four arithmetic expressions are generated.
The arithmetic expression here is for calculating a plurality of original blocks with a predetermined operator (XOR or the like) and making each metadata non-essential data for restoring the original file.
Further, the arithmetic expression may be one in which one original block is used as metadata as it is, such as m2 = b2.

  The metadata calculation unit 12b of the metadata generation unit calculates a plurality of original blocks b1, b2, and b3 according to the generated calculation formula. In FIG. 7, the operation of metadata m1 = (b1 xor b3), metadata m2 = b2, metadata m3 = (b2 xor b3), metadata m4 = (b1 xor b2 xor b3) is performed. Metadata m1-m4 is generated.

Metadata can be generated innumerably depending on the type of original block to be operated, the number of original blocks to be calculated, the type of operator, and the like.
On the decoding side (search node), by acquiring a predetermined number of metadata, the original original block can be obtained by solving simultaneous equations composed of a plurality of arithmetic expressions, and the original file can be restored.
In other words, the metadata decoding unit 53 uses several pieces of metadata and the arithmetic expressions used for generating the metadata to solve the simultaneous equations of the arithmetic expressions to obtain an original block.

  Since the original file is restored by solving the simultaneous equations, if there are more than the required number (for example, 4 in FIG. 7), the original file can be restored with a high probability. That is, each piece of metadata is not indispensable for restoring the original file, and the restoration side (search node) only needs to have a required number or more of combinations of metadata and arithmetic expressions that generate the metadata.

  In general, the number of metadata necessary for restoration is slightly larger than the number of original blocks. In FIG. 7, four pieces of metadata m1 to m3 are generated from three original blocks b1 to b3. ing. Details of the policy for generating metadata will be described later.

[Example of file transfer]
Hereinafter, a simple example of file transfer in the present embodiment will be described with reference to FIGS. Here, a P2P file sharing network system in which nodes N1 to N14 exist is assumed. The file transfer rules in this example are as follows.

File transfer rules
The original file to be transferred is assumed to be composed of five original blocks. The search node can restore the original file by acquiring six (six types) different metadata blocks.
The original node transmits the metadata block encoded with the erasure correction type rateless code to other nodes together with the arithmetic expression information. An arithmetic expression for generating metadata is randomly determined using a random number.

  When the relay node relays the metadata block, the relay node becomes a cache node, stores a maximum of two metadata blocks at random with respect to one original file, and discloses the same as a cache block to other nodes. If the cache node relays a metablock of the same original file multiple times, the cache node discards the old cache block and overwrites it with a newly transferred block.

  After restoring the original file, the search node determines a new arithmetic expression using a random number, and saves and releases the metadata blocks encoded by the erasure correction type rateless code as two cache blocks. That is, the search node functions as a cache node after acquiring the original file. When a search node generates and publishes a cache block as a cache node, an arithmetic expression is newly generated using a random number. Therefore, the arithmetic expression (arithmetic equation) can be compared with the arithmetic expression generated by the original node only with a very low probability. Does not match.

  In order to avoid concentration of access to the original node, a node that newly searches for a file tries to acquire a metadata block (cache block) in preference to the cache node over the original node. In addition, it is assumed that a metadata block (cache block 9 is tried to be acquired from a cache node as close as possible to its own search node among a plurality of caches.

[First file transfer]
FIG. 8 shows the first file transfer. The node N1 discloses the original file that is divided into blocks b1 to b5, and is the original node O1. The node N9 that tries to acquire the file becomes the search node S9 and makes a file request to the network. At this time, since there is no cache node for the file requested by the search node 9, the file is requested from the original node O1.

The original node O1 generates and transmits metadata blocks m1 to m6. At this time, the node N4 becomes a relay node, and transfers data from the original node O1 to the search node S9.
The relay node N4 after the transfer stores two arbitrarily selected metadata blocks m1 and m5 among the six relayed metadata blocks m1 to m6 as cache blocks, and becomes the cache node C4.

  After receiving six (six types) metadata blocks m1 to m6, the search node S9 performs an operation for solving the simultaneous equations from these metadata blocks m1 to m6 and their arithmetic expressions, and obtains the original file. Restore. The search node S9 re-encodes the original file acquired by the restoration using a randomly determined arithmetic expression, generates two metadata blocks m7 and m8 to be cached, and stores them. That is, the search node S9 becomes the cache node C9. Note that the arithmetic expression for generating the metadata blocks m7 and m8 and the arithmetic expression for generating the metadata blocks m1 to m6 are different.

  When the first file transfer is completed, four (four types) metadata blocks exist as cache blocks on the network (see FIG. 9).

[Second file transfer]
FIG. 9 shows the second file transfer. The node N11 trying to acquire the file becomes the search node S11 and requests the file from the network. Here, since the cache nodes C4 and C9 exist, first, the search node S11 acquires the metadata block from these nodes C4 and C9.

  Specifically, metadata blocks m1 and m5 that are cache blocks are transmitted from the cache node C4 to the search node S11. At this time, the node N5 becomes a relay node and transfers to the search node S11. At the same time, the relay node N5 stores the relayed metadata blocks m1 and m5 as cache blocks and becomes the cache node C15.

