JP7196052B2 - Information processing device and information processing method - Google Patents

Description

The present invention relates to an information processing apparatus and an information processing method for specifying the arrangement combination of multiple data blocks in multiple storage areas used in a distributed system.

Conventionally, there has been a technique for improving fault tolerance and parallelism of data processing by managing data in units of blocks and distributing them (for example, see Non-Patent Document 1 and Non-Patent Document 2).

Ghemawat, S., Gobioff, H., and Leung, S. T., The Google file system, 2003.

Shvachko, K., Kuang, H., Radia, S., and Chansler, R., The hadoop distributed file system, In MSST, Vol.10, pp.1-10, 2010.

In the conventional technology, data is arranged in multiple storage areas according to predetermined rules. On the other hand, since the data to be arranged have different usage frequencies, it is preferable to increase the processing execution speed of the data with high usage frequency. Therefore, it is required to arrange data in a suitable storage area based on the frequency of use of the data.

Accordingly, the present invention has been made in view of these points, and aims to provide an information processing apparatus and an information processing method that can specify a storage area suitable for arranging data based on the frequency of use of the data. aim.

An information processing apparatus according to a first aspect of the present invention is a distributed system in which a plurality of data blocks are arranged in a plurality of storage areas having different addresses. a transmission history identification unit that identifies a transmission history indicating a frequency of transmission between storage areas; a storage capacity identification unit that identifies the storage capacity of each of the plurality of storage areas; and the plurality of memories identified by the storage capacity identification section including a plurality of variables corresponding to a placement combination for arranging the plurality of data blocks in the plurality of storage areas so as not to exceed the storage capacity of each of the areas; Quantum Ising is an optimization function that identifies a combination of variable values that minimizes the total transmission cost based on the transmission frequency and the transmission cost of the data block when re-executing the processing executed in Quantum Ising a creation unit that creates an optimization function in a machine-evaluable format; and a quantum Ising machine that evaluates the optimization function created by the creation unit to determine values of the plurality of variables that bring the total cost closer to a minimum. an arrangement combination specifying unit that specifies the arrangement combination so that the total cost is minimized by specifying the combination.

The creation unit may create, as the optimization function, a second-order polynomial described only with binary variables.
The data capacity of each of the plurality of data blocks is constant, and the creation unit specifies a transmission source storage area and a transmission destination storage area of each of the plurality of data blocks corresponding to the arrangement combination. creating a first binary variable group for the storage area and creating a second binary variable group indicating the number of the plurality of data blocks stored in the plurality of storage areas corresponding to the arrangement combination; The optimization function may be created that includes a set of binary variables of 1 and the second set of binary variables.

Based on the transmission frequency, the creation unit creates the first binary variable group by limiting the combinations of transmission sources and transmission destinations to which transmission was performed in the processing executed during the predetermined period. may
The creation unit is an optimization function that allows duplication and placement in a plurality of storage areas of each of the plurality of data blocks, and specifies the combination, and is an optimization function that can be evaluated by a quantum Ising machine. You can also create a function to
The creation unit specifies the combination so that the number of copies when each of the plurality of data blocks is copied and arranged in a plurality of storage areas is equal to or greater than a first predetermined number and equal to or less than a second predetermined number. An optimization function may be created that is a form of optimization function that can be evaluated by a quantum Ising machine.

The transmission history identifying unit identifies the transmission history indicating the frequency of transmission of the data blocks between the plurality of storage areas and the transmission cost of the data blocks associated with processing executed in a predetermined period, and the creating unit , the total cost of the transmission costs based on the transmission costs indicated by the transmission frequency and the transmission history is minimized when the distributed system re-executes the processing executed in the predetermined period for the arrangement combination. The optimization function may be created to identify combinations of variable values.
The information processing apparatus may further include an arrangement control section that arranges the plurality of data blocks in the plurality of storage areas based on the arrangement combination specified by the arrangement combination specifying section.

