JPH07200389A - Data management method - Google Patents

Abstract

PURPOSE:To enhance the system capability easily and sufficiently by deciding new data allocation in a data storage area based on data storage information and correlation information representing a data storage area. CONSTITUTION:A correlation information management section 3 generates correlation information from access information received newly and collates it with correlation information stored in a correlation information storage table 2 and updates the correlation information when the information if correlation information already registered and registers the correlation information on the correlation information storage table when the information is correlation information not registered yet. In this case, the information not updated for a prescribed period or over or not satisfying a predetermined condition is deleted from the correlation storage table 2. The correlation information management section 3 repeats update of the correlation information every time an access request comes and gives the correlation information to a data allocation decision section 4 when sufficient correlation information is obtained. The data allocation decision section 4 decides new data allocation based on the received correlation information and data storage information stored in a data storage information storage section 5.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data management method for a data management device having a plurality of areas for storing data.

[0002]

2. Description of the Related Art On a computer system using a data management device having a plurality of areas for storing data and storing and managing the data in a distributed manner in the plurality of areas, a plurality of processing executed for each processing request is performed. Data access accesses data in a plurality of data storage areas. in this case,
If the data required to execute each processing request is in the same data storage area, the time required to execute the processing request will be shorter. However, when multiple data required for execution are not in the same data storage area, even if they are stored in different data storage areas in the same storage device, access to them is in the same data storage area. It takes extra time to read and write compared to when there is. Further, when a plurality of data are in the data storage areas on different storage devices, it is necessary to request the reading from another data management device that manages the storage device or transfer from the storage device. The access time is further increased, and the processing request execution time is lengthened.

On the other hand, when a plurality of data storage areas can be accessed at the same time, it is better to equalize the access frequency to each data storage area as much as possible, so that the processing request execution capability is improved as a whole system. Therefore, in order to shorten the execution time of each processing request and to fully exhibit the processing request execution performance of the system, the data required by each processing request should be placed in the same data storage area as much as possible and It is desirable to arrange the data so that the access frequency to is equalized.

Conventionally, the data arrangement in the storage area has been performed by the system administrator explicitly designating the data arrangement. That is, the system administrator analyzes the data accessed by each processing request, the frequency of occurrence of each processing request, etc. to determine whether or not the system is currently exhibiting sufficient performance.
Furthermore, if the performance is not fully exerted, the performance of the system should be determined by analyzing what is the cause of the data arrangement and allocating each data to which data storage area based on the analysis. He decided on his own whether or not he could make the most of it. Then, based on the new data arrangement, the data is arranged by explicitly designating the correspondence between each data and the data storage area in which it should be stored in the data management device.

On the other hand, when the correspondence between the data and the area for storing the data is not given, the system appropriately arranges each data in an arbitrary storage device.

On the other hand, as a method of arranging data without performing such analysis, there is a method of arranging data according to a predetermined rule. As these methods, as described in Database Benchmarking (edited by Jim Gray, Yu Kitsuregawa, Eiichi Watanabe, Nikkei BP), data is randomly arranged in the data storage area using a hash function, etc. Are arranged in a round robin manner with respect to the data storage area, and the range of data values is divided into the number of data storage areas and assigned one by one to the data storage area. In addition, a method has also been proposed in which data is divided into a predetermined size for each value range of data, and the data is allocated to a data storage area by a round robin method. (Shahram
Ghandeharizadeh, David J. DeWitt, Proc. 16th.VLDB Con
f. 1990) Here, the execution time of the processing request executed on the computer system differs depending on the data arrangement in the system. Therefore, in order to make full use of the performance of the system, it is necessary to place the data required by the processing requests executed by the system in the same data storage area as much as possible, and when using multiple storage devices, It was necessary to minimize the load difference between the two.

Conventionally, in order to make such an arrangement, the system administrator has to access the data accessed by each processing request operating on the system, the frequency of occurrence of each processing request, and the load on each storage device. After measuring and analyzing, it was necessary to determine the data arrangement.

However, in order to analyze what data each processing request accesses, it is necessary to analyze, for example, the source code. However, the source code is analyzed for all the processing requests executed on the computer system. It was very difficult to analyze. In addition, even if the source code is analyzed, not all data access destinations can be identified, and a branch occurs or a data access destination is determined due to a factor that is dynamically determined when a processing request is executed. Cannot determine the data accessed by the processing request. Therefore, even if a new data arrangement that satisfies the above requirements is required based on the result of analyzing the source code, the system performance cannot be sufficiently exerted with that arrangement.

Further, it is necessary to re-execute the analysis each time a new processing request is added to the system or the frequency of occurrence of processing requests executed on the system changes. Therefore, it takes a lot of effort to make the data arrangement match the characteristics of the system. Also, after arranging the data appropriately without such analysis, measure the load of each data and the data storage device, and rearrange the data so that the load difference between the data storage devices is eliminated as much as possible. However, there is a problem in that the performance of the system cannot be exhibited because it does not take into consideration which data and which data requires a processing request to be executed on the system.