  Further, the metadata blocks m7 and m8, which are cache blocks, are transmitted from the cache node C9 to the search node S11. At this time, the node N10 becomes a relay node and transfers to the search node S11. At the same time, the relay node N10 stores the relayed metadata blocks m7 and m8 as cache blocks and becomes the cache node C10.

  Further, as described above, the search node S11 acquires four types of metadata blocks m1, m5, m7, and m8 from the cache node, but six types are necessary for original file restoration. Therefore, the search node S11 requests the remaining two types (two) (or more) metadata blocks from the original node O1.

  The original node O1 generates two new metadata blocks m9 and m10 by an arithmetic expression determined at random, and transfers them to the search node N11 together with the arithmetic expression. At this time, the node N5 becomes a relay node and transfers to the search node S11. At the same time, the relay node N5 discards either one of the metadata blocks m1 and m5 (m1 in this case) held so far, and either one of the metadata blocks m9 and m10 relayed this time. (Here, m9) is saved. Note that the block m9 is discarded by overwriting the block m9 in the storage area of the block m1. Which of the relayed blocks is to be stored may be determined randomly, or may be determined according to a predetermined policy (storage rule). In addition, a block to be discarded (overwritten) may be determined at random or according to a predetermined policy (discard rule).

  The search node S11 receives six (six types) metadata blocks m1, m5, m7, m8, m9, and m10, and then these metadata blocks m1, m5, m7, m8, m9, and m10 and their operations. Perform an operation to solve the simultaneous equations from the equation and restore the original file. The search node S11 re-encodes the original file acquired by the restoration using a randomly determined arithmetic expression, generates two new metadata blocks m11 and m12 to be cached, and stores them. That is, the search node S11 becomes the cache node C11.

When the second file transfer is completed, 10 (seven types) metadata blocks exist as cache blocks on the network (see FIG. 10). In this way, it is possible to obtain a state where there are more types of metadata blocks on the network than the number of types of metadata blocks (= 6 types) necessary for restoring the original file. After this state is reached, the original file can be restored simply by acquiring the cache block from the cache node.
That is, even if the node O1 having the original file itself does not exist on the network, a sufficient number (type) of metadata blocks can be obtained for decoding the original file.

As described above, since a search node for newly acquiring a file generates a metadata block calculated by an arithmetic expression (arithmetic equation) different from that at the time of decoding, each time a file is acquired, the metadata block The number increases.
The original file O1 also generates a metadata block calculated by a new calculation expression (calculation equation) every time there is a file request or metablock request. Therefore, every time a file request or metablock request is made, The types of data blocks are increasing.

[Network after N file transfers]
If the above file transfer is performed N times, as shown in FIG. 11, it is assumed that there are twelve eight types of metadata blocks as cache files in addition to the original file on the network.

[Case where some nodes leave after N times of file transfer]
FIG. 12 shows a case where the original node O1 and the cache node C6 have left the network after N times of file transfers. Note that node detachment occurs when the computer is turned off or the program is uninstalled.
Even after the nodes O1 and C6 leave, there are ten types of metadata blocks m7, m8, m14, m15, m16, and m17 on the network as cache blocks.

Even in the state of FIG. 12, since the number of types of metadata blocks (six types) necessary for original file restoration (decryption) is sufficient, even if a node (search node S9) for acquiring a new file appears, It can be acquired.
In other words, in the present embodiment, the original file is converted to metadata, and if any six (six types) metadata blocks can be acquired, the original file can be restored, and the file retention efficiency of the network is increased.

[Comparative example]
Hereinafter, the high file retention efficiency in the network of the present embodiment will be described in comparison with the comparative examples shown in FIGS. In the comparative example, the original file is transferred without being converted into metadata. That is, in this comparative example, not the metadata block but the five original blocks b1 to b5 obtained by dividing the original file are transferred.
The original block is stored in the cache node as a cache block.

In the comparative example, a search node that requires a file needs to align all five blocks b1 to b5. That is, the blocks b1 to b5 are all indispensable blocks for restoring the original file.
In the comparative example, other points such as the transfer rule are the same as those of the network of the present embodiment.

[First file transfer in comparative example]
FIG. 13 shows the first file transfer. The node N1 discloses the original file that is divided into blocks b1 to b5, and is the original node O1. The node N9 that tries to acquire the file becomes the search node S9 and makes a file request to the network. At this time, since there is no cache node for the file requested by the search node 9, the file is requested from the original node O1.

The original node O1 transmits original blocks b1 to b5. At this time, the node N4 becomes a relay node, and transfers data from the original node O1 to the search node S9.
The relay node N4 after the transfer stores the two blocks b1 and b5 among the six relayed original blocks m1 to m6 as cache blocks, and becomes the cache node C4.

  The search node S9 acquires the original file by receiving the five blocks b1 to b5. The search node S9 stores the blocks b2 and b4 as cache blocks and becomes the cache node C9.

