JP5756959B2 - Distributed archive system and data archive device - Google Patents

Description

  The present invention relates to a distributed archive technology.

  For distributed archives, from the standpoint of data availability and maintenance, a technique called RAID (Redundant Array Of Inexpensive Disks) has been used particularly for the purpose of protecting data from disk failure. For example, in RAID1, when recording data, the same contents are recorded on two independent disks (mirroring), and even if one disk fails, data can be extracted from the other disk. On the other hand, in RAID3, data is distributed and recorded on n disks, but one of the disks is exclusively used for recording parity, and even if one of the n disks fails, the remaining disk remains. The original data can be restored from the (n-1) discs.

  However, when data is distributed and recorded on an archive server on the Internet such as cloud computing, two or more servers are used due to reasons such as Internet failure, server failure, and denial of service attack (Denial Of Service Attack). On the other hand, it is possible to envisage a situation where access cannot be made at the same time, and the RAID system that can cope with only inaccessibility to one server has a problem in data availability and maintenance.

  Further, since the RAID system does not include a mechanism for guaranteeing the confidentiality of archived data, it is necessary to use it together with other encryption techniques, resulting in an overhead in apparatus configuration and execution efficiency.

  On the other hand, in Japanese Patent No. 3607700 (Patent Document 1), a secret sharing method by Shamir (Adi Shamir, How To Share A Secret, Communications Of ACM, 22 (11), pages 612-613, 1979). By using k, it is possible to restore data if it is possible to access k servers out of n, so that the application to the Internet is flexible, and data from k-1 servers is temporarily assumed. There has been described an invention with safety that theoretically guarantees that the original data will not be restored even if it leaks.

  However, in Shamir's secret sharing method, calculation of a power residue with a large amount of calculation is required for each data fragment obtained by dividing the data, and the length of the data fragment should be several thousand bits at most because of the limitation of the computer capacity. Inevitably, the purpose of archiving large data of several megabytes and several gigabytes is not practical from the viewpoint of computational efficiency.

  On the other hand, a method of deciding a secret pattern and repeatedly archiving data fragments in a distributed manner while permitting redundancy is possible according to the pattern, but it is difficult to prove safety. In fact, since the same pattern appears repeatedly, it can be easily assumed that at least a part of the data can be restored by pattern analysis. As long as the security cannot be proved, it is necessary to use it together with other cryptographic techniques, and as described above, it becomes an overhead in apparatus configuration and execution efficiency.

Japanese Patent No. 3607700

Adi Shamir, How To Share A Secret, Communications Of ACM, 22 (11), pages 612-613, 1979

  In view of the problems of the prior art described in the background art, in the present invention, for k and n (0 <k ≦ n) arbitrarily selected, data is distributed and recorded in n servers or storage media. In addition, the following two items related to availability and confidentiality are satisfied, a high-speed data archiving / restoration process is possible, and a theoretical support for confidentiality is given.

  If data archived from k locations among n locations of servers or storage media can be acquired, the original data can be completely restored. This requirement is called n: k threshold availability.

  Even if an attacker can illegally acquire data from k-1 locations of n servers or storage media, the original data cannot be restored from the acquired data. This requirement is called n: k threshold confidentiality.

  In information security, especially cryptography, it is not allowed to claim that a method is safe due to the fact that no effective attack method has been found for the method, and it is impossible to attack based on a certain mathematical model. Only after demonstrating that there is (in fact, a very small probability of successful attack) is allowed to claim security. From this point of view, the most important problem to be solved by the present invention is to achieve both efficient processing and proof of safety in a system that satisfies the above two requirements.

  It should be noted that the present invention should not be understood in a limited manner due to the above-described problems, the contents of which are defined in the claims and will be described in detail below using examples.

  According to the first aspect of the present invention, the pointer means that points to the beginning of the data in the initial state and moves in one direction toward the end of the data as the process proceeds, and a data fragment having a predetermined length from each pointer position is held. Taking out (n−k + 1) storage media to store the data fragment independently and randomly from the n storage media, and duplicating the data fragment on the (n−k + 1) storage media. The pointer position is advanced by the length of the data fragment, and the above process is repeated until the pointer position reaches the end of the data.

  In this configuration, each data fragment is recorded in duplicate on (n−k + 1) storage media. Therefore, when arbitrarily selecting k storage media, at least one storage medium records the data fragment. ing. This ensures n: k threshold availability.

On the other hand, when (k-1) storage media are arbitrarily selected, the probability that the data fragment is included in the storage media is
1-1 / _{nC (k-1)}
Therefore, on average, 1 / _n C _(k−1) data is lost for the entire data, and n: k threshold confidentiality is guaranteed.

  Furthermore, the security of the data according to the first aspect of the present invention is analyzed in detail as follows, and it can be seen that confidentiality in a stronger sense is guaranteed. In fact, the percentage of data fragments covered by archive data that is illegally obtained from a storage medium is not important, and if you cannot estimate where the data fragments in the archive data will appear, It cannot be restored. In order to evaluate the success probability of this estimation by an attacker, the following considerations are made.

N is the total number of final data fragments obtained by dividing the data, and m is the number of data fragments included in a specific storage medium. Since the archive destination of each data fragment is determined independently and randomly, the expected value of the value of m is given by the following equation.
m≈N · (n−k + 1) / n
a is the order from the beginning of the data fragment in the original data, j is the order from the beginning of the data fragment in the archive data held in the storage medium,
Succ (a, j; N, m)
This represents the probability that the jth data fragment in the archive data is a copy of the ath data fragment in the original data. This probability is evaluated by the following formula.
Succ (a, j; N, m)
= _(A-1) C _{(j-1). (N-a)} C _(m-j) / _N C _m
For a given a, when j = [am / (N + 1)], Succ (a, j; N, m) takes the maximum value. However, [x] is an integer obtained by rounding down the decimal part of x. Therefore,
Succ (a, [am / (N + 1)]; N, m)
Is the maximum success probability when the position of the jth data fragment from the beginning in the archive data is assumed to be a (where [α] is the maximum integer smaller than α, Is a Gaussian symbol).

