JP2003296150A - File relocation system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a file relocation system capable of efficiently performing the relocation of files which simultaneously satisfies various purposes in a short time. <P>SOLUTION: In this file relocation system, a data working processing means 12 generates the input and output count number and file capacity of each file and locating state information showing whether each file is actually located to each volume or not from the access record to the file group 2 of a host online system 1 outputted from a resource information collecting system 3. A file optimum location processing means 15 determines the optimum location of files by operation by genetic algorithm using each information generated from the data working processing means 12. A file moving processing means 19 performs the relocation of files in the file group 2 according to the optimum location of files. <P>COPYRIGHT: (C)2004,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a file relocation system for determining a file relocation position, and more particularly to a file relocation system capable of reducing the time and labor required for relocation.

[0002]

2. Description of the Related Art In online processing or batch processing of a general-purpose computer, the time required to access a file on a disk greatly affects the performance of the computer. For this reason, in a general-purpose computer, it is important to relocate the file to the optimum position on the disk periodically in order to reduce the access time. Conventionally,
The relocation of files was done manually by experts, relying on intuition.

[0003]

However, the above-mentioned conventional file relocation method has a problem that it is not possible to perform efficient file relocation in a short time.
In particular, since the host online system operates in full time, there is a restriction that maintenance time cannot be taken sufficiently. Therefore, it may take, for example, 6 weeks to complete the rearrangement of files.

Further, since it is necessary to relocate files so as to simultaneously satisfy various purposes such as the number of times of access or leveling of file capacity, it is difficult to determine the location of relocation. It was

The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a file rearrangement system capable of efficiently rearranging files that simultaneously satisfy various purposes in a short time. .

[0006]

DISCLOSURE OF THE INVENTION The present invention for solving the above-mentioned problems of the prior art is, in a file relocation system, relocating a file placed in a plurality of volumes in a computer to a plurality of volumes in a computer.
A genetic algorithm is used to calculate the I / O load and file capacity of each file, and the placement status information that indicates whether each file is actually placed in each volume. The optimal placement determination process that determines the optimal placement volume is performed, and each file is relocated to the corresponding volume based on the processing result. Relocation of files that simultaneously satisfy various purposes can be performed efficiently in a short time. Can be done on a regular basis. The computer is preferably a general purpose computer.
The I / O load is represented by I / O count × block size. In an online system, the block size of a file is almost constant, so it can be replaced by the I / O count.

Further, the present invention is a file relocation system for relocating files arranged in a plurality of volumes in a computer to a plurality of volumes in a file relocation system. A data processing unit for generating and outputting the input / output load and file capacity of the file and the arrangement state information indicating whether or not each file is actually arranged for each volume, and the data processing unit. Based on the processing result of the optimum file placement section, and the file best placement section that performs the best placement decision process that determines the volume that is the best or the best placement for each file by performing a genetic algorithm operation using the data And a file move processing unit that relocates each file to the corresponding volume. Are those, can be performed efficiently relocation of files that meet a variety of purposes at the same time in a short time.

In the file rearrangement system according to the present invention, the genetic algorithm is a gene generating means for generating a plurality of random gene data in which the arrangement state of each file in the volume is expressed in binary notation, Gene evaluation means that evaluates gene data using an evaluation function that indicates the suitability of file allocation for volume in file relocation, and the remaining genes except for a specified number of gene data with low evaluation based on the evaluation result by the evaluation function. Group extracting means for extracting a group of data and two arbitrary gene data are extracted from the group, a part of the extracted gene data is exchanged to generate new gene data, and the generated gene data is used to identify the group. Reconstructed crossing means and arbitrary gene data extracted from the group, extracted data Generating new gene data by changing a part, and having a mutation means for reconstituting a group with the generated gene data, the gene data generated by the gene generation means is evaluated by the gene evaluation means. , The group extraction means extracts the groups, the crossing means and the mutation means reconstruct the groups, and further, the gene evaluation means evaluates the gene data in the groups reconstructed by the crossing means and the mutation means. However, it is an algorithm that repeats the processing in the crossing means, the mutation means and the gene evaluation means until the gene evaluation means satisfies the prescribed condition. It outputs the highest or almost the highest gene data to the file movement processing unit, using the evaluation function. To evaluate the genetic data generated by the heat algorithm can be efficiently performed in a short time relocation of files that meet a variety of purposes at the same time.