[Second file transfer in comparative example]
FIG. 14 shows the second file transfer. The node N11 trying to acquire the file becomes the search node S11 and requests the file from the network. Here, since the cache nodes C4 and C9 exist, first, the search node S11 acquires four blocks b1, b2, b4, and b5 from these nodes C4 and C9.

  Specifically, blocks b1 and b5 that are cache blocks are transmitted from the cache node C4 to the search node S11. At this time, the node N5 becomes a relay node and transfers to the search node S11. At the same time, the relay node N5 stores the relayed blocks b1 and b5 as cache blocks and becomes the cache node C15.

  The cache node C9 transmits the blocks m2 and m4, which are cache blocks, to the search node S11. At this time, the node N10 becomes a relay node and transfers to the search node S11. At the same time, the relay node N10 stores the relayed metadata blocks b2 and b4 as cache blocks and becomes the cache node C10.

Also, the original node O1 transfers the missing block b3 to the search node 11. At this time, the node N5 becomes a relay node and transfers to the search node S11. At the same time, the relay node N5 discards the cache block b1 held so far and stores the relayed block b3.
After receiving all of b1 to b5, the search node S11 stores the blocks b1 and b3 as cache blocks and becomes the cache node 11.

  As shown in FIG. 15, in the comparative example, when the second file transfer is completed, there are 10 cache blocks on the network, and each block b1 to b5 exists in any one of the cache nodes. is doing. Therefore, after that, the original file can be acquired only by acquiring the cache block from the cache node.

[Network after N file transfers in comparative example]
If the file transfer as described above is performed N times in the comparative example, as shown in FIG. 16, there are 12 (five types) cache files in addition to the original file of the original node O1 on the network. To do.

[Case where some nodes leave after file transfer N times in the comparative example]
FIG. 17 shows a case where the original node O1 and the cache node C6 have left the network after N times of file transfers.
However, the block b5 that has only the cache node C6 does not exist on the network. That is, even if there are 10 cache blocks on the network, there are only four types of blocks b1 to b4, and the search node S9 cannot arrange all the necessary blocks b1 to b5. Therefore, the original file cannot be restored on this P2P network.
Thus, in the comparative example, all of the blocks b1 to b5 are essential for file restoration, and even if there are many cache blocks, the file restoration cannot be performed only by lacking one kind of cache block.

  In FIG. 17 of the comparative example and FIG. 12 of the present embodiment, there are a total of 10 cache blocks, but the original file cannot be restored in FIG. 17, and the original file can be restored in FIG. That is, in FIG. 12 of the present embodiment, the file can be restored with a relatively small number of cache blocks, and it can be said that the file sharing efficiency in the network is high.

In the comparative example, the maximum number of types of cache blocks that can exist on the network is equal to the number of original blocks (= 5) obtained by dividing the original file. In other words, there are no more cache block types on the network than are necessary to restore the original file.
On the other hand, in the present embodiment, an infinite number of arithmetic expressions for generating metadata blocks can be generated, so that there are an infinite number of types of metadata blocks that can exist on the network. Therefore, the number of cache blocks on the network is larger than the number (type) necessary to restore the original file. In addition, each piece of metadata is optional, not essential for file restoration. As a result, even if the original node or a part of the cache node leaves the network, the probability that the original file can be restored becomes very high.

In the comparative example, the types of original blocks for a certain original file are limited. In the above example, there are only five types (= number of divisions). Therefore, even if many cache nodes hold cache blocks, only the number of nodes holding duplicate blocks increases.
On the other hand, in the present embodiment, since it is converted to metadata, the number of types of blocks is virtually unlimited, and the possibility of holding duplicate blocks is reduced. In this embodiment, file sharing efficiency is also high from this point.

  Hereinafter, each function in the node will be described in more detail.

[Details of original node processing]
Hereinafter, the processing of the original node will be described with reference to FIGS.

[Step S11: Release of the original file]
When the file to be disclosed by the user is placed in the original file storage unit 11 (upload folder) of the original node, the file (original file) is disclosed to the network (step S11). When the file is published, the original node creates file information (key information) of the file. The file information includes a file name, a file identifier (file hash value), a file size, original file position information (original node IP address and port number), and the like.

  Hereinafter, the original file itself is referred to as “file body”, and the file body and the file information of the file body are collectively referred to as “original file”. FIG. 19A shows an original file in which file information is added to the file body. Since the original file includes file information, information sharing between nodes for file transfer is facilitated.

[Steps S12, S13, S14: File Request Waiting and Metadata Generation]
The original node waits until a request for a file (or metadata) is received from another node (search node). When the original node receives the file request (steps S12 and S13), the original node performs a metadata generation process.
Specifically, the file request is received by the P2P network communication processing unit (file request receiving means) 21. When the file request is received, the control unit 20 starts the metadata generation process.

  Hereinafter, the metadata itself generated from the original file is referred to as “metadata body”, and the metadata body, file information, and arithmetic expression information are collectively referred to as “metadata”.