An information processing method according to a second aspect of the present invention is a computer-executed distributed system in which a plurality of data blocks are arranged in a plurality of storage areas having different addresses. identifying a transmission history indicating the frequency of data block transmission between a plurality of storage areas; identifying a storage capacity of each of said plurality of storage areas; and a storage capacity of each of said plurality of identified storage areas. including a plurality of variables corresponding to an arrangement combination for arranging the plurality of data blocks in the plurality of storage areas so as not to exceed An optimization function that specifies a combination of variable values that minimizes the total transmission cost based on the transmission frequency and the transmission cost of the data block in the case of re-execution, and is in a format that can be evaluated by a quantum Ising machine. and causing the quantum Ising machine to evaluate the created optimization function to identify a combination of values of the plurality of variables that bring the total cost closer to the minimum, thereby reducing the total cost to and identifying said arrangement combination to be the minimum.

According to the present invention, it is possible to specify a storage area suitable for arranging data based on the frequency of use of the data.

It is a figure explaining the outline|summary of the information processing apparatus which concerns on this embodiment. It is a figure which shows the structure of the information processing apparatus which concerns on this embodiment. It is a figure which shows an example of execution result information. FIG. 4 is a diagram showing a table corresponding to a transmission frequency matrix; FIG. FIG. 4 is a diagram showing a table corresponding to a transmission cost matrix; It is a figure which shows an example of storage capacity information. FIG. 10 is a diagram showing a table corresponding to an arrangement matrix; FIG. It is the figure which showed an example of the 1st auxiliary variable in the form of a table. It is the figure which showed an example of the 2nd auxiliary variable in the form of a table. FIG. 10 is a diagram showing a table corresponding to a constellation matrix corresponding to x _ij specified by a quantum Ising machine; 4 is a flow chart showing the flow of processing in the information processing apparatus;

[Overview of information processing device 1]
FIG. 1 is a diagram illustrating an outline of an information processing device 1 according to this embodiment. The information processing apparatus 1 uses a quantum Ising machine to specify the optimal allocation of a plurality of data blocks to a plurality of storage areas in a distributed system in which a plurality of data blocks are arranged in a plurality of storage areas having different addresses. computer.

In this embodiment, each of the plurality of storage areas is provided, for example, in each of the plurality of servers, and when the distributed system uses data stored in each of the plurality of storage areas, A processing cost is assumed. The processing cost includes the transmission cost and the CPU usage time.

In a distributed system, the information processing apparatus 1 acquires transmission history information indicating the frequency of data block transmission between a plurality of storage areas and the history of data block transmission costs associated with processing executed in a predetermined period, and By obtaining storage capacity information indicating the storage capacity of each of the plurality of storage areas, the transmission history and storage capacity are specified ((1) and (2) in FIG. 1).

When the information processing apparatus 1 re-executes a process executed in a predetermined period for an arrangement combination in which a plurality of data blocks are arranged in a plurality of storage areas so as not to exceed the storage capacity of each of the plurality of storage areas. create an optimization function for calculating the total transmission cost in ((3) in FIG. 1). The optimization function is created in a form that can be evaluated by a quantum Ising machine. The information processing apparatus 1 specifies an arrangement combination that minimizes the total transmission cost by having the quantum Ising machine evaluate the created optimization function ((4) in FIG. 1). After that, the information processing device 1 performs placement control of the plurality of data blocks based on the identified placement combination ((5) in FIG. 1). By doing so, the information processing apparatus 1 can specify a suitable arrangement from among a large number of combinations regarding the arrangement of the plurality of data blocks in the plurality of storage areas based on the frequency of use of the data.
The configuration of the information processing apparatus 1 will be described below.

[Configuration example of information processing device 1]
FIG. 2 is a diagram showing the configuration of the information processing device 1 according to this embodiment. The information processing device 1 includes a storage section 11 and a control section 12 .

The storage unit 11 is, for example, a ROM (Read Only Memory) and a RAM (Random Access Memory). The storage unit 11 stores various programs for causing the information processing device 1 to function. For example, the storage unit 11 stores a program that causes the control unit 12 of the information processing device 1 to function as a transmission history identification unit 121, a storage capacity identification unit 122, a creation unit 123, an arrangement combination identification unit 124, and an arrangement control unit 125. do. Note that this program may be composed of a plurality of programs.