According to the present invention, in the above file relocation system, the evaluation function is a value obtained by multiplying a sum of ratios of input / output counts in each volume to an average value of input / output counts per volume by a predetermined weighting factor. A value obtained by multiplying the sum of the ratio of the file capacity in each volume to the average value of the file capacity per volume by a predetermined weighting coefficient, and a sum of the ratio of the number of files in each volume to the average value of the number of files per volume The value obtained by multiplying by is used to evaluate the gene data by adding these values, and the I / O count, file capacity and number of files in each volume can be leveled simultaneously under certain conditions. The optimum arrangement can be efficiently obtained in a short time.

Further, according to the present invention, in the above file relocation system, the evaluation function uses a value obtained by multiplying a total sum of penalties determined corresponding to the number of files in each volume by a predetermined weighting coefficient. Is added to evaluate the gene data, and it is possible to efficiently find the optimum arrangement of files that can level the number of files in each volume under a certain condition in a shorter time.

Further, according to the present invention, in the above file relocation system, the evaluation function uses a value obtained by multiplying the number of volumes whose input / output count is not within a certain range by a predetermined weighting coefficient, and adding this value. Then, the gene data is evaluated, and the optimum arrangement of files that can level the input / output counts in each volume under a certain condition can be efficiently obtained in a shorter time.

According to the present invention, in the above file relocation system, the evaluation function uses a value obtained by multiplying the number of volumes whose file capacity is not within a certain range by a predetermined weighting coefficient, and adding this value. The gene data is evaluated by using the above-mentioned method, and the optimum file arrangement that can equalize the file capacity in each volume under a fixed condition can be efficiently obtained in a shorter time. Further, the present invention is
In the file relocation system described above, the file optimum placement unit can set as a condition the files to be placed in different volumes from the viewpoint of avoiding input / output conflict and exclusive control. Access conflict of the same file between transactions can be avoided.

According to the present invention, in the above file rearrangement system, the file optimum arrangement unit extracts, in the group extraction means, a prescribed upper number of gene data from the gene data belonging to the group, and in the group reconstruction means, If a part of the gene data that has not been extracted is partially exchanged or changed, and in the gene evaluation means, there is gene data that has a higher evaluation than the extracted gene data, the gene data that has a high evaluation and the It replaces the lowest-evaluated gene data among the extracted gene data, and can promote the generation of higher-evaluated gene data, so the rearrangement of files that simultaneously satisfy various purposes can be made even shorter. It can be done efficiently in time.

Further, in the above-mentioned file rearrangement system according to the present invention, the file optimum arrangement part is partially changed at the time of group rearrangement which is continuously performed when the gene data of the highest evaluation is not generated in the gene evaluation means for a prescribed period. Since the mutation can be appropriately generated, the rearrangement of files that simultaneously satisfy various purposes can be efficiently performed in a shorter time.

Further, the present invention is a host online system characterized in that files are reallocated using the file reallocation system described above, and the reallocation of files satisfying various purposes at the same time is achieved in the host online system. It can be done efficiently in a short time and can improve the performance of the host online system.

[0016]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described with reference to the drawings. The file relocation system according to the embodiment of the present invention, when relocating files arranged in a plurality of volumes in a host online system to a plurality of volumes, the input / output count number and file capacity of each file, and each volume By the genetic algorithm using the placement state information of each file to the, to find the optimal or near-optimal placement of the file, to relocate the file according to the placement,
This makes it possible to relocate files that simultaneously satisfy various purposes simultaneously in a short time, and reduce the time and labor required for relocating files.

The data processing unit in the claims corresponds to the resource usage status storage unit 11, the data processing unit 12 and the processed data storage unit 13 of FIG. 1, and the file optimal placement unit is the control storage unit 14 and the file optimal placement unit. The determination processing unit 15 and the file arrangement data storage unit 16 correspond to the file movement processing unit, and the file movement processing unit corresponds to the file movement JCL creation processing unit 17, the file movement processing JCL storage unit 18, and the file movement processing unit 19, respectively. The gene generating means in the claims is S1 in FIG. 5, and the gene evaluating means is S3.
The group extracting means corresponds to S5, the crossing means corresponds to S6, and the mutation means corresponds to S7.

The configuration of the file relocation system according to the embodiment of the present invention will be described with reference to FIG.
FIG. 1 is a block diagram of an overall configuration including a file relocation system according to an exemplary embodiment of the present invention. File relocation system according to an embodiment of the present invention (this system)
As shown in FIG. 1, has a host online system 1, a file group 2, a resource information collection system 3, and a file relocation system 4 as a whole.

Each part of the apparatus will be specifically described.
The host online system 1 is connected to a terminal device (not shown) via a network such as a LAN, performs online processing based on a processing command input from the terminal device, and outputs the processing result via the network. It is output to the terminal device.

The file group 2 is a set of files used in the host online system 1. File group 2 is
It is stored on a disk consisting of multiple volumes,
Reading or writing is performed by access from the host online system 1.

