JP6201385B2 - Storage apparatus and storage control method - Google Patents

Description

  The present invention relates to a storage apparatus and a storage control method.

  Conventionally, a hash value is calculated before data is written to a nonvolatile storage device such as an HDD (Hard Disk Drive), and if data having the same hash value is already stored, the data is shared without writing the data. There is deduplication storage technology.

For example, if the number of bits of hash is 512 bits, the possibility that hash values collide is about once every 10 ¹⁵⁰ times, and the storage device that performs data deduplication should check whether the data is the same. In addition, it can be determined whether or not the data is the same only by the hash value.

  Here, the data is a file or a block having a size of, for example, 64 KB (kilobytes), and if it is attempted to check the identity using the data, the time required for writing increases. Therefore, the storage apparatus can shorten the data writing time by determining the identity of the data using the hash value.

  When the original image is converted into the JPEG format, secret information is obtained by using a quantized DCT coefficient obtained by performing DCT on each of the 8 × 8 pixel blocks obtained by dividing the original image, and for each block. There is a conventional technique for preventing image falsification by embedding confidential information. Further, there is a conventional technique that makes it difficult to falsify content or metadata by performing conversion processing that is difficult to separate between input content and metadata.

JP 2003-289435 A JP 2004-72184 A

  In the deduplication storage technology, the possibility that the hash values collide accidentally can be sufficiently reduced, but a technology for intentionally colliding the hash values may be developed. When such a technique is developed and it becomes possible to create data having the same hash value as that of a certain data, a denial of service attack becomes possible.

  For example, an attacker creates data B having the same hash value as data A that can be expected to be created in the future, and data B is written to the storage device in advance, so that data A is written when data A is written. A is replaced with data B. As a result, the attacker can cause the information processing apparatus that uses the data A to perform an incorrect process.

  Note that it is possible to change the hash function when data with the same hash value can be generated. However, in the storage device, it is necessary to recalculate the hash value for all data. Is a heavy burden.

  An object of one aspect of the present invention is to provide a storage apparatus and a storage control method capable of preventing a denial-of-service attack using deduplication storage technology.

In one aspect, the storage device disclosed in the present application includes a first generation unit that generates a hash value as a first value from data when the data is written to the nonvolatile storage device, and the data is the nonvolatile A second generator that generates a second value using a portion of the data extracted from the data based on a number determined when written to the storage device; the first value and the second And a control unit for controlling the redundant storage of the data based on the value.

  According to one embodiment, a denial of service attack using deduplication storage technology can be prevented.

FIG. 1 is a diagram illustrating a functional configuration of the storage apparatus according to the embodiment. FIG. 2 is a diagram illustrating an example of the correspondence table. FIG. 3 is a diagram illustrating an example of the data structure of the additional key. FIG. 4 is a diagram illustrating an example of the hash value table. FIG. 5 is a flowchart showing the processing procedure of the writing unit. FIG. 6 is a flowchart showing a processing procedure for duplicate data processing. FIG. 7 is a flowchart showing a processing procedure for new data processing. FIG. 8 is a flowchart showing the processing procedure of the reading unit. FIG. 9 is a diagram illustrating an example of a service denial attack. FIG. 10 is a diagram illustrating a hardware configuration of the storage apparatus.

  Embodiments of a storage apparatus and a storage control method disclosed in the present application will be described below in detail with reference to the drawings. Note that this embodiment does not limit the disclosed technology.

  First, the functional configuration of the storage apparatus according to the embodiment will be described. FIG. 1 is a diagram illustrating a functional configuration of the storage apparatus according to the embodiment. As illustrated in FIG. 1, the storage device 1 includes a volatile storage unit 10, a control unit 20, and a nonvolatile storage unit 30.

  The volatile storage unit 10 is a storage unit in which stored data disappears when the power is turned off. The control unit 20 controls the storage device 1 using data stored in the volatile storage unit 10. The nonvolatile storage unit 30 stores data that the server 2 connected to the network stores in the storage device 1. The data stored in the non-volatile storage unit 30 does not disappear even when the power is turned off.