The control unit 12 is, for example, a CPU (Central Processing Unit). The control unit 12 controls functions related to the information processing apparatus 1 by executing various programs stored in the storage unit 11 . By executing programs stored in the storage unit 11, the control unit 12 functions as a transmission history identification unit 121, a storage capacity identification unit 122, a creation unit 123, an arrangement combination identification unit 124, and an arrangement control unit 125. .

In a distributed system in which a plurality of data blocks are arranged in a plurality of storage areas having different addresses, the transmission history identification unit 121 determines the frequency of transmission of data blocks between the plurality of storage areas associated with processing executed during a predetermined period of time, and A transmission history is identified that indicates the transmission cost of the data block.

For example, the distributed system stores query execution result information as a statement for executing processing using data blocks in a predetermined storage area. FIG. 3 is a diagram showing an example of execution result information. As shown in FIG. 3, the execution result information includes a query executed during a predetermined period, a data block used for the query, the address of the storage area of the transmission source of the data block, and the address of the data block. This is information that associates the address of the storage area of the transmission destination with the transmission cost that is the cost of transmitting the data block from the storage area of the transmission source to the storage area of the transmission destination. The transmission cost shall be defined based on the time a data block occupies network, memory or CPU.

Execution result information corresponding to queries executed in a predetermined period is stored in a predetermined storage area. , the frequency of transmission of data blocks between multiple storage areas and the transmission cost of data blocks. Specifically, based on the acquired execution result information, the transmission history identification unit 121 determines the transmission frequency matrix F indicating the transmission frequency between the plurality of storage areas of the data blocks associated with the processing executed in a predetermined period, and the data A transmission cost matrix C indicating the transmission cost of the block is specified.

The transmission frequency matrix F indicates the frequency of transmission to the storage area of each address of each data block by queries executed during a predetermined period. 4 is a diagram showing a table corresponding to the transmission frequency matrix F. FIG. In the table shown in FIG. 4, the row direction indicates the index of the data block, and the column direction indicates the address of the storage area of the transmission destination. An element _fik corresponding to row i and column k indicates the transmission frequency, which is the number of times the i-th data block has been transmitted to the storage area at address k. For example, the element _f12 of the transmission frequency matrix F indicates that the frequency of transmission of data block A (first data block) to the storage area at address 2 is 5 times.

The transmission cost matrix C indicates the transmission cost per data block from the transmission source storage area to the transmission destination storage area. FIG. 5 is a diagram showing a table corresponding to the transmission cost matrix C. FIG. In the table shown in FIG. 5, the row direction indicates the address of the transmission source storage area, and the column direction indicates the address of the transmission destination storage area. In the table shown in FIG. 5, the element _cjk corresponding to j row and k column indicates the transmission cost per data block from the storage area at j address to the storage area at k address. For example, the element c ₃₂ of the transmission cost matrix C indicates that the transmission cost is 2 when the data block is transmitted from the storage area at the 3rd address to the storage area at the 2nd address.

The transmission cost is 1) the cost associated with reading a data block at the same address, 2) the cost associated with writing a data block at the same address, 3) the cost associated with reading a data block at a different address, and 4) the cost associated with a different address. It is also possible to divide and manage the cost associated with writing data blocks in . Furthermore, the transmission frequency may be managed in accordance with 1) to 4). In the following, for the sake of simplicity of explanation, these are evaluated without being separated.

The storage capacity specifying unit 122 specifies the storage capacity of each of the plurality of storage areas. For example, a distributed system stores storage capacity information that associates the addresses and storage capacities of each of a plurality of storage areas in a predetermined storage area. The storage capacity specifying unit 122 specifies the storage capacity of each of the plurality of storage areas by acquiring storage capacity information stored in a predetermined storage area. FIG. 6 is a diagram showing an example of storage capacity information. In this embodiment, the storage capacity is indicated by the number of data blocks that can be stored in the storage area. As shown in FIG. 6, for example, the storage capacity of the storage area at address 3 is 2, indicating that two data blocks can be stored.