The resource information collecting system 3 is connected to the host online system 1, monitors the access status to the file group 2 in the host online system 1, collects the access record (resource usage status), and recollects the file. Output to the placement system 4.

The file relocation system 4 is a characteristic part of the present invention. From the access record of the host online system 1 output from the resource information collection system 3, the IO (Input / Output) count number and file capacity of each file And the file allocation status information for each volume is generated, and based on each generated information, the process of determining the volume that provides the optimum allocation for all files under certain conditions (hereinafter, the file optimal allocation determination process) I do. The file relocation system 4 relocates the files in the file group 2 based on the result of the file optimal placement determination process.

Next, the configuration of the file relocation system (present system) 4 will be described. As shown in FIG. 1, the present system includes a resource usage status storage unit 11, a data processing unit 12, a processed data storage unit 13, a control information storage unit 14, and a file optimal placement determination processing unit 15.
And the file arrangement data storage unit 16 and the file move J
It has a CL (Job Control Language) creation processing means 17, a file movement processing JCL storage section 18, and a file movement processing means 19.

Next, each part of this system will be specifically described. The resource usage status storage unit 11 stores and stores the access record in the host online system 1 output from the resource information collection system 3. The data processing unit 12 reads the access record stored in the resource use status storage unit 11 and processes the access record into information on the IO count number and file capacity of each file and the file allocation state in each volume. , Processing data storage unit 1
Output to 3. The processed data storage unit 13 stores and stores information processed by the data processing unit 12.

The control information storage unit 14 stores various processing instructions used by the file optimal placement determination processing means 15, and the processing instructions are sent to the file optimal placement determination processing means 15 by access from the file optimal placement determination processing means 15. Output. File optimal placement determination processing means 15
Is a genetic algorithm (Genetic Algorithm) described later.
m: hereinafter referred to as GA), the optimum placement determination processing of the file is performed, and the optimum placement information of the file which is the processing result is output to the file placement data storage unit 16. The file optimum layout determination processing means 15 accesses the control information storage unit 14 and reads a necessary processing command if necessary in the optimum layout determination processing.

The file allocation data storage unit 16 stores and stores the optimum allocation information of the file obtained by the file optimum allocation determination processing means 15. File move J
The CL creation processing means 17 includes the file arrangement data storage unit 1
A processing instruction for allocating a file according to the optimum allocation information of the file stored in 6 is generated by using JCL, and is output to the file move JCL storage unit 18. The file move processing JCL storage unit 18 stores the file move JC.
The file placement processing command generated by the L creation processing means 17 is stored and stored. The file migration processing means 19 relocates each file of the file group 2 to the corresponding volume based on the file layout processing command stored in the file migration JCL storage unit 18.

Next, the GA used in the file optimum placement determination processing means 15 of this system will be described. In the file relocation processing problem, the main purpose is to equalize the IO counts of disks of multiple volumes (hereinafter abbreviated as volumes), but in the sense of speeding up disk access, Leveling the file capacity is also important. In an actual host online system, distributed file processing is performed to distribute one file and allocate it to multiple volumes in order to avoid transaction conflicts, which causes file relocation problems. Is more complex.

As one of the solutions to the file relocation processing problem, the problem is modeled as an integer programming problem consisting of a plurality of objective functions and constraints corresponding to the objective functions,
A multi-objective integer mixed programming method for obtaining Pareto optimal solutions of these objective functions is known, but there is a restriction that the constraint condition must be linear (linear expression), and it is optimal when the constraint condition is complicated. It is difficult to find a solution. On the other hand, according to the GA used in the present system, the constraint condition is not limited to the linear expression and can be set relatively freely, so that it is very effective for the integer programming problem that requires complicated constraint.

In the GA used in the present system, the data indicating the arrangement state of files in each volume is regarded as gene data (hereinafter referred to as genes) in the first step, and a plurality of random genes are generated, and in the second step. Genes generated as are evaluated and classified based on the evaluation result, and a gene having a higher evaluation is obtained by performing a crossover or generating a mutation with respect to genes belonging to a highly evaluated gene group in the third step. Is generated to obtain the optimum solution for file relocation. It is preferable that the plurality of genes have a constant length.

First, the structure of the gene generated in the first step and the method for generating it will be described with reference to FIG. FIG. 2 is an explanatory diagram of a gene generation method in GA used in the present system, in which FIG. 2 (a) is a table showing the arrangement state of each file in a volume,
FIG. 2B is a configuration diagram of genes generated by this system. The table of FIG. 2A shows the correspondence between the file number and the volume number in which the file is to be arranged. In FIG. 2A, in addition to the file number and the number of the allocation destination volume, the binary representation of the volume number (see FIG.
3 bits) are shown. For example, since the file 1 is arranged in the volume 1, the binary representation of the volume number is "001".