  FIG. 19B shows metadata including file information and arithmetic expression information. The metadata generation unit 12 of the present embodiment generates the metadata of FIG. 19B from the original file of FIG. 19A (step S14).

The metadata generation policy in the metadata generation unit 12 in the present embodiment is as follows. First, the number of arithmetic expressions for generating metadata is required to be at least the number corresponding to the number of metadata necessary for restoration (number of types). The number of metadata necessary for restoration is calculated as [number of original blocks × redundancy (for example, about 105% of the number of original blocks)] or more. For example, if the number of original blocks is three, the arithmetic expression generation unit 12a of the metadata generation unit 12 randomly determines four arithmetic expressions to generate four (four types) metadata.
Note that the redundancy is appropriately set as a value that can solve the simultaneous equations including a plurality of arithmetic expressions and restore the original block.

The arithmetic expression generation unit 12a is configured by, for example, an arithmetic expression library shown in FIG. The arithmetic expression library randomly determines a necessary number of arithmetic expressions when a combination of the file size of the original file and a random number is given as an argument.
For example, it is assumed that when the size of the original block is set as 64 KB, 180 KB is given to the arithmetic expression library as the original file size. In this case, the number of original blocks is 3, and the integer value obtained by multiplying this by the redundancy (105%) and rounding up after the decimal point is 4. The arithmetic expression library randomly determines four arithmetic expressions based on random numbers as arguments. For example, as shown in FIG. 7, the generated arithmetic expression is preferably one in which one or a plurality of original blocks are operated with an XOR operator.

In the above description, the original file size is given to the library to determine the number of arithmetic expressions. However, the number of original blocks or the number of necessary arithmetic expressions themselves may be given, and the number of arithmetic expressions can be determined. Any information that can be used is not particularly limited.
Further, when the library itself can generate a random number, the random number may not be given to the library.

The metadata calculation unit 12b of the metadata generation unit 12 calculates the metadata main body by calculating between the original blocks according to a randomly determined calculation formula.
Further, the metadata generation unit 12 adds the file information of the original file and the calculation formula information indicating the calculation formula used for generating the metadata main body to each metadata main body, thereby obtaining metadata m1 to m4. .

The arithmetic expression information may be the arithmetic expression itself or an identifier for specifying the arithmetic expression. Here, in the arithmetic expression library, an identifier that identifies the arithmetic expression is generated together with the arithmetic expression, and therefore the identifier is added to the metadata as arithmetic expression information.
Each node has a common arithmetic expression library, and each node can obtain an arithmetic expression specified by the identifier by giving the identifier to the arithmetic expression library (see FIG. 20).

[Steps S15 and S16: Metadata transmission]
When the metadata m1 to m4 are generated, the metadata transmission unit 13 performs file transmission to the search node that has requested the file in units of individual metadata (metablocks).

[Transmit metadata when the number of metadata is specified]
The original node searches for the metadata request with the designation of the number of metadata (number of types) as in the generation / transfer of the metadata blocks m9 and m10 shown in FIG. 9 instead of the request for the entire file as described above. The metadata generation unit 12 generates a specified number (two) of arithmetic expressions and transmits the two pieces of metadata m9 and m10 to the search node. In this case, a specified number is given to the arithmetic expression library instead of the original file size, and a number of arithmetic expressions corresponding to the specified number are generated.
Note that a metadata request accompanied by designation of the number of metadata is received by the P2P network communication processing unit 21.

[Metadata randomness]
The arithmetic expression generation unit 12a formed of an arithmetic expression library determines the arithmetic expression using a random number every time a metadata request with a file request or number specification is received, and therefore a metadata request with a file request or number specification. Each time it is received, substantially different metadata can be generated. Therefore, since various metadata are generated and transmitted from the same original file, the number of types of metadata on the network increases innumerably.
That is, when file transfer is performed a plurality of times, it is possible to easily obtain a state in which the number of types (number of types) of metadata necessary for restoring the original file exists on the network.

Note that the original node is the type of metadata generated in the past and / or the metadata of the original file and is present on the network when the metadata generation for the original file is performed for the second time or more. It is preferable to check the kind of thing. In the unlikely event that the newly generated metadata type overlaps with the previously generated metadata type and / or the original file metadata that currently exists on the network, the duplicate metadata Data is not transmitted, and by regenerating metadata that does not overlap, a predetermined number of metadata can be secured and transmitted.
In the case of a metadata request with the number of metadata, the original node receives information indicating the type of metadata that the search node has already acquired from the search node, and the type of metadata that the search node has and Send the required number of unique metadata.

[Details of relay node processing]
Hereinafter, the processing of the relay node will be described with reference to FIGS. 21 and 4.

[Steps S21 to S23: Reception of Relay File]
The relay node relays metadata and then functions as a cache node.
First, when the P2P network communication processing unit determines that the metadata transfer relay is necessary for file transfer from another node to another node (steps S21 and S22), the metadata receiving unit of the relay node 31 (metadata receiving unit 41) receives the metadata in units of blocks (step S23).