The creating unit 123 includes a plurality of variables corresponding to a placement combination in which a plurality of data blocks are placed in a plurality of storage areas, and transmits data when the distributed system re-executes a process executed in a predetermined period for the placement combination. Based on the transmission frequency indicated by the frequency matrix F and the transmission cost indicated by the transmission cost matrix C, an optimization function is created that specifies a combination of variable values that minimizes the total transmission cost. The creating unit 123 creates an optimal data block including a plurality of variables corresponding to a layout combination for arranging a plurality of data blocks in a plurality of storage areas so as not to exceed the storage capacity of each of the plurality of storage areas specified by the storage capacity specifying unit 122. create a function The creation of the optimization function by the creation unit 123 will be described in detail below.

First, the creation unit 123 defines a binary variable _xij indicating whether or not the i-th data block is to be placed in the j-address storage area, and assigns each of the plurality of data blocks to the j-address storage area. An arrangement matrix X indicating whether or not to arrange is created. x _ij is represented by the following formula (1).

In the quantum Ising machine, the spin s _ij , which is the variable to be optimized, is a variable that takes either +1 or −1 _. can be transformed into a binary variable x _ij that takes Therefore, the binary variables x _ij can be the variables to be optimized in the quantum Ising machine.

Since the arrangement matrix X is a set {x _ij } of binary variables x _ij , it is also expressed as X={x _ij }. FIG. 7 is a diagram showing a table corresponding to the arrangement matrix X. FIG. As shown in FIG. 7, it can be confirmed that the binary variable _xij , which is an element of the arrangement matrix X, is either 1 or 0. FIG.

Optimal placement of a plurality of data blocks in each storage area is specified by solving the optimization problem represented by the following equations (3) and (4).

Equation (3) represents an objective function for obtaining an arrangement matrix X that minimizes the total cost based on the transmission frequency and transmission cost when the processing executed in a predetermined period based on the arrangement matrix X is re-executed. C _ik (X) indicates the total transmission cost for transmitting the i-th block to the destination storage area (k-address storage area).

Equation (4) indicates a constraint that the number of data blocks stored in each storage area should be less than or equal to the storage capacity. By combining the objective function shown in Equation (3) and the constraint conditions shown in Equation (4), the optimization problem of finding the arrangement matrix X that minimizes the total transmission cost is obtained.

Furthermore, for each of a plurality of data blocks, one data block may be duplicated and arranged in a plurality of storage areas, and a restriction may be imposed that the number of duplicates of the data block is a predetermined number r. In this case, the optimal allocation of the plurality of data blocks to each storage area is specified by solving the optimization problem given by Equations (3), (4) and (5) below.

[Create binary variable]
In a quantum Ising machine, all variables must be binary variables that take either 1 or 0. The creation unit 123 creates an optimization function in a form that can be evaluated by a quantum Ising machine, expressed by a binary variable, for a part that takes a large number of values in the optimization problem represented by the equations (3) and (4). do.

[Creation of binary variable for formula (3)]
First, the creating unit 123 defines a binary variable for Equation (3). C _ik (X) in equation (3) indicates the transmission cost when transmitting the i-th data block to the k-th storage area, but a plurality of storage areas can be taken as the transmission source storage area ( can take many values). On the other hand, the creation unit 123 creates a first binary variable group as a first binary variable group for specifying the transmission source storage area and the transmission destination storage area of each of the plurality of data blocks corresponding to the arrangement combination. Create an auxiliary variable v _ijk . The first auxiliary variable v _ijk is 1 when the storage area at address j is used as the storage area of the transmission source when the i-th block is transmitted to the storage area at address k, and the storage area at address j is not used. is a variable that becomes 0 when

FIG. 8 is a table showing an example of the first auxiliary variable. In FIG. 8, each element has a value outside the parentheses and a value inside the parentheses. Values outside parentheses indicate the value of the first auxiliary variable v _ijk . The values in parentheses indicate the values of the elements x _ij of the arrangement matrix X. In FIG. 8, for example, when (i, k) is (2, 3) and j is 3, it is indicated as 1 (1). This means that the memory area at address 3 (j=3) is used as the transmission source memory area when the second data block (i=2) is transmitted to the memory area at address 3 (k=3). It also shows that the second (i=2) data block exists in the third (j=3) storage area.