  The volatile storage unit 10 includes a data unit 11, a correspondence table 12, and a hash value table 13. The data unit 11 temporarily stores a part of data stored in the nonvolatile storage unit 30 by the server 2.

  The correspondence table 12 stores information indicating correspondence between data identifiers stored in the nonvolatile storage unit 30 and data keys. FIG. 2 is a diagram illustrating an example of the correspondence table 12. As shown in FIG. 2, the correspondence table 12 stores an identifier, a hash value, and an additional key for each data.

  The identifier is a value given to the data in order to identify the data. Here, the data is a 64 KB block, and the identifier is a block number. If the data is a file, the identifier is the ID of the file.

  The hash value is a value calculated from data using a hash function. Examples of the hash function include MD5 (Message Digest Algorithm 5), SHA (Secure Hash Algorithm) 1, SHA256, SHA512, and the like. For example, if the hash function is SHA512, the length of the hash value is 64 bytes.

  The additional key is a value determined when data is written to the storage apparatus 1, and is used as a data key together with a hash value. FIG. 3 is a diagram illustrating an example of the data structure of the additional key.

  As shown in FIG. 3, the additional key is a concatenation of four “positions” having a length of 2 bytes and “content” having a length of 2 bytes. “Position” is a random number generated when data is written to the storage apparatus 1. “Content” is 2-byte data of the byte indicated by “position” from the top of the data.

  For example, the correspondence table 12 stores a block number “28391893” as an identifier, “af49389...” As a hash value, and “128“ aa ”,. “128” aa “” indicates that the 2-byte data of the 128th byte from the top of the data is “aa”.

  The storage device 1 uses a set of a hash value and an additional key as a data key. By using the additional key as the data key in addition to the hash value, the storage apparatus 1 can prevent a denial of service attack using the deduplication storage technology.

  For example, even when data B having the same hash value as the correct data A is written to the storage device 1 before the data A is written, the additional key is different between the data A and the data B, so the data A is stored in the storage device 1. Is written to.

  Since the additional key is a value determined when the data A is written in the storage device 1, the attacker cannot specify the additional key in advance. Therefore, even if the attacker can generate the data B having the same hash value as the data A, the attacker cannot abuse the deduplication storage technology to replace the data A with the data B.

  In this example, four random numbers are used and four 2-byte data are used as additional keys. However, the storage apparatus 1 may accidentally collide with additional keys by using a larger number of random numbers. Can be reduced. In addition, the storage device 1 generates an additional key using another number such as a number based on the time when data is written instead of generating a random number as a value determined when the data is written to the storage device 1. You can also

  Returning to FIG. 1, the hash value table 13 stores information related to data stored in the nonvolatile storage unit 30. FIG. 4 is a diagram illustrating an example of the hash value table 13. As illustrated in FIG. 4, the hash value table 13 stores a hash value, an additional key, position information, and a reference number for each data.

  The position information is information indicating a position where the data specified by the combination of the hash value and the additional key is stored in the nonvolatile storage unit 30. The reference number indicates the number of data that is referred to by different identifiers. Since the storage apparatus 1 does not store the same data repeatedly, one data may correspond to a plurality of identifiers.

  For example, the hash value table 13 stores “af49389...” As the hash value, “128“ aa ”,...” As the additional key, “4324932” as the position information, and “1” as the reference number. To do.

  Returning to FIG. 1, the control unit 20 includes a writing unit 21 and a reading unit 22. The writing unit 21 writes data to the nonvolatile storage unit 30 based on an instruction from the server 2. When writing the data, the writing unit 21 generates a hash value from the data, and refers to the hash value table 13 to determine whether both the hash value and the data specified by the additional key match.

  Then, when both the hash value and the data specified by the additional key match, the writing unit 21 performs processing when there is duplicate data, and both the hash value and the data specified by the additional key are If they do not match, a process of writing new data to the nonvolatile storage unit 30 is performed.