On the other hand, when Succ (a, [am / (N + 1)]; N, m) is approximated using the Stirling formula,
Succ (a, [am / (N + 1)]; N, m) ≈
(1 / π ^1/2 ) · (m / N) · [(N−1) / (a−1) (N−a)] ^1/4
Get. FIG. 1 is a graph showing this approximate expression. When a is close to 1 or N, that is, when the data fragment is close to the beginning or end of the original data, the success probability when the attacker assumes j = [am / (N + 1)] is close to 1. It can be seen that the success probability decreases rapidly when the data fragment moves away from the head or tail. In other words, for a very small number of data fragments close to the beginning and end of the data, an attacker can estimate the position in the archive data with a high probability, but with the majority of other data fragments, the probability is low. This means that the estimation can only be done with. FIG. 2 is a graph showing a result of simulation when N = 100 and m = 75, and FIG. 3 is a graph showing a result of simulation when N = 1000 and m = 750. Both graphs show that the theoretically derived approximation is appropriate. In addition, an important point is the fact that the probability of the estimation success is suppressed to about 5% even when the number is divided into a very small number of N = 1000. For example, when a 1-megabyte file is archived by dividing it into 1-byte data fragments, N = 1000000, and the estimated success probability is extremely small.

  Furthermore, in reality, if the positions of a considerable number of data fragments cannot be estimated, it cannot be said that the data has been partially restored. Therefore, the position of each data fragment must be estimated, and the success probability of correctly estimating the position of all data fragments decreases rapidly as the number of data fragments increases.

  According to the second aspect of the present invention, in relation to the first aspect of the present invention, the length of the data fragment to be extracted is determined independently and randomly at each pointer position, thereby making it possible to further estimate the attacker. Make it difficult.

  According to a third aspect of the present invention, in relation to the first aspect of the present invention, at each pointer position, a data fragment is recorded on (n−k + 1) storage media and the pointer position is advanced, or , It is possible to estimate the attacker by independently and randomly determining whether or not to store random dummy data generated randomly instead of data fragments in any number of storage media and keep the pointer position unchanged. Make it more difficult.

  According to a fourth aspect of the present invention, in relation to the second aspect of the present invention, at each pointer position, a data fragment is recorded on (n−k + 1) storage media and the pointer position is advanced, or , It is possible to estimate the attacker by independently and randomly determining whether or not to store random dummy data generated randomly instead of data fragments in any number of storage media and keep the pointer position unchanged. Make it more difficult.

  According to a fifth aspect of the present invention, in the first aspect of the present invention, in relation to the first aspect of the present invention, a first random number for generating a random number to randomly determine each factor A method is described in which the generation means is held, an independent random number is generated by the first random number generation means at each pointer position, and the archive destination storage medium is determined from the random number by a predetermined method.

  According to a sixth aspect of the present invention, in relation to the fourth aspect of the present invention, in order to determine each factor randomly, the first random number generating means for generating a random number is held, and each pointer position The first random number generating means generates an independent random number, determines an archive destination storage medium based on the random number, and determines the length of the data fragment or dummy data insertion based on the random number. It describes a method for determining the presence or absence, or both the length of the data fragment and the presence or absence of dummy data insertion.

  According to a seventh aspect of the present invention, in relation to the sixth aspect of the present invention, a key recording means for recording a series of random numbers generated by the first random number generating means is held, and the key recording means Based on the series of recorded random numbers, the storage medium from which the data fragment is to be acquired is determined, and based on the series of random numbers, the length of the data fragment, the presence or absence of dummy data, or the data fragment A method is described in which both the length and the presence or absence of dummy data are determined and the original data is restored.

  According to an eighth aspect of the present invention, in association with the fifth aspect or the sixth aspect of the present invention, the second random number generating means for generating a random number is held, and the first random number generating means is a seed. Is used to generate a random number generation algorithm that generates an arbitrary number of random numbers, and the random number generated by the second random number generation means at the start of archiving is input to the first random number generation means as a seed. Describes a method of performing distributed archiving based on the random number generated by the first random number generation means.

  According to the ninth aspect of the present invention, in relation to the eighth aspect of the present invention, the key recording means for recording the seed generated by the second random number generating means is held and recorded on the key recording means. The configuration is described in which the seed is input to the first random number generation means, the storage medium from which the data fragment is obtained is determined from the series of random numbers generated by the first random number generation means, and the original data is restored. .

  According to the tenth aspect of the present invention, in relation to the first to ninth aspects of the present invention, a part or all of the storage medium for recording data fragments is a server that can be accessed via a network. Describes a configuration that assumes an Internet archive service.

  According to the eleventh aspect of the present invention, in connection with the seventh or ninth aspect of the present invention, the key recording means is provided in a device that can be carried by the user who owns the data. A configuration is described in which it is possible to perform data archiving and decryption processing using the host computer, and to improve the convenience for the user.

  According to a twelfth aspect of the present invention, in relation to the eleventh aspect of the present invention, the device that the user can carry is an IC card, a mobile phone, or a PDA (personal digital assistant). Describes the configuration.

  According to a thirteenth aspect of the present invention, in relation to the eleventh aspect of the present invention, the device portable by the user holds a microcomputer, and the microcomputer controls access to data in the device. And a configuration is described in which a part or all of the data archiving and restoring process is executed.

  According to a fourteenth aspect of the present invention, in relation to the thirteenth aspect of the present invention, in the invention according to the thirteenth aspect, the device that can be carried by the user includes a first random number generation means and a second random number generation means. The configuration is characterized in that either or both of the random number generation means are held.

  According to a fifteenth aspect of the present invention, in connection with the fourteenth aspect of the present invention, the microcomputer held by the user-carryable device and the random number generating means are an IC chip having tamper resistance. A configuration by a so-called smart chip, which is characterized by being mounted inside, is described.

  According to the sixteenth aspect of the present invention, in relation to the first to fifteenth aspects of the present invention, using the above properties, dummy data of a small size is added to the beginning and end of the original data. Thus, a method is described in which the original data is included in a region that is difficult for an attacker to estimate, thereby improving the airtightness of the data.