Here, the transmission frequency f _ik , the arrangement matrix element x _ij , and the first auxiliary variable v _ijk must satisfy the following equation (6).

In equation (6), the condition f _ik >0 is provided to save the number of first auxiliary variables v _ijk . f _ik =0 indicates that the i-th data block was not transferred to the storage area at address k due to the processing executed within the predetermined time, _and the first auxiliary variable v For _ijk , there is no need to consider the transmission cost. For this reason, based on the transmission frequency f _ik , the creating unit 123 limits the combination of the transmission source storage area and the transmission destination storage area to which the transmission was performed in the process executed during the predetermined period, and selects the first Create an auxiliary variable v _ijk . In the example shown in FIG. 8, it can be seen that the first auxiliary variables v _ijk are limited to the first auxiliary variables v _ijk corresponding to f _ik >0 shown in FIG.

By doing so, the number of first auxiliary variables v _ijk can be suppressed by not creating the first auxiliary variables v _ijk corresponding to f _ik =0. Although the number of qubits that can be handled by the quantum Ising machine is limited, by suppressing the number of the first auxiliary variables v _ijk , the number of the first auxiliary variables v _ijk exceeds the number of qubits that can be handled by the Ising machine. It is possible to suppress the process from becoming unexecutable in

Further, in Equation (6), for the i-th data block transmitted to the storage area at address k, Σ _j v _ijk =1 means that the storage area of the transmission source (storage area of address j) is the storage area of the transmission destination. This indicates that one area (storage area at address k) must be fixed. Σ _j x _ij v _ijk =1 indicates that the i-th data block exists in the storage area of the transmission source (storage area at j address). In the example shown in FIG. 8, for example, when (i, k) is (2, 3), the storage area of the transmission source (storage area of j address) becomes the storage area of the transmission destination (k=3). It can be confirmed that the second data block exists in the transmission source storage area (storage area at address 3), as well as the storage area at address 3 (j=3).

By using the first auxiliary variable v _ijk , C _ik (X) in Equation (3) can be expressed by Equation (7) below.

As described above, c _jk indicates the transmission cost when transmitting one data block from the transmission source storage area (storage area at j address) to the transmission destination storage area (storage area at k address). The frequency f _ik indicates the transmission frequency at which the i-th block is transmitted to the storage area of the transmission destination (storage area at address k). Here, since c _jk f _ik includes an element j, _j is set to one like the first auxiliary variable v _ijk shown in _{FIG .} By multiplying C ik (X), C _ik (X) is the total transmission cost if all the processes performed within the predetermined period are re-executed.

Also, since the first auxiliary variable v _ijk is constrained by the elements x _ij of the placement matrix X, it can be said that it is a function related to the placement matrix X. From equation (7), equation (1) can be expressed by equation (8) below. In equation (8), V denotes the set {v _ijk } of first auxiliary variables v _ijk .

[Creation of binary variable for formula (4)]
Next, the creating unit 123 defines a binary variable for Equation (4). Σ _i x _ij in equation (4) indicates the number of data blocks arranged in each of the plurality of transmission source storage areas, and can take many positive integer values. On the other hand, the creating unit 123 creates a second auxiliary variable _yjm as a second binary variable group indicating the number of data blocks stored in each of the plurality of storage areas. The second auxiliary variable y _jm is given by the following formula (9) for each positive integer value (1≦m≦t _j ) less than or equal to the storage capacity specified by the storage capacity specifying unit 122 for each storage area. Created as shown.