In the GA of this system, the arrangement of each file
Combines the binary representation of the previous volume numbers into a file
Genes showing the arrangement state of Inheritance in Figure 2 (b)
The child corresponds to the table in Fig. 2 (a), and the distribution of each file is
The binary number of the destination volume number is the file number
They are arranged in order from the left. Ie this system
Genes generated by the GA of the system are located in each file
It can be said that it is a combination of volume numbers. In FIG.
This system uses 3 bits for the ryume number.
, There are N files to be relocated and
Bit length of volume number when there are M volumes
Is log_TwoM bits, the gene bit length is Nlo
g _TwoIt becomes M bits.

Next, the G of the present system in the second stage
Classification of genes in A will be described with reference to FIG. FIG. 3 is an explanatory diagram showing a preferable gene extraction and classification situation in this system. In FIG. 3, a gene group 31, which is a set of genes generated in the present system, is evaluated for each gene in GA, and genes that have been highly evaluated are extracted. Among the extracted genes, the upper specified number of genes are separated from the gene group 31 and classified as the upper gene group 32, and the remaining highly evaluated genes are classified as the crossover target gene group 34. Genes with a low evaluation are removed from the gene group 31 as the lower gene group 33.

Further, the GA of the present system extracts two genes out of the genes classified in the crossover target gene group 34 and crosses them to generate a new gene. This is performed a specified number of times (n times in this system). In some cases,
An appropriate number of genes are extracted from the genes classified in the crossover target gene group 34 to generate mutations. After performing the crossover or mutation, the GA of the present system evaluates the genes classified into the crossover target gene group 34, and the gene having the higher evaluation (highest evaluation gene 341) is higher than any of the genes in the upper gene group 32. If present, the gene is replaced with the lowest-evaluated gene in the upper gene group 32. The gene evaluation method will be described later.

Crossing and mutation generally performed in GA will be described with reference to FIG. FIG. 4 shows G
It is explanatory drawing of the crossover and mutation in A. In FIG. 4, gene A and gene B are both 18 bits long. In FIG. 4, the GA extracts the upper 12 bits of the gene A and the lower 6 bits of the gene B and connects them (crossover). With respect to the gene C obtained as a result of the crossover, the GA reverses (mutates) the numbers in the upper 6th bit and the 10th bit to obtain a new gene C ′.

In GA, crossing is performed for the purpose of multiplying highly evaluated genes to generate a more highly evaluated gene. In addition, when evolution is stagnant and a highly evaluated gene cannot be generated, an abnormal gene is generated as a mutation, which has the effect of promoting new evolution. In the GA of this system, gene crossover and mutation are realized by performing the same operation in the third step.

The specific operation of the file optimum layout determination process by the GA in this system will be described with reference to FIGS. FIG. 5 is a flow chart of the file optimal placement determination process by the GA in this system.
As shown in FIG. 5, when GA is operated in this system, first, the allocation state of all files in the volume is represented by an integer expression of 0 or 1 (0-1), and the file allocation is formulated. (S1). Furthermore, the system randomly generates a plurality of genes using the integer representation of the file arrangement formulated in S1 (S2).

In this system, in S1 of FIG. 5, the volume number of the file allocation destination is replaced with a binary number representation as shown in FIG. 2A for formulation, and in S2, it is formulated as shown in FIG. 2B. Genes with the constitution shown in are randomly generated. That is, the present system randomly determines the volume of the allocation destination for all files and repeats the process of generating a combination of those volume numbers as a gene to generate a plurality of genes.

After generating the genes, the system applies the generated genes to an evaluation function to evaluate each gene (S3). Hereinafter, a method for evaluating a gene will be described. First, the present system calculates the IO count, file capacity, and number of files in each volume when files are arranged according to gene information for each gene, and further calculates the IO count and file capacity of each file and the actual The average value of the IO count, the file capacity, and the number of files per volume is calculated from the file allocation state information on the volume. The system then converts the IO count, file capacity, and number of files in each volume into a quantity that represents the ratio of each to the average value. The conversion is performed for each volume as shown in equations (1) to (3), and IO count,
The file capacity and the number of files are IO _i ,
S _i and N _i (i is a volume number) are obtained. The obtained quantity is used in the evaluation function described later.

[0039]

[Equation 1]

In the gene evaluation, the present system also evaluates the number of files arranged in each volume. In this system, in addition to N _i, the number of files placed in one volume is checked, and as the number of placed files increases, a penalty is deducted, and the obtained penalty P _i is evaluated by an evaluation function. I am using. In addition, a penalty is applied when two or more files of the same divided file group are contained in one volume group. Specifically, if the distributed file is in two or more for one volume, setting the penalty D _i as the Di> 0.