  The writing unit 21 includes a generation unit 211. The generation unit 211 generates an additional key from the new data when the writing unit 21 writes the new data to the nonvolatile storage unit 30. If the identifier specified in the write request is in the correspondence table 32, the writing unit 21 updates the hash value and the additional key of the correspondence table 32. If not, the identifier, the hash value, and the additional key are updated. Is used to create a new entry in the correspondence table 32.

  The reading unit 22 reads data from the nonvolatile storage unit 30 based on an instruction from the server 2. The reading unit 22 refers to the correspondence table 12 and the hash value table 13, acquires data position information from the data identifier, and reads data from the nonvolatile storage unit 30 using the acquired position information. Then, the reading unit 22 transmits the data read from the nonvolatile storage unit 30 to the server 2.

  The nonvolatile storage unit 30 includes a data unit 31, a correspondence table 32, and a hash value table 33. The data unit 31 stores data written to the storage device 1 by the server 2. A part of the data stored in the data unit 31 is temporarily stored in the data unit 11.

  The correspondence table 32 is the same as the correspondence table 12 included in the volatile storage unit 10. When the storage apparatus 1 is activated, data stored in the correspondence table 32 is read into the correspondence table 12. When the correspondence table 12 is updated, the correspondence table 32 is also updated.

  The hash value table 33 is the same as the hash value table 13 included in the volatile storage unit 10, and data stored in the hash value table 33 is read into the hash value table 13 when the storage apparatus 1 is activated. Further, when the hash value table 13 is updated, the hash value table 33 is also updated.

  The correspondence table 32 and the hash value table 33 are updated in synchronization with the update of the correspondence table 12 and the hash value table 13, respectively. However, when the storage apparatus 1 is stopped, the correspondence table 32 and the hash value table 33 are collectively updated. The hash value table 33 can also be updated. The reason why the storage device 1 updates the correspondence table 32 and the hash value table 33 in synchronization with the correspondence table 12 and the hash value table 13 is to prepare for a failure of the device.

  Next, the processing procedure of the writing unit 21 will be described. FIG. 5 is a flowchart showing a processing procedure of the writing unit 21. As illustrated in FIG. 5, when an application operating on the server 2 requests data writing, the writing unit 21 calculates a hash value of data to be written (step S1).

  Then, the writing unit 21 searches the hash value table 13 with the calculated hash value (step S2) and determines whether there is data having the same hash value (step S3). As a result, when there is data having the same hash value, the writing unit 21 reads the additional key from the hash value table and determines whether or not the data at the position specified by the additional key matches ( Step S4) If the data specified by the additional key also matches, duplicate data processing is performed (step S5).

  On the other hand, if the data specified by the additional key does not match, or if there is no data having the same hash value, the writing unit 21 performs new data processing (step S6).

  Thus, when the writing unit 21 writes data, it determines whether there is matching data including the additional key, so that the storage apparatus 1 prevents a denial-of-service attack using the deduplication storage technology. Can do.

  Next, a processing procedure for duplicate data processing will be described. FIG. 6 is a flowchart showing a processing procedure for duplicate data processing. As illustrated in FIG. 6, the writing unit 21 increases the reference number of the entry having the same hash value and additional key in the hash value table 13 by 1 (Step S <b> 11).

  Then, the writing unit 21 searches the correspondence table 12 using the identifier specified by the write request from the server 2 (step S12), and determines whether or not the identifier specified by the write request is in the correspondence table 12. Determination is made (step S13).

  As a result, when the data is in the correspondence table 12, the data specified by the identifier is updated, so the writing unit 21 uses the hash value and the additional key corresponding to the identifier of the correspondence table 12 as the new hash value. And updated with the additional key (step S14). Then, the writing unit 21 decreases the reference number of the hash value table 13 of the data before update by 1 (step S15).