FIG. 9 is a table showing an example of the second auxiliary variable y _jm . In FIG. 9, for example, the value in row 3 (j=3) indicates that the number of data blocks arranged in the storage area at address 3 is one.

With the second auxiliary variable y _jm , equation (4) can be transformed into equation (10) shown below.

Equation (10) indicates the total storage capacity of data blocks arranged in each storage area (the number of data blocks), and the total storage capacity of data blocks arranged in each storage area is the total storage capacity of each storage area. It indicates that it is less than the storage capacity of the area. Let Y be the set {y _jm } of the second auxiliary variables y _jm .

[Create optimization function]
Subsequently, the creating unit 123 creates an optimization function H that can be solved by the quantum Ising machine, including the first auxiliary variable v _ijk and the second auxiliary variable y _jm . The format of the optimization function H is called QUBO (Quadratic Unconstrained Binary Optimization), which is an unconstrained quadratic polynomial described only with binary variables.

The optimization function H is represented by the following formula (11), for example, by combining formulas (5), (6), (7), and (10).

Formula (7) is a formula for optimizing V, but Formulas (6) and (10) include X, Y, and V, so they are variable groups in which X, Y, and V change. Therefore, the optimization function H optimizes these variable groups X, Y, and V. FIG.

The fourth and fifth terms on the right side of equation (11) are constraint terms corresponding to the constraint expressions of equations (6) and (10). Equations (6) and (10) are constraint equations for the first auxiliary variable v _ijk and the second auxiliary variable y _jm . , and a constraint term for minimizing this is provided, and in the solution of the optimization function H, (left side - right side) = 0, that is, left side = right side, so that the constraint indicated by the constraint expression is satisfied. Similarly, the sixth term is also a constraint term corresponding to the constraint expression of Equation (5), and is in the form of squaring (left side - right side). The second term is a constraint term corresponding to Σ _j v _ijk =1 in the constraint expression of Equation (6), and the second term is in the form of (left side−right side) squared. On the other hand, the third term is a constraint term corresponding to Σ _{j x ij} v _ijk =1 in the constraint expression of Equation (5) _{. ijk} ) is never negative, and squaring it results in a quartic, so it is not squared. Here, λ ₀ to λ ₄ in Equation (11) are coefficients of constraint terms, and constraint terms to be satisfied more strongly can be set.

It should be noted that the third and fourth terms in equation (11) can be converted to λ ₂ Σ _j (Σ _i x _ij −Σ _{m: 1≦m≦tj} y _im ) to achieve the same effect. is known (in this case the auxiliary variables take different values than in equation (9), but the effect of constraining them to satisfy the same constraints as the third and fourth terms is preserved). Therefore, the creation unit 123 may create equation (12) instead of equation (11). By doing so, the coefficients of the constraint terms can be reduced, making it easier to adjust the coefficients of each constraint term.

In the above explanation, it is assumed that the data _capacity of the data block is constant. You may In equation (13), _tj indicates the data size rather than the maximum number of allocatable data blocks, and _cjk indicates the transmission cost per unit data size.

In the above description, the number of copies of the data block is assumed to be the predetermined number r, but the number of copies of the data block may be a first predetermined number or more and a second predetermined number or less. For example, when the number of replications is set to 1 or more and 2 or less, the last terms of equations (11) to (13) should be changed to equation (14).

Also, when the number of replications is 1 or more and 3 or less, an auxiliary variable z _il indicating the number of replications of each data block is introduced, and the last term of Equations (11) to (13) is changed to Equation (15). Just do it. The idea is the same as the transformation in equation (12).

The arrangement combination identifying unit 124 causes the quantum Ising machine to evaluate the optimization function generated by the generating unit 123 to obtain a suboptimal solution, and causes the quantum Ising machine to identify a combination of multiple variables that brings the total cost closer to the minimum. specifies the placement combination that minimizes the total cost. FIG. 10 is a diagram showing a table corresponding to the arrangement matrix X corresponding to x _ij specified by the quantum Ising machine. As shown in FIG. 10, it can be confirmed that data blocks are arranged in each storage area within a range smaller than the storage capacity of each storage area.