Next, this system uses the above quantities IO _i , S
Genes are evaluated using an evaluation function based on _i , N _i and penalties P _i , D _i . The evaluation function is (4)-
It is expressed by equation (11).

[0042]

[Equation 2]

Among the evaluation functions, equations (4) to (7) are for calculating the sum total of the quantities IO _i , S _i , N _i and the penalty P _i in all the volumes. Expression (8) calculates the number of volumes whose quantity IO _i is not within a certain range, and expression (9) calculates the number of volumes whose quantity S _i is not within a certain range. Α _IO and β _IO in the formula (8), and α _SIZE and β _SIZE in the formula (9) are values arbitrarily determined according to the operation status of the system. Further, D _{i in the} equation (10) is a penalty for files in the divided file group.

Then, the values E _{1 to} E ₇ obtained by the equations (4) to (10) are respectively multiplied by the weighting factors ε _{1 to} ε _{7 in the} equation (11), and the sum E thereof is calculated. The value of E is used as the evaluation scale. The evaluation function used in this system shows the purpose to be achieved in file relocation, that is, the suitability for leveling the IO count, file capacity, and number of files in each volume. The smaller the value of E, the better the evaluation. high,
This is the optimal file layout. The weighting factors ε _{1 to} ε ₇ in the equation (11) may be arbitrarily determined according to the operational status of the system.

The present system performs the series of evaluation processes described above for each gene. When the evaluation process ends, the system confirms whether or not a certain standard is cleared (S4). The criterion in S4 is a criterion for deciding whether to end the operation of GA. For example, in the case where a gene having a level higher than a prescribed evaluation exists or has occurred, or a gene with the highest evaluation does not occur in a prescribed cycle, etc. , Assuming that there is no further gene evolution, terminate GA (S4 N
O), if not (YES in S4), continue GA. In this system, the GA is not terminated immediately after the first evaluation. Then, the system extracts a predetermined number of highly evaluated genes based on the obtained evaluation of each gene (S5). In S3, in the present system, as shown in FIG. 3, the generated gene group 31 is classified into an upper gene group 42, a lower gene group 43, and a crossing target gene group 34 based on the evaluation result.

Next, the present system uses the cross target gene group 3
Two genes are extracted from 4 and crossed to generate one new gene (S6). The system repeats the above intersection a specified number of times. After the predetermined number of crossovers, the system extracts a proper number of genes and inverts some of the genes to generate mutations (S
7). Gene crossovers and mutations in this system are performed as shown in FIG. Genes newly generated by crossing and mutation are included in the crossing target gene group 34, and thereby the crossing target gene group 34 is reconstructed. The site of the gene to be crossed and the site of the gene to be inverted by mutation may be arbitrarily determined.

After completing the crossover and mutation operations, this system evaluates the genes belonging to the crossover target gene group 34 (S3). As a result of the evaluation, when there is a gene having a higher evaluation than the genes in the upper gene group 32 (the highest evaluation gene 341 in FIG. 3), the present system is
As shown in, the lowest-evaluated gene among the upper genes 32 and the highest-evaluated gene 341 are exchanged. In this system, the upper defined gene data is stored in the upper gene group 32.
By performing an operation of crossing or mutation in the gene data of the crossover target gene group 34 and exchanging with the generated highest evaluation gene 341, stagnation of evolution caused by the dominant gene is prevented, and The generation of highly evaluated genes can be promoted.

The present system tries to generate a gene having a higher evaluation by repeating the above-described operations of S3 to S7 for the crossing target gene 34. And S
In 3, when the criteria such as the occurrence of the highest evaluation gene 341 of the prescribed evaluation has occurred or the highest evaluation gene 341 does not occur in the prescribed cycle (YES in S4), the present system ends GA. If the criterion is not cleared (NO in S4), the system returns to S5 and repeats crossing and mutation. After the GA is completed, the system can obtain the optimal solution for the file arrangement by extracting the highest gene from the upper genes 32. The above is the operation of the file optimal layout determination processing by the GA of the present system.

In this system, the mutation generated in S7 is the highest evaluated gene 3 in the evaluation in S3.
You may make it generate | occur | produce when 41 does not exist.
With such an operation, mutation can be appropriately generated, and the highest evaluation gene 341 can be efficiently generated.