  Then, the writing unit 21 determines whether or not the reference number decreased by 1 is 0 (step S16). If the reference number is 0, the data before update is no longer referred to, so the hash value table 13 The entry of the data before update is deleted from (Step S17). Then, the writing unit 21 reflects the change in the correspondence table 12 and the hash value table 13 in the nonvolatile storage unit 30 (step S18).

  On the other hand, when the identifier specified by the write request is not in the correspondence table 12, the data specified by the new identifier is written, so the writing unit 21 uses the identifier specified by the write request. A new entry is created in the correspondence table 12 (step S19). Then, the writing unit 21 proceeds to step S18.

  As described above, the writing unit 21 performs duplicate data processing, whereby the storage apparatus 1 can prevent duplicate storage of the same data.

  Next, a processing procedure for new data processing will be described. FIG. 7 is a flowchart showing a processing procedure for new data processing. As illustrated in FIG. 7, the generation unit 211 generates an additional key (step S30). Specifically, the generation unit 211 generates four 2-byte long random numbers and combines the 2-byte data of the byte corresponding to the random number from the top of the data with the random numbers to generate 4-byte data. For random numbers. Then, the generation unit 211 generates an additional key by combining the four 4-byte data. Subsequently, the writing unit 21 secures a new area in the nonvolatile storage unit 30 and writes the data (step S31).

  Then, the writing unit 21 creates a new entry in the hash value table 13 using the hash value and the additional key, and stores the data position information and the reference number (step S32). Here, the value of the reference number is an initial value of 1.

  Then, the writing unit 21 searches the correspondence table 12 using the identifier specified by the write request from the server 2 (step S33), and determines whether or not the identifier specified by the write request is in the correspondence table 12. Determination is made (step S34).

  As a result, when the data is in the correspondence table 12, the data specified by the identifier is updated, so the writing unit 21 uses the hash value and the additional key corresponding to the identifier of the correspondence table 12 as the new hash value. And updated with the additional key (step S35). Then, the writing unit 21 decreases the reference number of the hash value table 13 of the data before update by 1 (step S36).

  Then, the writing unit 21 determines whether or not the reference number decreased by 1 is 0 (step S37). If the reference number is 0, the data before update is no longer referred to, so the hash value table 13 The entry of the data before update is deleted from (Step S38). Then, the writing unit 21 reflects the change of the correspondence table 12 and the hash value table 13 in the nonvolatile storage unit 30 (step S39).

  On the other hand, when the identifier specified by the write request is not in the correspondence table 12, the data specified by the new identifier is written, so the writing unit 21 uses the identifier specified by the write request. A new entry is created in the correspondence table 12 (step S40). Then, the writing unit 21 proceeds to step S39.

  As described above, the writing unit 21 performs new data processing, whereby the storage apparatus 1 can store data in which the same data is not stored in the nonvolatile storage unit 30.

  Next, the processing procedure of the reading unit 22 will be described. FIG. 8 is a flowchart showing the processing procedure of the reading unit 22. As shown in FIG. 8, when an application running on the server 2 requests to read data, the reading unit 22 searches the correspondence table 12 using the identifier specified by the read request, and acquires a hash value and an additional key. (Step S51).

  And the reading part 22 searches the hash value table 13 using the acquired hash value and an additional key, and acquires position information (step S52). And the reading part 22 reads data from the non-volatile memory | storage part 30 using the acquired positional information, and transmits to the server 2 (step S53).

  As described above, when the reading unit 22 reads data using the additional key in addition to the hash value, the storage apparatus 1 may transmit appropriate data from the data having the same hash value to the server 2. it can.

  As described above, in the embodiment, the writing unit 21 generates a random number when writing data to the nonvolatile storage unit 30, and generates an additional key based on the generated random number. Then, the storage device 1 uses the additional key generated by the writing unit 21 as a key together with the hash value to eliminate data duplication. Therefore, even when data having the same hash value can be generated, the storage apparatus 1 can prevent a denial of service attack using the deduplication storage technology.