[Step S24: Save Cache Block]
The metadata storage unit 32 (metadata storage unit 42) of the relay node can store a certain number of blocks as cache blocks. Here, it is assumed that the metadata storage unit 32 can store two metadata blocks (number smaller than the number of metadata necessary for file restoration) for each original file.
The relay node stores, in the metadata storage unit 32, a randomly determined number (maximum number that can be stored in the metadata storage unit 32) of the metadata blocks received for relaying.

[Steps S25 and S26: Transmission of metadata]
Further, the metadata transmission unit 33 of the relay node transmits all the metadata received for the relay to other nodes.

[Step S27: Make the relay node a cache node]
When the relay is completed, the relay node publishes the metadata stored in the metadata storage unit 32 (metadata storage unit 42: cache folder) on the network as a cache block.
For example, the relay node receives four pieces of metadata (metadata blocks) m1 to m4, and among these blocks m1 to m4, two blocks m1 and m4 randomly selected by the control unit 20 A state stored in the data storage unit 32 (metadata storage unit 42) is shown in FIG.

  As shown in FIG. 22, when the metadata m1 and m4 are disclosed as cache blocks, metadata information such as metadata location information is added to the metablocks m1 and m4. The metadata location information indicates the location of the metablocks m1 and m4 on the network, and is configured by the IP address and port number of the cache node.

  File information, metadata location information (metadata information), and calculation formula information if necessary are periodically exchanged between adjacent nodes on the network as key information. Therefore, the key information gradually spreads over the network. That is, each node has a function of notifying key information to other nodes and a function of receiving key information from other nodes. Also, these functions are not functions unique to the relay node, but can function with other types of nodes.

The key information is accumulated in a key table (key storage unit; not shown) of each node.
As a result, each node can grasp which node on the network has what original file or what metadata. The key information is exchanged by the P2P network communication processing unit 21.

[Example of metadata cache blocking policy]
For example, the control unit 20 stores metadata that is a cache block in accordance with the following policy.

[Policy 1: New saving of metadata]
When the metadata storage unit 32 (metadata storage unit) does not have metadata about a certain original file as a cache block, the maximum number of metadata stored as a cache block is stored in the metadata storage unit 32. Equal to the number of metadata that can be stored.
Part or all of the relayed metadata is stored as a cache block, and when a part of the relay metadata is stored, the stored metadata is determined at random (see FIGS. 8 and 9).

[Policy 2: Overwrite metadata]
When the metadata storage unit 32 (metadata storage unit) has metadata about a certain original file as a cache block, the metadata stored in the metadata storage unit 32 is replaced with the relayed metadata. (See FIGS. 9 and 10).
The metadata replacement may be performed for all metadata in the metadata storage unit 32, but as shown in FIGS. 9 and 10, a part of the metadata of the relayed metadata is stored in the metadata storage unit 32. It is preferable to replace a part of the cache block.

Further, the control unit 20 determines which metadata to be overwritten from among the metadata already stored in the metadata storage unit 32, that is, the metadata already stored in the metadata storage unit 32. Select which metadata of the data to discard.
The metadata to be discarded may be determined at random by the control unit 20, but it is preferable that metadata of the type held by other cache nodes is preferentially discarded. By preferentially discarding metadata held by a plurality of cache nodes, it becomes easy to obtain a variety of metadata in the network.
Each node can grasp, by the P2P network communication processing unit 21, which cache node has which original file and what type (metadata type determined by the arithmetic expression). That is, the P2P network communication processing unit 21 is a unit that acquires the type of metadata from another node, and is also a unit that notifies the other node of the type of metadata.

In addition, even if a cache node overwrites metadata, the cache block (metadata) held by other adjacent cache nodes is not subject to discarding types that are not held by other cache nodes. ) May be included in at least one metablock of its own metadata storage unit 32 (metadata storage unit 42).
If there is no metadata to be discarded determined by the above policy, any metadata may be forcibly determined as a discard target, or the metadata may not be overwritten and saved. .

[Policy 3: Determination of metadata to save]
The control unit 20 may include a meta block that is as different as possible from the cache blocks (metadata) held by other adjacent cache nodes in its own metadata storage unit 32 (metadata storage unit 42). Next, the metablock to be stored is determined.

  For example, when storing the metadata relayed by the control unit 20 of the cache node, the type of metadata held by another cache node is compared with the type of metadata received for relaying, and for relaying. From the received metadata, select the metadata that is not held by other cache nodes (excluding the source of the relayed metadata) or the number of cache nodes held is less than the predetermined number as the storage target .

  By doing so, duplication of the types of metadata blocks on the network can be reduced and the file sharing efficiency of the network can be increased.

[Cache node processing details]
Hereinafter, the processing of the cache node will be described with reference to FIGS.

[Steps S31 and S32: File Request Reception]
As shown in FIG. 23, when the cache node publishes the metadata as a cache block, the cache node waits for a file request from another node (search node) (step S31).