  According to a seventeenth aspect of the present invention, in a data archiving apparatus for archiving archiving target data to n (n is an integer of 2 or more) archiving destination storage means, a plurality of archiving target data are determined in a predetermined procedure from the archiving target data A data fragment, and at least a plurality of extracted data fragments including all of the archive target data, and each of the plurality of data fragments extracted by the extraction unit Data that is cut out by the cut-out means, having an association means that randomly relates to n−k + 1 (k is a positive integer less than n) archive-destination storage means independently of the archive destination storage means Each fragment is duplicated in n−k + 1 archive destination storage means associated with the association means. Data archive device for causing archived Te is achieved.

  According to an eighteenth aspect of the present invention, in a data archiving computer program for archiving archive target data to n (n is an integer of 2 or more) archive destination storage means, a computer is stored in the archive target data. A plurality of data fragments cut out in a predetermined procedure, and at least a plurality of data fragments cut out include all of the archive target data, and the plurality of data cut out by the cut-out means Each fragment is made to function as an associating means for independently associating each of the fragments independently with (n−k + 1) (k is a positive integer less than or equal to n) archive destination storage means among the n archive destination storage means. Each of the data fragments cut out by the cut-out means is Data archiving computer program is realized for causing the attached relationship (n-k + 1) archived duplicate pieces of archive destination storage means.

  According to a nineteenth aspect of the present invention, in the restoration device for restoring data archived in n (n is an integer of 2 or more) archive destination storage means by the archive device according to the seventeenth aspect of the present invention, Association information obtaining means for obtaining association information used by the association means to randomly associate each of the data fragments of the data independently with (n−k + 1) archive destination storage means; and For each, the data fragment acquisition means for extracting the data fragment from any of the (n−k + 1) archive destination storage means associated with the data fragment by the association information, and the extracted data fragment as the data fragment Connecting means for connecting in a connecting procedure corresponding to the predetermined procedure for cutting out Data recovery device can be realized, wherein.

  According to the twentieth aspect of the present invention, the archive device according to the seventeenth aspect of the present invention is used for restoring data archived in n (n is an integer of 2 or more) archive destination storage means. In the computer program for data restoration, the computer acquires association information used by the association unit to independently associate each of the data fragments of the data independently with (n−k + 1) archive destination storage units. Association information acquisition means, for each of the data fragments, data fragment acquisition means for retrieving the data fragment from any of (n−k + 1) archive destination storage means associated with the data fragment by the association information; and The extracted data fragment is converted into the data fragment to be cut out in advance. Data recovery computer program for functioning as a connecting means for connecting a connecting procedure corresponding to the order obtained procedure is realized.

  According to a twenty-first aspect of the present invention, the archive target storage unit includes n archive target data (n is an integer of 2 or more), and the n archive target data (n is an integer of 2 or more). In the data archiving method for archiving in the archive destination storage means, the cutout means cuts out a plurality of data fragments from the archive target data in a predetermined procedure, and at least the plurality of cut out data fragments are all of the archive target data And a correlating unit assigns each of the plurality of data fragments cut out by the cutout unit to (n−k + 1) (k is n or less) out of the n archive destination storage units. And a step of randomly relating to the archive destination storage means independently of the archive destination storage means. Each of issued data fragment, the data archiving method for causing archived duplicate n-k + 1 single archive destination storage means associated respectively with the association unit is realized.

  According to the present invention, a distributed archiving and / or data restoration method that achieves both high-speed processing and safety is realized.

It is a principle explanatory view of the present invention. It is explanatory drawing of the simulation result of the principle of this invention. It is explanatory drawing of the other simulation result of the principle of this invention. It is a figure explaining the split key of 1st Example of this invention. It is a figure explaining the number of the division | segmentation keys of the said 1st Example. It is a figure explaining the whole structure of the above-mentioned 1st Example. It is a figure explaining the functional block of the above-mentioned 1st Example. It is a figure explaining the structural example of the correlation part of the above-mentioned 1st Example. It is a figure explaining the more preferable structural example of the correlation part of the above-mentioned 1st Example. It is a figure explaining the operation example at the time of the data archive of the above-mentioned Example. It is a figure explaining the operation example at the time of the data archive of the above-mentioned Example. It is a figure explaining the operation example at the time of the data restoration of the above-mentioned Example. It is a figure explaining the operation example at the time of the data restoration of the above-mentioned Example. It is a figure explaining the 2nd Example of this invention. It is a figure explaining the 2nd Example of this invention. It is a figure explaining the 3rd Example of the present invention. It is a figure explaining the 3rd Example of the present invention.

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

[First embodiment]
The first embodiment of the present invention is an implementation of the basic configuration of the present invention. Here, prior to describing the configuration and operation of the first embodiment, a principle description will be given.

  In the archive system 100 (FIG. 6) of the first embodiment, in order to realize n: k threshold availability and n: k threshold confidentiality, one piece of data is sequentially cut out from the data to be archived. The data fragment is recorded in duplicate on (n−k + 1) storage media among the n storage media. An n: k split key is introduced as information for specifying which (n−k + 1) storage media are selected from the n storage media. An identifier represented by a number is given to the n: k split key, and (n−k + 1) storage media are specified. By randomly outputting n: k split keys independently, (n−k + 1) storage media that are randomly specified independently from n storage media are selected and data fragments are recorded in duplicate. .

FIG. 4 is an example in the case of n = 4 and k = 2. The number in the case of selecting (n−k + 1) storage media from n storage media is _n C _{(n−k + 1)} . In this case, ₄ C _{(4-2 + 1)} = 4 split keys exist. Therefore, consider split keys d_key1, d_key2, d_key3, and d_key4 given identifiers 1 to 4. D_key1 having identifier 1 becomes the target when the first (dst_file1), second (dst_file2), and third (dst_file3) storage media in the four storage media are specified and this split key is selected. The data fragment is recorded in the storage media dst_file1, dst_file2, and dst_file3. The identifier of the n: k split key is output independently and randomly using a random number generator.