  FIG. 9 is a diagram illustrating an example of a service denial attack. FIG. 9 shows an example of an attack when a file of the main site 7 is copied to the mirror site 8. In FIG. 9, the attacker 6 confirms the data contents before the file is copied to the mirror site 8 (1). Then, the attacker 6 generates a fake file having the same hash value and writes it in the mirror site 8 (2).

  Thereafter, the file of the main site 7 is mirrored on the mirror site 8 (3). Then, conventionally, since the fake file having the same hash value has already been written in the mirror site 8, the file to be mirrored is determined to be the same as the fake file, and the contents of the file are replaced at the mirror site 8. (4).

  However, even when the attacker 6 writes a fake file on the mirror site 8, the storage apparatus 1 according to the embodiment generates an additional key using a random number when mirroring the file, and adds it in addition to the hash value. Data identity is determined using the key as a key. Therefore, the storage apparatus 1 can prevent the mirrored file from being replaced with the fake file by generating an additional key different from the fake file for the mirrored file.

  In the embodiment, the correspondence table 12 stores hash values, additional keys, and data identifiers in association with each other, and the hash value table 13 associates hash values, additional keys, data storage positions, and reference numbers. Let me remember. Therefore, the storage apparatus 1 can manage a plurality of identifiers having the same data.

  In the embodiment, the storage apparatus 1 generates an additional key using four random numbers, but can also generate an additional key using more random numbers. Accordingly, when the hash value and the additional key are used as the data key, the storage apparatus 1 can reduce the possibility that the key coincides with different data by chance.

  Note that the storage apparatus described in the embodiment can also be realized by operating a program with a CPU. Therefore, a hardware configuration of a storage apparatus realized by operating a program with the CPU will be described.

  FIG. 10 is a diagram illustrating a hardware configuration of the storage apparatus. As shown in FIG. 10, the storage device 40 includes a main memory 41, a CPU (Central Processing Unit) 42, a host interface 43, and an HDD (Hard Disk Drive) 44.

  The main memory 41 is a memory for storing a program, a program execution result, and the like, and corresponds to the storage unit 10 in FIG. The CPU 42 is a central processing unit that reads and executes a program from the main memory 41, and corresponds to the control unit 20 in FIG.

  The host interface 43 is an interface for connecting the storage device 40 to the server 2. The HDD 44 is a disk device that stores programs and data, and corresponds to the nonvolatile storage unit 30 of FIG. The storage device 40 can also include an SSD (Solid State Drive) instead of the HDD 44.

DESCRIPTION OF SYMBOLS 1 Storage apparatus 2 Server 6 Attacker 7 Large site 8 Mirror site 10 Volatile memory | storage part 11 Data part 12 Correspondence table 13 Hash value table 20 Control part 21 Writing part 22 Reading part 30 Non-volatile storage part 31 Data part 32 Correspondence table 33 Hash value table 40 Storage device 41 Main memory 42 CPU
43 Host interface 44 HDD
211 generator

Claims (5)

A first generation unit that generates a hash value as a first value from the data when the data is written to the nonvolatile storage device;
A second generator that generates a second value using a portion of the data extracted from the data based on a number determined when the data is written to the non-volatile storage device;
And a control unit that controls the redundant storage of the data based on the first value and the second value. A first table for storing the first value and the second value in association with the storage position of the data;
A second table for storing the first value and the second value in association with an identifier for identifying the data;
The storage apparatus according to claim 1, wherein the control unit controls duplication storage of the data using the first table and the second table. The storage apparatus according to claim 1, wherein the second generation unit generates the second value using the number and the extracted data. Before numeration storage device according to claim 1, 2 or 3, characterized in that a random number. Generating a hash value as a first value from the data when the data is written to the non-volatile storage device;
Generating a second value using a portion of the data extracted from the data based on a number determined when the data is written to the non-volatile storage device ;
Controlling the overlapping storage of the data based on the previous SL first value and second value
A storage control method , wherein a process of the nonvolatile storage device is executed by a processor .

Images