[Steps S33 to S36: Metadata transmission]
When the P2P network communication processing unit 21 receives a file request from the search node (step S32), the control unit 20 checks whether or not the metadata storage unit 42 has a cache block (step S33). If there is a cache block in the metadata storage unit 42, it is determined whether or not the cache block is metadata of a file requested by the search node (step S34).
If the cache block held by the cache node matches the one requested by the search node, the cache block is transmitted to the search node by the metadata transmission unit 43 (steps S35 and S36).
When the search node receives a metadata request accompanied by designation of the metadata type, the cache node transmits the requested type of metadata if it is held.

In addition, the search node may make a file request together with metadata operation formula information (unnecessary metadata information) already possessed. In this case, even if the metadata of the file requested by the search node is stored in the metadata storage unit 42, if the metadata already exists in the search node, the control unit displays the metadata as the search node. Therefore, the metadata is not transmitted.
On the other hand, if the metadata is not duplicated, the cache node transmits the metadata in the metadata storage unit 42 to the search node.

[Details of search node processing]
Hereinafter, the processing of the search node will be described with reference to FIGS.

[Steps S41 and S42: File Request and Search]
When a file request is processed by a user or the like in the search node (step S41), the P2P network communication processing unit 21 of the search node transmits a query for searching a cache block or an original file in order to obtain a file from the network. (Step S42). The query is transferred between the nodes, and by referring to the key table of each node, the location information of the cache node having the metadata of the file to be obtained as a cache block and the location information of the original file are obtained. The query is transferred back and forth between the nodes and then sent back to the search node.

The location information of the cache node can be acquired from the metadata location information of the metadata. Thus, it can be said that the query collects metadata in each node or key information (additional information) of the original file.
Further, since the key information acquired by the query includes the file size of the file to be acquired, the search node can grasp the file size of the file to be acquired.

[Steps S43 to S49: Acquisition of Required Number of Metadata]
Based on the returned query, the search node control unit 20 determines whether there is a cache node that holds the metadata of the file to be acquired as a cache block (step S43).
If there is a cache node that holds the necessary metadata, the search node requests a cache block from the closest cache node among those nodes (step S44), and the metadata receiving unit 51 determines the metadata block. Is received (step S45) and stored in the metadata storage unit 52.

The search node acquires the file size of the original file to be acquired from the key information of the returned query. When the key information includes the number of original blocks or the number of types of metablocks necessary for file restoration, the search node may acquire these pieces of information.
This is because the original file size, the number of original blocks, and the number of metablock types necessary for restoration can all be information used for specifying the number of metablock types necessary for restoration.

  When a cache block is sent from the cache node, the search node determines whether or not the cache block (metadata) is of a different type from the metadata already held (whether the arithmetic expression is different). If the type is different and the type is different, the acquired cache block is used as a metadata block used for restoration, and if the type is the same, the acquired cache block is discarded.

The search node repeats the above cache block acquisition process until the number of metablock types (for example, four types or more) necessary for restoring the original file is obtained.
That is, the cache blocks are acquired in order from the cache node closest to the search node (in terms of the logical path) (step S44), and the cache node is gradually cached until the necessary number (number of types) of metadata blocks is obtained. A block is requested (step S47).
The number of metablock types necessary for restoration is calculated by obtaining the original file size from the key returned to the search node and multiplying the search node by a predetermined redundancy (for example, 105%).

As a result of the processing in steps S44 to S47, when the necessary number of metadata blocks are not prepared but there is no cache node that holds the metadata of the original file to be acquired as a cache block, the search node returns. The presence of the original node is determined based on the query (step S48). Also, if there is no cache node from the beginning (step S43), the presence of the original node is determined (step S48).
If it is determined in step S48 that the original node does not exist, file acquisition fails because the required number of metadata blocks are not complete (step S50).

If it is determined in step S48 that the original node exists, the search node requests and acquires metadata blocks for the shortage number (shortage type number) from the original node (step S49).
Note that the processing when the original node receives a metadata request with the designation of the number of metadata (number of missing types) is the same as described in [Sending metadata when the number of metadata is designated]. is there.

[Steps S51 to S55: File Restoration]
When the number of metadata blocks necessary for file restoration is obtained in the search node in step S46 or S49 (step S51), the metadata is decrypted (step S52). First, the search node obtains an arithmetic expression based on arithmetic expression information (arithmetic expression identifier) included in the metadata. That is, the arithmetic expression generation unit 53a of the metadata decoding unit 53 of the search node obtains an arithmetic expression from the arithmetic expression identifier of the acquired metadata using the arithmetic expression library shown in FIG.

  Then, the metadata decoding operation unit 53b performs a decoding operation from a predetermined number of arithmetic expressions and metadata. Decoding is performed by calculating an original block by solving simultaneous equations of a plurality of arithmetic expressions and metablocks.

  If the decoding in step S52 fails, that is, if the collected metadata number (number of types) is not sufficient for restoration, an additional metadata block is acquired (step S54). Additional block acquisition is realized by performing the processing of steps S43 to S49 again.

  If the decoding is successful and the original block (original file) can be restored, the file acquisition is completed (step S55). The restored original file is stored in the original data storage unit 54. The restored original file (file body and file information) itself may be disclosed to the network, but in the present embodiment, it is disclosed as a cache block as follows.