The placement control unit 125 places the plurality of data blocks in the plurality of storage areas based on the placement combination specified by the placement combination specifying unit 124 . As a result, the allocation of data blocks to storage areas is optimized. For example, the placement control unit 125 places each data block based on the difference between the placement matrix before optimization and the placement matrix specified by the placement combination specifying unit 124 .

The allocation control unit 125 may rearrange data during data transmission associated with data processing. For example, when the i-th data block is located in the j-address storage area, and the allocation destination storage area of the data block specified by the allocation combination specifying unit 124 is the k-address storage area, the allocation control unit 125 Alternatively, the data may be rearranged when the distributed system executes the process of transmitting the data block to the k address. Also, the allocation control unit 125 may rearrange the plurality of data blocks all at once at the time set by the administrator of the distributed system.

[Flow of processing in information processing device 1]
Next, an example of the flow of processing in the information processing device 1 will be described. FIG. 11 is a flow chart showing the flow of processing in the information processing apparatus 1. As shown in FIG.
First, the transmission history identifying unit 121 identifies a transmission history indicating the frequency of data block transmission between multiple storage areas and the transmission cost of the data block associated with processing executed during a predetermined period (S1).
Subsequently, the storage capacity specifying unit 122 specifies the storage capacity of each of the plurality of storage areas (S2).

Subsequently, the creating unit 123 creates an optimization function for specifying an arrangement combination for arranging a plurality of data blocks in a plurality of storage areas (S3).
Subsequently, the placement combination identifying unit 124 causes the quantum Ising machine to evaluate the optimization function created in S3, and specifies a placement combination that minimizes the total transmission cost (S4).
The placement control unit 125 places the plurality of data blocks in the plurality of storage areas based on the placement combination specified by the placement combination specifying unit 124 (S5).

[Effects of this embodiment]
As described above, the information processing apparatus 1 according to the present embodiment stores a plurality of variables corresponding to arrangement combinations for arranging a plurality of data blocks in a plurality of storage areas so as not to exceed the storage capacities of the plurality of storage areas. Optimal for specifying a combination of variable values that minimizes the total transmission cost based on the transmission frequency and the transmission cost when the distributed system re-executes the processing executed in the predetermined period for the arrangement combination Create an optimization function in a form that is an optimization function that can be evaluated by a quantum Ising machine. Then, the information processing apparatus 1 causes a quantum Ising machine to evaluate the created optimization function, and specifies a combination of values of a plurality of variables that brings the total cost closer to the minimum. identify. By doing so, the information processing apparatus 1 can specify a storage area suitable for arranging data based on the frequency of use of the data.

Although the present invention has been described above using the embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments, and various modifications and changes are possible within the scope of the gist thereof. be. For example, all or part of the device can be functionally or physically distributed and integrated in arbitrary units.

For example, in the above-described embodiment, the transmission cost is specified based on the transmission history specified by the transmission history specifying unit 121, but the present invention is not limited to this. The information processing device 1 may specify the transmission cost by accepting input of information indicating the transmission cost from the user. In addition, new embodiments resulting from arbitrary combinations of multiple embodiments are also included in the embodiments of the present invention. The effect of the new embodiment caused by the combination has the effect of the original embodiment.

Reference Signs List 1 information processing device 11 storage unit 12 control unit 121 transmission history identification unit 122 storage capacity identification unit 123 creation unit 124 Layout combination identification unit 125 . . . Layout control unit

Claims (7)