Next, in the present system, the initial placement setting, which is preferably performed prior to the file optimum placement determination process, will be described with reference to FIGS. Figure 6
FIG. 7 is an explanatory diagram showing the relationship between the IO count number or file capacity and volume in a general host online system, FIG. 7 is an explanatory diagram of the IO count in this system, and FIG. 8 is a file in this system. It is an explanatory view of capacity.

In FIG. 6, FIG. 6A is a distribution chart showing the distribution of the IO count numbers in each volume, where the horizontal axis represents the volume number and the vertical axis represents the IO count number. FIG. 6B is a distribution diagram showing the distribution of the file capacity in each volume, where the horizontal axis represents the volume number and the vertical axis represents the file capacity. In each figure, it can be seen that the IO count number is high in a specific volume and the file capacity is also large in the specific volume. Further, comparing the two figures, the volume with a high frequency of IO counts and the volume with a large file capacity do not necessarily match, but rather have a negative correlation.

It is generally known that a significant portion of online system access is concentrated in a particular file. Therefore, in this file relocation system, first, the IO count numbers of all files are checked, and a list of files is created in the order of higher ranks. Then, each file is provisionally allocated to the volume in the order of higher ranks.

In FIG. 7, the present system will be described assuming that N files are arranged in m volumes. First, this system uses IO for N files.
The number of counts is checked, and a list of file numbers is created in order from the top based on the IO count.

Here, the present system obtains the file number n that is the IO count average value. In this system, n
Is used to classify the upper n files of the N files into the upper count number file group and the remaining n + 1 to N-th remaining files as the lower count number file group.

The method of calculating the value of n will be described in detail. When the IO count number of the upper i-th file is represented as IO _i , the IO count number IO _n of the n-th file
The average IO count of the lower count files is
It is necessary for the relationship to be expressed by equation (12).

[0056]

[Equation 3]

Based on equation (12), this system is
IO count of the eye file IO _nAnd the lower count
Minimal difference from the average IO count of several files
Determine n as such. When expressed by a mathematical formula, as in formula (13)
become.

[0058]

[Equation 4]

When the value of n is determined, the system provisionally allocates each file to the volume. As the allocation method, the file with the largest IO count number is assigned to volume 1, the second file is assigned to volume 2, and so on. The illustration of FIG. 6 shows the volume and I after allocation.
It shows the relationship of the O count number, the horizontal axis shows the volume number, and the vertical axis shows the IO count number.

The process after each file is allocated to the volume will be specifically described with reference to FIG. Figure 8
Is a diagram showing the relationship between the volume and the file capacity of the upper count number file group after file allocation, where the horizontal axis represents the volume number and the vertical axis represents the file capacity.

In a general host online system, the difference between the file capacities of the volumes is
It is not as large as the difference between the IO count numbers. Further, since the IO count number and the file capacity do not have a positive correlation, even if the IO count number is high in each volume, the file capacity is not always high. Therefore, the distribution of the file capacity in the upper count number file group has a waveform with no regularity as shown in FIG.

In the preferred system according to the present invention, GA is performed after the allocation of the upper count number file group. Specifically, GA is performed using the IO count excluding allocated files and the total file capacity in each volume for files that have not been allocated yet. Here, the penalty is applied only to the n + 1 to m-th volumes. Thereby, the rearrangement can be performed more efficiently.

Next, the operation of the file relocation processing in this system will be described with reference to FIG. The resource information collection system 3 monitors the access status to the file group 2 in the host online system 1, collects the monitor status as an access record, and outputs it to the file relocation system 4.

In the file relocation system 4, the access record is stored and stored in the resource use status storage unit 11. The access record is recorded in the data processing means 12 by
It is processed into information on the IO count number and file capacity of each file and the file allocation state in each volume, and is output to the processed data storage unit 13. Then, the file optimum placement determination processing means 15 accesses the processed data storage unit 13, reads the stored information, calculates the IO count average value and the file capacity average value, and performs the initial placement setting of the file. , GA to determine the optimum file layout. Further, the file optimum layout determination processing means 15 accesses the control information storage unit 14 to read a necessary processing command, and executes the optimum layout determination processing. When the optimum placement determination processing is completed, the file optimum placement determination processing means 15 outputs the optimum placement information of the file, which is the processing result, to the file placement data storage unit 16.

The file move JCL creation processing means 17
When the optimal placement information of a new file is input to the file placement data storage unit 16, a processing instruction for placing a file according to the optimal placement information (hereinafter, file placement processing instruction) is generated using JCL, Output to the file move JCL storage unit 18. File move processing means 1
When a new file placement processing command is input to the file migration processing JCL storage unit 9, the server 9 relocates each file in the file group 2 to the corresponding volume based on the processing command.