  FIG. 5 shows the change in the number of n: k split keys when the values of n and k change. Identifiers for the number of split keys are prepared, and these identifiers are output independently and randomly using a random number generator.

FIG. 6 shows the archive system 100 of the first embodiment as a whole. In this figure, the archive system 100 starts the communication network 30 with the host device 10 and the plurality of storage media 20 _{1 to} 20 _n. Connected. The host device 10 is a device having an information processing function such as a personal computer, a mobile phone, a personal digital assistant, an information home appliance, and the like. The host device 10 includes a source / restoration file storage unit 10a. The host device 10 can communicate with an information device such as a smart card, a mobile phone, or a personal digital assistant, which will be described later, via an interface. The storage media 20 _{1 to} 20 _n receive and manage data transmitted through various communication protocols via the communication network 30 and are, for example, internal or external storage devices of the server computer.

  The host device 10 implements an archive controller 40 that controls data archiving and restoration, for example, by installing a computer program 10b. A part of the archive controller 40 may be realized by a smart card or the like. The computer program 10b may realize only one of the function of archiving data and the function of restoring data.

At the time of archiving, the archive controller 40 of the host device 10 sequentially extracts data fragments from the archive target source file (also simply referred to as data) stored in the source / restore file storage unit 10a, According to the n: k split key output randomly at random, recording is performed redundantly on (n−k + 1) storage media that are randomly specified independently from the _n storage media 20 _{1 to} 20 _n. . Further, at the time of restoration, the archive controller 40 randomly specifies (n−k + 1) pieces of independent _number from the _n storage media 20 _{1 to} 20 _n according to the n: k split key output at the time of archiving. A data fragment is extracted from any accessible storage medium, and the extracted data fragment is additionally written to finally restore the source file.

  Detailed operations of the archive controller 40 during archiving and restoration will be clarified based on the specific configuration described below.

  FIG. 7 shows the configuration of the archive controller 40 realized by the host device 10. In this figure, the archive controller 40 includes a pointer 41, a duplication / distribution unit 42, and a data restoration unit 46. The duplication / distribution unit 42 includes a data fragment extraction unit 43, an association unit 44, and a data fragment storage control unit 45.

The pointer 41 specifies a data fragment by shifting the pointer position from the beginning to the end of the archive target data (file). The data fragment extraction unit 43 sequentially extracts data fragments based on the pointer position of the pointer 41. Adjacent data fragments are directly adjacent and typically do not overlap, but may be extracted with some overlapping. The associating unit 44 associates one extracted data fragment with (n−k + 1) storage media randomly specified independently of the _n storage media 20 _{1 to} 20 _n. The n: k split key is output. The data fragment storage control unit 45 stores one extracted data fragment in duplicate on (n−k + 1) storage media that are randomly specified independently of the _n storage media 20 _{1 to} 20 _n. It is.

Based on the pointer position from the pointer 41, the duplication / distribution unit 42 sequentially extracts data fragments from the archive target data, and the sequentially extracted data fragments are stored in the n storage media 20 _{1 to} 20 _n . Are redundantly stored in (n−k + 1) storage media specified at random independently. When in a manner duplicate all the data fragments in the data in the storage medium 20 ₁ to 20 _n finished dispersed storage, typically archive operation ends when the pointer position reaches the end of the data is.

Although the writing of data fragments to the storage media 20 _{1 to} 20 _n is performed directly via a communication network or the like, the data fragments correspond to the host device 10 or its storage means corresponding to the storage media 20 _{1 to} 20 _n. A storage location (file) is provided and additional data fragments are written, and after the final writing of the last data fragment is completed, the files at the respective storage locations are transmitted to the storage media 20 _{1 to} 20 _n. Also good.

The data restoration unit 46 acquires the n: k split key from the association unit 44 based on the pointer position from the pointer 41 and stores the data fragment at the pointer position in the _n storage media 20 _{1 to} 20 _n. (N−k + 1) storage media are identified, data segments are extracted from any one or more of these storage media, and concatenated and written to the file being restored in the source / restoration file storage unit 10a. The data restoration unit 46 also manages the restored data fragment positions of the _n storage media 20 _{1 to} 20 _n .

FIG. 8 shows a configuration example of the associating unit 44. In this figure, the associating unit 44 includes a random number generator 441 and a key storage unit 442. The random number generator 441 outputs the n: k split key described above, and based on this, the data fragment storage control unit 45 of the duplication / distribution unit 42 stores the data fragments in n storage media 20 _{1 to} 20 _n . Are redundantly stored in (n−k + 1) storage media specified at random independently. Also, the key storage unit 442 stores and manages the n: k split key output from the random number generator 441 in a time series, and sequentially supplies the data to the data recovery unit 46 at the time of data recovery. The data restoration unit 46 performs data restoration by determining the acquisition destination of the data fragment based on the n: k split key from the key storage unit 442.

  The association unit 44 in FIG. 8 is typically implemented by a portable device such as a smart card (IC card), a mobile phone, and a personal digital assistant (PDA) that can be carried by the user, but is not limited thereto. The portable device realizes a microcomputer and random number generating means, which are preferably mounted in an IC chip having tamper resistance.

  FIG. 9 shows a preferred configuration example of the associating unit 44. In this figure, the associating unit 44 includes a seed random number generator 443, a seed register 444, a key random number generator 445, a management table storage unit 446, and the like. Consists of including. The seed random number generator 443 supplies a random number as a seed to the key random number generator 445 that generates a random number by a pseudo random number generation algorithm. The seed is held in the seed register 444 and stored and managed together with the file ID in the management table storage unit 446. The key random number generator 445 outputs a series of random numbers based on the seed, and based on this, data is archived. The seed used for archiving is stored and managed in the management table storage unit 446 in association with the file ID of the archive target data, and the seed is retrieved from the management table storage unit 446 based on the file ID of the recovery target data at the time of data restoration. Using this, the key random number generator 445 outputs the same random number sequence as that at the time of archiving and is used for data restoration.

  The association unit 44 of FIG. 9 is also typically implemented by a portable device such as a smart card (IC card) that can be carried by the user, but is not limited thereto.