In a distributed system in which a plurality of data blocks with a fixed data capacity are arranged in a plurality of storage areas with different addresses, the frequency of transmission of the data blocks between the plurality of storage areas associated with processing executed during a predetermined period of time is indicated. a transmission history identifying unit that identifies a transmission history;
a storage capacity specifying unit that specifies the storage capacity of each of the plurality of storage areas;
including a plurality of variables corresponding to an arrangement combination for arranging the plurality of data blocks in the plurality of storage areas so as not to exceed the storage capacity of each of the plurality of storage areas specified by the storage capacity specifying unit; A combination of variable values that minimizes the total transmission cost based on the transmission frequency and the transmission cost of the data block when the distributed system re-executes the processing executed during the predetermined period. a creation unit that creates a second-order polynomial optimization function in which the variables are described only by binary variables, which is a format that can be evaluated by a quantum Ising machine;
The optimization function created by the creating unit is evaluated by the quantum Ising machine, and a combination of values of the plurality of variables that brings the total cost closer to the minimum is identified, thereby minimizing the total cost. an arrangement combination identifying unit that identifies an arrangement combination;
with
The creation unit creates a first binary variable group for specifying a transmission source storage area and a transmission destination storage area of each of the plurality of data blocks corresponding to the arrangement combination, creating a second binary variable group indicating the number of the plurality of data blocks stored in the plurality of storage areas corresponding to the arrangement combination, the first binary variable group and the second binary variable creating said optimization function comprising a group and
Information processing equipment. Based on the transmission frequency, the creation unit creates the first binary variable group by limiting to combinations of transmission sources and transmission destinations to which transmission was performed in the processing executed during the predetermined period. ,
The information processing device according to claim 1 . The creation unit allows each of the plurality of data blocks to be duplicated and arranged in a plurality of storage areas, and is an optimization function that identifies the combination, and is an optimization function in a format that can be evaluated by a quantum Ising machine. create a function,
The information processing apparatus according to claim 1 or 2 . The creation unit specifies the combination so that the number of copies when each of the plurality of data blocks is copied and arranged in a plurality of storage areas is equal to or greater than a first predetermined number and equal to or less than a second predetermined number. creating an optimization function in a form that can be evaluated by a quantum Ising machine,
The information processing apparatus according to claim 3 . The transmission history identifying unit identifies the transmission history indicating the frequency of transmission of the data blocks between the plurality of storage areas and the transmission cost of the data blocks associated with processing executed during a predetermined period of time,
The creation unit calculates, for the arrangement combination, a total cost of the transmission costs based on the transmission costs indicated by the transmission frequency and the transmission history when the distributed system re-executes the processing executed during the predetermined period. creating the optimization function that identifies the combination of values of the variables that minimizes
The information processing apparatus according to any one of claims 1 to 4 . further comprising an arrangement control unit for arranging the plurality of data blocks in the plurality of storage areas based on the arrangement combination specified by the arrangement combination specifying unit;
The information processing apparatus according to any one of claims 1 to 5 . the computer runs
In a distributed system in which a plurality of data blocks with a fixed data capacity are arranged in a plurality of storage areas with different addresses, the frequency of transmission of the data blocks between the plurality of storage areas associated with processing executed during a predetermined period of time is indicated. identifying a transmission history;
identifying the storage capacity of each of the plurality of storage areas;
including a plurality of variables corresponding to an arrangement combination for arranging the plurality of data blocks in the plurality of storage areas so as not to exceed the storage capacity of each of the specified plurality of storage areas, and for the arrangement combination, Optimization for identifying a combination of variable values that minimizes the total transmission cost based on the transmission frequency and the transmission cost of the data block when the distributed system re-executes the processing executed during the predetermined period creating a quadratic polynomial optimization function in which the variables are described only by binary variables, the function being in a form that can be evaluated by a quantum Ising machine;
The arrangement combination is specified so as to minimize the total cost by causing the quantum Ising machine to evaluate the created optimization function and specifying a combination of values of the plurality of variables that brings the total cost closer to the minimum. and
has
In the creating step, the computer creates a first binary variable group for specifying a transmission source storage area and a transmission destination storage area for each of the plurality of data blocks corresponding to the arrangement combination. and creating a second binary variable group indicating the number of the plurality of data blocks stored in the plurality of storage areas corresponding to the arrangement combination, the first binary variable group and the first creating the optimization function comprising a binary variable group of 2;
Information processing methods.

Images