The file rearrangement process in this system may be automatically executed periodically by using a batch process, or may be executed by an execution instruction from the user. In addition, in this system,
For example, if there are many volumes, divide the volume into multiple volume groups, relocate each file to the volume group using this system, and then use this system in each volume group. The files relocated to the volume group may be relocated to the volumes belonging to the volume group.

The rearrangement of files by this system is preferably performed when the access status of a file changes, such as when a disk is replaced or when a new application is installed.

As described above, according to this system, when rearranging each file in the volume, the IO count number and file capacity of each file and whether each file is actually arranged in each volume or not. Relocation of files satisfying various purposes at the same time by performing the optimal placement determination process for obtaining the optimal placement of files by GA using the placement status information indicating that, and relocating the files to the volume based on the processing result. There is an effect that can be efficiently performed in a short time.

In particular, in this system, the GA creates a random gene in which the arrangement state of all files in the volume is expressed by a binary number, and the gene evaluation is performed using an evaluation function showing the suitability of the file arrangement for the volume. Then, the ones with low evaluation among the genes are removed based on the evaluation result, the groups of the remaining genes are extracted, crossover is performed with respect to any two genes in the group, and / or with respect to predetermined gene data. Generating a new gene to generate a new gene to reconstruct the group, repeat the crossover, mutation and evaluation of the genes in the group until the specified condition is satisfied in the gene evaluation, and after the GA is completed, By outputting the gene with the highest evaluation as the optimal layout of the file, the gene algorithm generated by the genetic algorithm is filtered. Since it is possible to evaluate the suitability of relocation,
Files that satisfy various purposes at the same time can be relocated in a short time and efficiently.

Further, in this system, in the GA, the gene is evaluated by using the evaluation function showing the suitability of the IO count number, file capacity and the file number in each volume, so that the IO count number and the file in each volume are evaluated. There is an effect that it is possible to efficiently find the optimum arrangement of files that can equalize the capacity and the number of files under a certain condition in a short time.

Further, in the present system, among the genes in the extracted group, the upper defined number of genes are extracted as the upper gene group 32, the remaining genes are classified as the crossover target gene group 34, and the crossover or mutation is performed. As a result of the gene evaluation of the genes of the crossover target gene group 34 after performing the reconstruction, when there is a gene having a higher evaluation than the gene of the upper gene group 32, the gene having the lowest evaluation is exchanged with the upper gene group 32. As a result, it is possible to promote the generation of highly evaluated genes, and to rearrange files that simultaneously satisfy various purposes at the same time in an even shorter time.

In addition, in the present system, when a gene having the highest evaluation in the crossover target gene group 34 is not generated for a prescribed period, it is possible to appropriately generate a mutation by generating a mutation. There is an effect that the files to be simultaneously filled can be efficiently rearranged in a shorter time.

Further, by relocating the files in the host online system using this system, the relocation of the files satisfying various purposes at the same time can be performed efficiently in a short time. As a result, the access time to the file can be reduced, which has the effect of improving the performance of the host online system.

[0074]

According to the present invention, in the file relocation system, it is possible to efficiently relocate files that simultaneously satisfy various purposes in a short time.

[Brief description of drawings]

FIG. 1 is a block diagram of an overall configuration including a file relocation system according to an exemplary embodiment of the present invention.

FIG. 2 is an explanatory diagram of a gene generation method in GA used in the file rearrangement system according to the exemplary embodiment of the present invention.

FIG. 3 is an explanatory diagram showing gene extraction and classification states in the file rearrangement system according to the exemplary embodiment of the present invention.

FIG. 4 is an explanatory diagram of crossover and mutation in GA.

FIG. 5 is a flowchart diagram of a file optimal placement determination process by a GA in the file reallocation system according to the embodiment of the present invention.

FIG. 6 is an explanatory diagram showing the relationship between the IO count number, file capacity, and volume in the host online system.

FIG. 7 is a diagram of IO count number versus volume according to one embodiment of the invention.

FIG. 8 is a diagram of file capacity versus volume according to one embodiment of the invention.

[Explanation of symbols]

1 ... Host online system, 2 ... File group,
3 ... Resource information collection system, 4 ... File relocation system, 11 ... Resource usage status storage unit, 12 ... Data processing unit, 13 ... Processing data storage unit, 14 ... Control storage unit, 15 ... Optimal file placement determination processing unit , 16 ... File arrangement data storage unit, 17 ... File move JCL creation processing unit, 18 ... File move processing JCL storage unit, 19 ... File move processing unit, 3.
1 ... Gene group, 32 ... Upper gene group, 33 ... Lower gene group, 34 ... Cross target gene group, 341 ... Highest evaluation gene