[Steps S56 and S57: Metadata Regeneration]
The metadata generation unit 55 of the search node has the same function as the metadata generation unit 12 of the original node, and newly generates metadata from the original file in the original data storage unit 54.
That is, the arithmetic expression generation unit 55a of the metadata generation unit 55 generates an arithmetic expression different from the arithmetic expression of the received metadata by using a random number, using the arithmetic expression library. Here, the number of arithmetic expressions to be generated is a number (for example, 2) corresponding to the number of blocks stored as cache blocks in the metadata storage unit 56.
When the generated arithmetic expression and the received metadata arithmetic expression overlap, it is preferable to regenerate the arithmetic expression so that it does not overlap with the received metadata arithmetic expression. Further, the metadata calculation expression stored on the network may be compared with the generated calculation expression so that they do not overlap.

  From the generated arithmetic expression and the original block obtained by dividing the original file (file body), the metadata operation unit 55b performs an operation for generating metadata (step S56), and stores it as a cache block in the metadata storage unit 56. , It is made public on the network (step S57). That is, the search node becomes a cache node after file acquisition.

  In addition, this invention is not limited to the said embodiment, A various deformation | transformation is possible. For example, the original node may function as a cache node after transmitting metadata.

1 is a schematic diagram of a P2P file sharing network system indicated by nodes and logical paths. FIG. It is a block diagram of a node computer. It is a functional block diagram of an original node. It is a functional block diagram of a relay node. It is a functional block diagram of a cache node. It is a functional block diagram of a search node. It is a figure which shows the method of producing | generating metadata from an original file. It is a figure which shows the file transfer (1st transfer) in the network of this embodiment. It is a figure which shows the file transfer (2nd transfer) in the network of this embodiment. It is a figure which shows the state of the cache node after the completion of the 2nd file transfer in the network of this embodiment. It is a figure which shows the state of the cache node after completion | finish of the Nth file transfer in the network of this embodiment. In the network of this embodiment, it is a figure which shows the state from which node O1, C6 dropped out from the network after the Nth file transfer completion. It is a figure which shows the file transfer (1st transfer) in the network of a comparative example. It is a figure which shows the file transfer (2nd transfer) in the network of a comparative example. It is a figure which shows the state of the cache node after the completion of the 2nd file transfer in the network of a comparative example. It is a figure which shows the state of the cache node after completion | finish of the Nth file transfer in the network of a comparative example. In the network of a comparative example, it is a figure which shows the state from which node O1, C6 dropped out from the network after the Nth file transfer completion. It is a flowchart which shows the process sequence of an original node. It is a figure which shows the detailed data structure of an original file and metadata. It is a figure which shows an arithmetic expression library. It is a flowchart which shows the processing procedure of a relay node. It is a figure which shows the metadata memory | storage part of a cache node. It is a flowchart which shows the processing procedure of a cache node. It is a flowchart which shows the process sequence of a search node.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11 Original file storage part 12 Metadata production | generation part 12a Arithmetic expression generation part 12b Metadata calculation part 13 Metadata transmission part 20 Control part 21 P2P network communication processing part 31 Metadata reception part 32 Metadata storage part 33 Metadata transmission part 41 Metadata reception unit 42 Metadata storage unit 43 Metadata transmission unit 51 Metadata reception unit 53 Metadata decoding unit 55 Metadata generation unit 56 Metadata storage unit

Claims (38)