  Next, an operation example of the archive system 100 of the first embodiment will be described with reference to the associating unit 44 of FIG. Here, it is assumed that the associating unit 44 is implemented by a smart card.

  FIG. 10 shows the procedure of a smart card processing example.

[Step S01]: An identifier crr_fid for specifying a data file to be divided is received from the host device 10 side.
[Step S02]: The seed random number generator 443 (random number generation program) is called to generate a seed crr_seed for the key random number generator 445 (random number generation program).
[Step S03]: Seed crr_seed is associated with data file identifier crr_fid and written to management table storage unit 446 in the smart card.
[Step S04]: The key random number generator 445 is activated with the seed crr_seed as an input.
[Step S05]: The key random number generator 445 is set in a standby state to wait for a signal from the host device 10.
[Step S06]: A signal is received from the host device 10 side. If the signal indicates the end of the process, the key random number generator 445 is terminated and the process is terminated. If the signal is a key identifier request, the key random number generator 445 is called to generate a random number. Since the random number is used as an identifier of the n: k partition key, the random number is generated so that a natural number from 1 to _n C _{(n−k + 1)} is taken at random.
[Step S07]: The random number generated by the key random number generator 445 is output to the host device 10 as an identifier of the n: k split key.
[Step S08]: Return to Step S05 and wait for a signal from the host device 10.

  FIG. 11 illustrates a procedure of an operation example of the archive system on the host device 10 side. As shown in FIG. 11, corresponding to the associating unit 44 in FIG. 9, the data fragment storage control unit 45 of the host device 10 includes, for example, a destination file determining unit 451 and an additional writing unit 452. The destination file determination unit 451 obtains, for example, the table shown in FIG. 4 based on the randomly generated split key with reference to the destination table storage unit 451a and is randomly selected (n−k + 1) (4 in this example). Number of storage media. The additional writing unit 452 appends data fragments to these (n−k + 1) storage media. This will be described in detail below.

[Step S11]: The identifier crr_fid of the data file to be archived is sent to the smart card.
[Step S12]: A key identifier request signal is sent to the smart card.
[Step S13]: The identifier of the n: k split key is received from the smart card, and the storage medium to which the data fragment is recorded is specified from the identifier using means such as table reference. In the figure, the identifier ip is received, and dst_file2 and dst_file3 are specified as recording destination storage media.
[Step S14]: A data fragment (fgmntp) having a predetermined length is extracted from the position pointed to by the pointer of the archive source data file (src_file), and the pointer is advanced to the head of the next data fragment to be extracted.
[Step S15]: A copy of the extracted data fragment is added to the end of the specified storage medium (dst_file2 and dst_file3 in FIG. 11).
[Step S16]: If the pointer points to the end of the data file, a signal indicating the end of the process is sent to the smart card. If there is still data to be archived, the process returns to Step S12, and the key identifier request Send the signal to the smart card.

12 and 13 are for explaining an example of data restoration processing.
FIG. 12 shows a procedure of a restoration processing example in the smart card.

[Step S21]: The identifier crr_fid of the data file is received from the host device 10.
[Step S22]: The table in the management table storage unit 446 is referred to, and the seed crr_seed associated with the identifier crr_fid is acquired.
[Step S23]: The key random number generator 445 is activated with the seed crr_seed as an input.
[Step S24]: The key random number generator 445 is set in a standby state to wait for a signal from the host device 10.
[Step S25]: A signal is received from the host device 10 side. If the signal indicates the end of the process, the key random number generator 445 is terminated and the process is terminated. If the signal is a key identifier request, the key random number generator 445 is called to generate a random number. Since the random number is used as an identifier of the n: k partition key, the random number is generated so that a natural number from 1 to _n C _{(n−k + 1)} is taken at random.
[Step S26]: The random number generated by the key random number generator 445 is output to the host device 10 as an identifier of the n: k split key.
[Step S27]: Return to step S22 and wait for a signal from the host device 10.

  FIG. 13 shows a procedure of an operation example of the archive system on the host device 10 side.

[Step S31]: The identifier crr_fid of the data file to be restored is sent to the smart card.
[Step S32]: A key identifier request signal is sent to the smart card.
[Step S33]: The identifier of the n: k split key is received from the smart card, and the storage medium to which the data fragment is acquired from the identifier is specified using means such as table reference. In the figure, the identifier ip is received, and dst_file2 and dst_file3 are specified as recording destination storage media.
[Step S34]: The head data fragment is removed from all the specified storage media (dst_file2 and dst_file3 in FIG. 9). The head position of each storage means after removing the data fragment can be managed as a checkpoint.
[Step S35]: The data fragment extracted in step S34 is added to the end of the data file (src_file) to be restored.
[Step S36]: (When there is no data remaining in all the storage media, a signal indicating the end of processing is sent to the smart card, and when there is still data to be restored, the process returns to Step S32, and the key identifier request Send the signal to the smart card.
This is the end of the description of the first embodiment.

[Second Embodiment]
Next, an archive system according to a second embodiment of the present invention will be described. The second embodiment shows a method for improving safety by randomly determining the length of the data fragment each time.

  The difference from the first embodiment is that the output from the key random number generator 445 determines the length of the data fragment in addition to the identifier of the split key. Specifically, the output of the key random number generator 445 is a pair of a natural number ip specifying the split key and a natural number bp specifying the length of the data fragment.

  14 and 15 illustrate a processing example in the archive system of the smart card and the host device 10 in the second embodiment.

  FIG. 14 shows the procedure of a smart card processing example.