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Tatsuo Yoshikawa             Yamato Co., Ltd. 15-6 Fuyuki, Koto-ku, Tokyo             Soken F-term (reference) 5B082 AA01 CA05 CA11

Claims (11)

[Claims] 1. When relocating a file placed in a plurality of volumes in a computer to a plurality of volumes, the input / output load and file capacity of each file and whether each file is actually placed in each volume. A genetic algorithm is used to calculate the optimal placement determination process that determines the optimal or nearly optimal volume for each file by using the placement status information that indicates whether or not each file is matched based on the processing result. File relocation system characterized by relocating to a volume to be used. 2. A file relocation system for relocating files arranged in a plurality of volumes in a computer to a plurality of volumes, the input / output load and the file capacity of each file from the resource usage status in a general-purpose computer, Using the data generated by the data processing unit that generates and outputs the arrangement state information indicating whether or not each file is actually arranged for each volume, using the data generated by the data processing unit,
A file optimal placement unit that performs an operation by a genetic algorithm and performs an optimal placement determination process that determines a volume that is an optimal or substantially optimal placement for each file, and a file optimal placement unit that assigns each file based on the processing result of the optimal placement determination process. A file relocation system having a file migration processing unit for relocating to a corresponding volume. 3. The genetic algorithm comprises a gene generating means for generating a plurality of random gene data in which the arrangement state of each file in a volume is expressed in binary notation, and the file rearrangement determines the suitability of the file arrangement for the volume. Gene evaluation means for evaluating gene data using the evaluation function shown, and a group extraction means for extracting a group of the remaining gene data except for a prescribed number of low-evaluation gene data based on the evaluation result by the evaluation function, , Extracting any two gene data from the group,
Intersection means for exchanging a part of the extracted gene data to generate new gene data, and reconstructing a group with the generated gene data, and extracting arbitrary gene data from the group, and extracting the gene data. Generating new gene data by changing a part of the data, and having a mutation means for reconstructing a group with the generated gene data, the gene data generated by the gene generating means, The groups evaluated by the evaluation means, the groups extracted by the group extraction means, the groups reconstructed by the intersecting means and the mutation means, and the groups reconstructed by the intersection means and the mutation means The gene data in the medium is evaluated by the gene evaluation means, and the crossing operation is performed until the specified condition is satisfied by the gene evaluation means. An algorithm for repeatedly performing the processing in the mutation means and the gene evaluation means, wherein the file optimal placement unit is a file migration processing unit for the gene data having the highest or almost the highest evaluation result after the completion of the calculation by the genetic algorithm. 3. The file rearrangement system according to claim 2, wherein the file is rearranged. 4. The evaluation function is a value obtained by multiplying a sum of ratios of input / output counts in each volume with respect to an average value of input / output counts per volume by a predetermined weighting coefficient, and a value in each volume with respect to an average value of file capacity per volume. The value obtained by multiplying the total sum of the file capacity ratios by a predetermined weighting factor and the sum of the ratio of the number of files in each volume to the average number of files per volume by a predetermined weighting factor are used. The file rearrangement system according to claim 3, wherein the gene data is evaluated by adding the values. 5. The evaluation function uses a value obtained by multiplying the total sum of penalties determined according to the number of files in each volume by a predetermined weighting coefficient, and evaluates gene data by adding this value. 5. The method according to claim 4, wherein
File relocation system described. 6. The evaluation function is characterized in that a penalty is set when a plurality of distributed files formed by division for avoiding I / O contention and / or exclusive control of files are in the same volume. The file relocation system according to claim 5. 7. The evaluation function uses a value obtained by multiplying the number of volumes whose input / output count is not within a certain range by a predetermined weighting coefficient, and evaluates gene data by adding this value. The file relocation system according to any one of claims 4 to 6. 8. The evaluation function uses a value obtained by multiplying the number of volumes whose file capacity is not within a certain range by a predetermined weighting coefficient. 8. The file rearrangement system according to claim 4, wherein the gene data is evaluated by adding this value. 9. The file optimal arranging unit extracts, by the group extracting unit, a prescribed number of gene data out of the gene data belonging to the group, and the group reconstructing unit partially exchanges the gene data not extracted. Or, when some changes are made and the gene evaluation means has gene data with a higher evaluation than the extracted gene data, the highest evaluation of the gene data with the highest evaluation and the extracted gene data 4. Exchange with low genetic data.
8. The file relocation system according to any one of 8 to 8. 10. The file optimal arrangement unit makes a partial change in the subsequent group reconstruction when the gene evaluation unit does not generate the highest-evaluated gene data for a prescribed period. 9. The file relocation system according to item 9. 11. A host online system, wherein the file relocation system according to claim 1 is used to relocate a file.