In a file sharing network system,
A metadata generation unit that generates a plurality of metadata encoded so as to be erasure-correctable from the shared original file;
A metadata decoding unit that acquires and decodes the metadata to acquire the original file and restores the original file;
A file sharing network system characterized by comprising:   The file sharing network system according to claim 1, wherein the metadata generation unit generates metadata by an erasure correction type rateless code. The metadata generation unit performs encoding with an erasure correction code on a plurality of blocks obtained by dividing the original file, and generates non-essential metadata for restoration.
3. The file sharing network system according to claim 1, wherein the metadata decoding unit restores the original file by acquiring as many metadata as are necessary for restoration as necessary. A data structure used for file transfer in a file sharing network system,
A metadata body that encodes a shared original file so that it can be lost, and
Formula information for encoding used to generate the metadata body,
A data structure characterized by having   5. The data structure according to claim 4, further comprising location information on the network where the metadata exists.   6. The data structure according to claim 4, wherein the metadata further includes file information related to an original file that is a basis for generating the metadata. A file transfer method in a file sharing network system, comprising:
Generating a plurality of metadata encoded so as to be erasure-correctable from a shared original file, and enabling transmission to a network;
A file acquisition computer acquires the metadata from a network and restores the original file;
A file transfer method in a file sharing network system. A file transfer method in a file sharing network system, comprising:
Generating a plurality of metadata encoded to be erasure-correctable from the shared original file;
A plurality of cache computers on the network holding the metadata in a distributed manner;
A file acquisition computer acquires the metadata from a cache computer and restores the original file;
Including
A file transfer method for generating metadata such that the number of types of metadata held by a plurality of cache computers on the network is greater than the number of types of metadata required to restore the original file .   9. The file transfer method according to claim 8, wherein the cache computer notifies another cache computer of the type of metadata held by the cache computer.   The file transfer method according to claim 7, wherein a different arithmetic expression for encoding is used every time metadata is generated from an original file. A file publishing computer that publishes files on a network for file sharing,
A metadata generation unit for generating metadata encoded so as to be erasure-correctable from the original file to be shared;
A transmission unit for transmitting the generated metadata to the network;
A file publishing computer characterized by comprising: The metadata generation unit
An arithmetic expression generator for randomly determining an arithmetic expression for generating metadata;
A metadata operation unit that performs an operation for generating metadata using an operation expression randomly determined by the operation expression determining unit;
The file publishing computer according to claim 11, further comprising:   13. The file publishing computer according to claim 12, wherein the arithmetic expression generating unit determines the number of arithmetic expressions to be generated according to the file size of the original file.   12. The metadata generation unit performs encoding with erasure correction codes on a plurality of blocks obtained by dividing the original file, and generates a plurality of metadata that is not essential for restoration. The file public computer in any one of -13.   The metadata generation unit generates the metadata from the original file every time there is an original file transmission request or a metadata transmission request from another computer on the network. The listed file publishing computer.   16. The file publishing computer according to claim 15, wherein the metadata generation unit generates different metadata from the original file every time there is an original file transmission request or a metadata transmission request from another computer on the network. .   17. The file publishing computer according to claim 11, further comprising means for transmitting information indicating the existence of the original file to be shared to another computer on the network.   The computer program for functioning a computer as a file public computer in any one of Claims 11-17. A cache computer capable of holding a cache block for file sharing on a network,
A metadata storage unit that stores, as a cache block, metadata obtained by encoding an original file to be shared so as to be erasure-correctable;
A transmission unit for transmitting the metadata stored in the metadata storage unit to the network;
A cache computer comprising:   20. The cache computer according to claim 19, wherein the metadata storage unit stores a smaller number of metadata than is necessary for restoring the original file. Relay means for relaying one or more metadata transmitted from a computer on the network to other computers on the network;
21. The cache computer according to claim 19, wherein the metadata storage unit stores part or all of one or a plurality of metadata received for relay.   The metadata to be stored in the metadata storage unit is selected from one or a plurality of metadata received for relay, and the metadata to be discarded among the metadata already stored in the metadata storage unit. The cache computer according to claim 21, further comprising means for selecting data.   The cache computer according to any one of claims 19 to 22, further comprising means for acquiring a type of metadata held by another cache computer on the network.   The cache computer according to any one of claims 19 to 23, further comprising means for notifying another cache computer on the network of the type of metadata held by itself.   The metadata to be stored in the metadata storage unit is selected so that at least one type of metadata different from the metadata held by other adjacent cache computers is included in its own metadata storage unit. The cache computer according to any one of claims 19 to 24.   26. The metadata transmitted from the transmission unit to the network is metadata of a type requested by another computer among the metadata stored in the metadata storage unit. A cash computer according to the above.   Receiving unnecessary metadata information indicating a type of unnecessary metadata to be transmitted from another computer on the network, and transmitting metadata not corresponding to the metadata type indicated by the unnecessary metadata information to the network. 27. The cache computer according to claim 19, wherein the cache computer is a cache computer.   A computer program for causing a computer to function as the cache computer according to any one of claims 19 to 28. A file acquisition computer that acquires a shared file from a network,
A metadata receiver that obtains from the network metadata that is encoded so that the original file to be shared can be lost and corrected;
A decryption unit that decrypts the acquired metadata and restores the original file;
A file acquisition computer comprising:   30. The file acquisition computer according to claim 29, further comprising means for searching for metadata existing on the network.   31. The file acquisition computer according to claim 29, wherein the metadata reception unit acquires metadata of a number of types equal to or greater than the number of types of metadata necessary for restoring the original file from the network.   32. The file acquisition computer according to claim 29, wherein information necessary for calculating the number of types of metadata necessary for restoring the original file is acquired from a network.   If the original file restoration fails even if decryption is performed using the acquired metadata, metadata of a different type from the already acquired metadata is additionally acquired from the network. Item 33. The file acquisition computer according to any one of Items 29 to 32.   The metadata is preferentially acquired from a cache computer having metadata as a cache block rather than a file public computer having an original file that has not been converted to metadata. File acquisition computer.   35. The file acquisition computer according to claim 29, wherein metadata is acquired preferentially from other computers connected by a smaller number of logical paths. A metadata generation unit that regenerates metadata from the original file restored by the decoding unit;
36. The file acquisition computer according to claim 29, wherein the generated metadata is made public.   37. The file acquisition computer according to claim 36, wherein the metadata generation unit is capable of generating metadata having a different type from the metadata used for original file restoration.   A computer program for causing a computer to function as the file acquisition computer according to any one of claims 29 to 37.

Images