[Step S41]: An identifier crr_fid for specifying a data file to be divided is received from the host device 10 side to the smart card.
[Step S42]: The seed random number generator 443 is called to generate a seed crr_seed for the key random number generator 445.
[Step S43]: The seed crr_seed is associated with the data file identifier crr_fid and written to the table of the management table storage unit 446 in the smart card.
[Step S44]: The key random number generator 445 is activated with the seed crr_seed as an input.
[Step S45]: The key random number generator 445 is set in a standby state and waits for a signal from the host device 10.
[Step S46]: A signal is received from the host device 10 side. If the signal indicates the end of the process, the key random number generator 445 is terminated and the process is terminated. If the signal is a key identifier request, the key random number generator 445 is called to generate a random number. Since the random number represents the identifier ip of the n: k partitioning key and the length (bit length) bp of the data fragment, the random number is a naturally generated natural number from 1 to _n C _{(n−k + 1)} and a natural number representing the length. It is a pair. You may generate ip and bp separately using two pseudo random number generation means.
[Step S47]: The random number generated by the key random number generator 445 is output to the host device 10 as the identifier of the n: k split key and the length of the data fragment.
[Step S48]: Return to Step S45 and wait for a signal from the host device 10.

  FIG. 15 illustrates a procedure of an operation example of the archive system on the host device 10 side.

[Step S51]: The identifier crr_fid of the data file to be archived is sent to the smart card.
[Step S52]: A key identifier request signal is sent to the smart card.
[Step S53]: The identifier of the n: k split key and the length of the data fragment are received from the smart card, and the storage medium to which the data fragment is recorded is specified from the identifier using means such as table reference. In the figure, an identifier ip and a data fragment length bp are received, and dst_file2 and dst_file3 are specified as recording storage media from ip.
[Step S54]: A data fragment (fgmntp) having a length equal to the specified bp is extracted from the position pointed to by the pointer of the data file (src_file) of the archive source, and the pointer position is added to the head of the next data fragment to be extracted. Proceed.
[Step S55]: A copy of the extracted data fragment is added to the end of the specified storage media dst_file2 and dst_file3.
[Step S56]: If the pointer points to the end of the data file, a signal indicating the end of the process is sent to the smart card. If there is still data to be archived, the process returns to step S52, and the key identifier request is received. Send the signal to the smart card.

The procedure for restoring the original data file is the reverse procedure of the above archiving procedure, and can be easily assumed from the description of the first embodiment, so that the description is omitted.
This is the end of the description of the second embodiment.

[Third embodiment]
Next, a third embodiment of the present invention will be described. The third embodiment shows a method for further improving safety by archiving dummy data at random timing.

  The difference between the third embodiment and the first embodiment is that the output from the key random number generator 445 determines whether to insert dummy data in addition to the identifier of the split key. . Specifically, the output of the key random number generator 445 is a pair of a natural number ip that specifies the identifier of the split key and a logical value rp that specifies the insertion of dummy data. The logical value is true or false, and the probability distribution is not always uniform (the probability that true / false is output is not necessarily ½).

  FIG. 16 shows a procedure of a smart card processing example.

[Step S61]: An identifier crr_fid for specifying a data file to be divided is received from the host device 10 side to the smart card.
[Step S62]: The seed random number generator 443 is called to generate a seed crr_seed for the key random number generator 445.
[Step S63]: Seed crr_seed is associated with data file identifier crr_fid and written to the table of management table storage unit 446 in the smart card.
[Step S64]: The key random number generator 445 is activated with the seed crr_seed as an input.
[Step S65]: The key random number generator 445 is set in a standby state and waits for a signal from the host device 10.
[Step S66]: A signal is received from the host device 10 side. If the signal indicates the end of the process, the key random number generator 445 is terminated and the process is terminated. If the signal is a key identifier request, the key random number generator 445 is called to generate a random number. A random number is a pair of an identifier ip of an n: k partition key and a logical value rp that specifies insertion of dummy data.
[Step S67]: The random number generated by the key random number generator 445 is output to the host device 10 as the identifier of the n: k split key and the length of the data fragment.
[Step S68]: Return to Step S65 and wait for a signal from the host device 10.

  FIG. 17 shows a procedure of a processing example of the archive program on the host device 10 side. Hereinafter, the procedure will be described with a number.

[Step S71]: The identifier crr_fid of the data file to be archived is sent to the smart card.
[Step S72]: A key identifier request signal is sent to the smart card.
[Step S73]: An n: k partition key identifier and a logical value designating dummy data insertion are received from a smart card, and a storage medium to which a data fragment is recorded is identified from the identifier using means such as table reference. To do. In the figure, an identifier ip and a logical value rp are received, and dst_file2 and dst_file3 are specified as recording destination storage media from ip.
[Step S74]: If the value of rp is true, dummy data is randomly generated.
[Step S75]: If the value of rp is false, a data fragment (fgmntp) having a predetermined length is extracted from the position pointed to by the pointer of the archive source data file (src_file), and the pointer is extracted next. Advance to the beginning of the data fragment.
[Step S76]: Depending on the value of rp, the generated dummy data or a copy of the extracted data fragment is added to the end of the specified storage medium (dst_file2 and dst_file3 in FIG. 17).
[Step S77]: If the pointer points to the end of the data file, a signal indicating the end of processing is sent to the smart card. If there is still data to be archived, the process returns to step S72, and the key identifier request is received. Send the signal to the smart card.

Since the processing of the smart card and the host side archive system at the time of data restoration can be easily estimated from the description of the first embodiment, the description is omitted here.
This is the end of the description of the embodiment of the present invention.

  In addition, this invention is not limited to the above-mentioned Example, A various change is possible. Moreover, you may combine the characteristic of each Example. For example, the dummy data and the data fragment may be switched while the length of the data fragment of the second embodiment is made variable. In addition, the data fragment may be cut out according to a predetermined rule, and is not limited to an aspect in which data is extracted in order from the head to the tail. Any final data fragment set may cover all data. Data fragments may partially overlap. Dummy data may be added to the beginning and end of data to be archived. Data to be archived may be stored locally in a plurality of portable storage media, and then transported to different remote sites by distribution means.

  Further, the present invention may be applied to an apparatus or system that performs only data archiving, or an apparatus or system that performs only data restoration.

The technical features described here are listed below.
[Technical features 1]
In a distributed archive system in which data is distributed and stored in a plurality of n (n is an integer of 2 or more) storage media,
The n storage media;
Pointer means that points to the beginning of the data in the initial state and moves in one direction toward the end of the data as the process proceeds;
A data fragment having a predetermined length is extracted from the pointer position designated by the pointer means, and (n−k + 1) (k is a positive integer less than or equal to n) storage media in which the data fragment is to be stored A redundant distribution unit that selects the storage medium independently and randomly from the storage medium and records the data fragment in duplicate on the (n−k + 1) storage media;
Each time the duplication / distribution unit stores the data fragment in duplicate in the (n−k + 1) storage media, the pointer position is advanced by the length of the data fragment, and the duplication / distribution unit performs the data distribution. The distributed archiving system is characterized in that the storage processing of the duplicate distribution means is repeated until the processing is completed for all data fragments included in.
[Technical feature 2]
The distributed archiving system according to the technical feature 1, wherein the length of a data fragment to be extracted is independently and randomly determined at each of the pointer positions.
[Technical feature 3]
At each of the pointer positions, the data fragment is recorded on (n−k + 1) storage media and the pointer position is advanced, or randomly generated meaningless dummy data is arbitrarily substituted for the data fragment. 2. The distributed archiving system according to the technical feature 1, wherein the selection of whether to record the number of storage media and keep the pointer position unchanged is determined independently and randomly.
[Technical feature 4]
At each of the pointer positions, the data fragment is recorded on (n−k + 1) storage media and the pointer position is advanced, or randomly generated meaningless dummy data is arbitrarily substituted for the data fragment. The distributed archiving system according to the technical feature 2, wherein the selection of whether to record the number of storage media and keep the pointer position unchanged is determined independently and randomly.
[Technical feature 5]
First random number generating means for generating a random number, generating an independent random number by the first random number generating means at each of the pointer positions, and determining an archive destination storage medium based on the random number 2. The distributed archive system according to the technical feature 1 characterized by the above.
[Technical feature 6]
First random number generating means for generating a random number, and at each of the pointer positions, an independent random number is generated by the first random number generating means, and an archive destination storage medium is determined based on the random number; In addition, the technical feature 4 is characterized in that, based on the random number, the length of the data fragment, the presence / absence of dummy data insertion, or both the length of the data fragment and the presence / absence of dummy data insertion are determined. Described distributed archive system.
[Technical feature 7]
Having a key recording means for recording a series of random numbers generated by the first random number generation means, and determining a storage medium from which a data fragment is to be obtained based on the series of random numbers recorded in the key recording means; In addition, based on the series of random numbers, the length of the data fragment, the presence / absence of dummy data, or both the length of the data fragment and the presence / absence of dummy data insertion are determined to restore the original data. The distributed archiving system according to Technical Feature 6, characterized in that
[Technical feature 8]
The second random number generation means for generating a random number is held, and the first random number generation means realizes a pseudo random number generation algorithm for generating an arbitrary number of random numbers by using a seed as an input. The random number generated by the second random number generation means is input to the first random number generation means as a seed, and thereafter, the archive destination storage medium is determined based on the random number generated by the first random number generation means. The distributed archive system according to Technical Feature 5 or Technical Feature 6.
[Technical feature 9]
Holding a key recording means for recording the seed generated by the second random number generating means, inputting the seed recorded in the key recording means to the first random number generating means, wherein the first random number generating means 9. The distributed archive system according to technical feature 8, wherein a storage medium from which a data fragment is to be obtained is determined from a series of random numbers to be generated, and the original data is restored.
[Technical feature 10]
10. The distributed archive system according to technical feature 1 to technical feature 9, wherein a part or all of the n storage media are servers that can be accessed via a communication network.
[Technical feature 11]
By providing the key recording means in a device that can be carried by the user who owns the data, the user can perform data archiving and decryption processing using an arbitrary host computer. A distributed archiving system according to claim 7 or technical feature 9.
[Technical feature 12]
The distributed archive system according to the technical feature 11, wherein the device that the user can carry is an IC card, a mobile phone, or a personal digital assistant.
[Technical feature 13]
The device that can be carried by the user holds a microcomputer, performs access control to data in the device by the microcomputer, and executes part or all of data archiving and restoration processing, The distributed archive system according to the technical feature 11.
[Technical feature 14]
14. The distributed archive system according to the technical feature 13, wherein the device that the user can carry holds either one or both of the first random number generator and the second random number generator.
[Technical feature 15]
The distributed archive system according to the technical feature 14, wherein the microcomputer held by the device that can be carried by the user and the random number generation means are mounted in an IC chip having tamper resistance.
[Technical feature 16]
16. The distributed archiving system according to technical feature 1 to technical feature 15, wherein dummy data having an appropriate length is added to the beginning and end of data to be archived.
[Technical feature 17]
In a data archiving apparatus for archiving archive target data to n (n is an integer of 2 or more) archive destination storage means,
Cutout means for cutting out a plurality of data fragments from the archive target data in a predetermined procedure, and at least a plurality of cut out data fragments include all of the archive target data;
Each of the plurality of data fragments cut out by the cutout means is independent of (n−k + 1) (k is a positive integer less than n) archive destination storage means among the n archive destination storage means. And an association means for randomly relating,
A data archiving apparatus, wherein each of the data fragments cut out by the cut-out means is archived redundantly in n−k + 1 archive destination storage means respectively related by the correlation means.
[Technical feature 18]
In a data archiving computer program for archiving archive target data to n (n is an integer of 2 or more) archive destination storage means,
Computer
Cutting out means for cutting out a plurality of data fragments from the archive target data in a predetermined procedure, and at least a plurality of cut out data fragments including all of the archive target data; and
Each of the plurality of data fragments cut out by the cutout means is independent of (n−k + 1) (k is a positive integer less than n) archive destination storage means among the n archive destination storage means. Function as an association means to randomly relate,
A computer program for data archiving, wherein each of the data fragments cut out by the cut-out means is archived redundantly in n−k + 1 archive destination storage means respectively related by the correlation means.
[Technical feature 19]
In a restoration device for restoring data archived in n (n is an integer of 2 or more) archive destination storage means by the archive device described in the technical feature 17,
Association information acquisition means for acquiring association information used by the association means to randomly associate each of the data fragments of the data independently with (n−k + 1) archive destination storage means;
For each of the data fragments, a data fragment acquisition means for retrieving the data fragment from any of n−k + 1 archive destination storage means associated with the data fragment by the association information;
A data restoration apparatus comprising: a coupling means for coupling the extracted data fragment by a coupling procedure corresponding to the predetermined procedure for extracting the data fragment.
[Technical feature 20]
In a computer program for data restoration used for restoring data archived in n (n is an integer of 2 or more) archive destination storage means by the archive device described in the technical feature 17,
Computer
Association information acquisition means for acquiring association information used by the association means to randomly associate each of the data fragments of the data independently with (n−k + 1) archive destination storage means;
For each of the data fragments, a data fragment acquisition means for retrieving the data fragment from any of (n−k + 1) archive destination storage means associated with the data fragment by the association information; and
A computer program for data restoration for causing the extracted data fragment to function as a connection means for connecting in accordance with a connection procedure corresponding to the predetermined procedure for cutting out the data fragment.
[Technical feature 21]
In a data archiving method, archive target data is provided with n (n is an integer of 2 or more) archive destination storage means, and archive target data is archived in the n (n is an integer of 2 or more) archive destination storage means. ,
A step of cutting out a plurality of data fragments from the archive target data in a predetermined procedure, and at least the plurality of cut out data fragments including all of the archive target data;
The association means converts each of the plurality of data fragments cut out by the cut-out means to (n−k + 1) (k is a positive integer less than or equal to n) archive destinations out of the n archive destination storage means. Independently and randomly related to the storage means,
A data archiving method comprising: archiving each of data fragments cut out by the cutout unit in duplicate to n−k + 1 archive destination storage units respectively related by the association unit.

10 host device 10a the source / recovery file storage unit 10b the computer program ₂₀ 1 to 20 _n storage medium 30 communication network 40 archiving controller 41 pointer 42 overlaps distributed unit 43 data fragment extraction unit 44 association unit 45 data fragments storage control unit 46 data recovery Unit 100 archive system 441 random number generator 442 key storage unit 443 seed random number generator 444 seed register 445 key random number generator 446 management table storage unit

Claims (8)

In a distributed archive system in which archive target data is distributed and stored in a plurality of n (n is an integer of 2 or more) storage media,
The n storage media;
In the initial state, pointing to the data head of the archive target data, pointer means that moves in one direction toward the end of the data as the process proceeds,
First random number generating means for generating a random number at each pointer position designated by the pointer means;
A data fragment is extracted from the pointer position designated by the pointer means, and (n−k + 1) (k is a positive integer less than or equal to n) storage media from which the data fragment is to be stored is selected from the n storage media. The first random number generation means uniquely determines the data fragment, and in the (n−k + 1) storage media, the data fragment is recorded in advance as one or more data fragments. In some cases, in the mode of additional recording following this, in the case of no data fragment recorded in advance, in the mode of new recording, there is a duplicate decentralization means for recording redundantly,
Each time the duplication / distribution means records the data fragment in duplicate on the (n−k + 1) storage media, the pointer position is advanced by the length of the data fragment, and the duplication / distribution means Until the processing is completed for all data fragments included in
At the time of restoration of archive target data, (n−k + 1) storage media for duplication recording are determined according to the same random number as the random number generated by the first random number generating means at each pointer position designated by the pointer means. A distributed archiving system, wherein data fragments are read from any one of the storage media according to the recording order of appending the data fragments and sequentially linked to restore the archive target data.   2. The distributed archiving system according to claim 1, wherein the archiving process is performed using the data with dummy data to which dummy data of an appropriate length is added at the beginning and end of the data requested for archiving as the archiving target data. .   The distributed archive system according to claim 1, wherein the length of the data fragment to be extracted is determined independently and randomly at each of the pointer positions.   At each of the pointer positions, the data fragment is recorded on (n−k + 1) storage media and the pointer position is advanced, or randomly generated meaningless dummy data is arbitrarily substituted for the data fragment. The distributed archive system according to any one of claims 1 to 3, wherein the selection of whether to record on a number of storage media and keep the pointer position unchanged is determined independently and randomly.   At each of the pointer positions, the length of the data fragment to be extracted is determined independently and randomly, and at each of the pointer positions, the data fragment is recorded on (n−k + 1) storage media. Or whether to record meaningless dummy data randomly generated in place of the data fragment in an arbitrary number of storage media and keep the pointer position unchanged is determined independently and randomly. The distributed archive system according to any one of claims 1 to 3, wherein:   The distributed archive system according to any one of claims 1 to 5, wherein the first random number generation means generates an arbitrary number of random numbers based on a pseudo random number generation algorithm with a seed as an input.   The seed is generated by a second random number generating means, and further includes a key storage means for recording the seed, and the key storage means is provided in a device that can be carried by a user who owns the data. 7. The distributed archiving system according to claim 6, wherein the data can be archived and decrypted using an arbitrary host computer. In a data archiving apparatus for archiving archive target data to n (n is an integer of 2 or more) archive destination storage means,
A plurality of data pieces sequentially cut out from the archive target data in a predetermined procedure, and at least a plurality of cut out data pieces include all of the archive target data;
Based on the random numbers generated by the random number generation unit for each of the plurality of data fragments sequentially cut out by the cutout unit, each of the plurality of data fragments is stored in the n archive destination storage units. (N−k + 1) (k is a positive integer less than or equal to n) archive destination storage units, and a correlation unit that uniquely associates with (n−k + 1) correspondence relationships.
In each of the (n−k + 1) archive destination storage means associated with one pair of (n−k + 1) correspondences by each of the data fragments sequentially extracted by the extraction means, If there is one or more data fragments that have been recorded in advance, this is followed by additional recording. If there is no data fragment that has been previously recorded, new recording is used. Record,
When restoring data to be archived, (n−k + 1) storage media as duplicate recording destinations are determined according to the same random number as the random number generated by the random number generation unit for each data fragment, and the data is stored from any of the storage media. A data archiving apparatus, wherein data fragments are read out in accordance with a recording order of additional writing of fragments and sequentially linked to restore data to be archived.