On the other hand, there is also devised a method of arranging data so that the difference in loads does not appear in advance, instead of rearranging the data as described above so as to eliminate the difference in loads. As a method of such data arrangement, a method of dividing a data value into a predetermined range and allocating the range to each storage area, a method of randomly allocating to each storage area based on the value of the data , A method of allocating data to each data storage area in a round-robin manner, a method of dividing the data value into a predetermined range and allocating the divided ones to each data storage area in a round-robin manner, etc. Has been taken. All of these are intended to stochastically avoid concentration of load on a specific data storage device by such an arrangement.
If such an arrangement is used, there may be cases where the load is not concentrated, but these systems have different system configurations,
It does not guarantee that the load will not be concentrated on the frequency of processing requests. Also, when the load is concentrated, of course, even if the load is not concentrated, the data required by the processing request executed on the system is randomly arranged, so the execution time of each processing request is increased. However, there was a problem that the system performance could not be fully exhibited.

As described above, in the conventional data management method, it is very difficult to perform an analysis to make full use of the performance of the system, and furthermore, the analysis is not perfect, so that the performance of the system is sufficiently made. I couldn't let it go.

[0012]

As described above, according to the conventional data management method, it is very difficult to analyze the access request and the data arrangement for making full use of the system performance. Since the analysis was not complete, the system performance could not be fully demonstrated.

The present invention has been made in consideration of the above problems, and automatically creates and analyzes the information necessary for determining the data arrangement, thereby making it possible to easily and sufficiently bring out the capabilities of the system. It is an object of the present invention to provide a data management method capable of controlling.

[0014]

In order to solve the above problems, in a data management method according to the present invention, an access request receiving step of receiving an access request for data stored in a plurality of different data storage areas, and specifying Shows a correlation information creating step of creating correlation information of mutual data indicating that a plurality of data have been accessed together in a series of processing based on the access request of a predetermined number of times, and a data storage area in which the data is stored. A data arrangement determining step of determining a new data arrangement in the data storage area from the data storage information and the correlation information.

Preferably, in the correlation information creating step, the history of access requests is divided into access request sequences based on a predetermined rule, and the correlation information is created based on the data access executed in the access request sequence. It is characterized by In this case, further, in the correlation information creating step, at least one of whether to retain the received access request as a history or whether to use the divided access request sequence when creating the correlation information is determined. May be.

Preferably, the data stored in the data storage area is managed by being divided into a plurality of blocks, and in the correlation information creating step, the correlation information is acquired for each block. To do.

Preferably, the data management method according to the present invention further comprises a rearrangement step of rearranging data according to the determined new data arrangement. In this case, in the relocation step, the loads on the data storage devices are examined, the loads on the respective storage devices are compared,
The arrangement of data in a storage device with a light load may be changed preferentially, or the stored data may be copied to a new storage device that does not store data while changing the arrangement. You may do it.

Further preferably, in the data arrangement determining step, a significant one is selected from the correlation information, and the data arrangement is changed based on the selected correlation information.

Preferably, in the correlation information creating step, when the correlation information is updated, the information that has not been updated for a predetermined period or more is deleted.

Further, preferably, in the correlation information creating step, the significance of the correlation information is discriminated, and those which are discriminated to be insignificant when updating the correlation information are deleted.

Preferably, in the data arrangement determining step, the degree of association between the data is measured, and when there are a plurality of candidates for the data rearrangement destination based on the correlation information, a lot of data having a high degree of association are stored. The storage device in use is selected as the relocation destination.

Further, preferably, in the data arrangement determining step, when there are a plurality of candidates for the data rearrangement destination based on the correlation information, the plurality of candidates are presented to the administrator to specify the rearrangement destination. It is characterized by

On the other hand, the data management apparatus according to the present invention comprises means for receiving an access request for data stored in the data storage means comprising a plurality of data storage areas,
Means for creating correlation information between data indicating whether the stored data is accessed together in a series of processing based on an access request; data storage information indicating in which data area the data is stored; And means for determining a new data arrangement in the data storage area from the correlation information.

[0024]

According to the present invention, when a processing request that requires access to data stored in a distributed manner in a plurality of different data storage areas is transmitted from a computer or the like, it is first received and a specific plurality of data is stored. The mutual correlation information of the data indicating that the data are accessed together in a series of processing is updated.

This operation is repeated to obtain sufficient information about the correlation information in the system. Then, based on the correlation information and the data storage information indicating the data storage area in which the data is stored, the characteristics of the system can be determined. In addition, a new data arrangement in the data storage area is determined.

Therefore, according to the data management method of the present invention, when the data stored in the plurality of data storage areas are frequently referenced / updated in a series of processes, the correlation is automatically determined. The data arrangement can be automatically changed so that such data is stored in the same data storage area.

Further, since the system administrator does not need to analyze the processing on the system to determine the data allocation because the information required for the data allocation is automatically acquired, the execution time of each process is shortened. It will be possible to easily bring out the performance of the.

[0028]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows a schematic configuration of a data management device according to an embodiment of the present invention. As shown in the figure, the data management device 10 includes an access request reception unit 1, a correlation information storage table 2, a correlation information management unit 3, a data allocation determination unit 4, a data storage information storage unit 5, a data allocation changing unit 6, and The data access unit 7 is provided.

The data management apparatus 10 receives a data access request from a processing request executed on a computer group 9 composed of one or more computers, and the history of access requests satisfies a predetermined condition. Create and update correlation information when After acquiring a sufficient amount of correlation information in this way, new data arrangement information is created based on this and the data storage information.

The data management device 10 also stores data on one or more storage devices. Here, the data management device 10 manages a plurality of data storage areas, but it is also possible to allocate them to each storage device one by one as shown in FIG. 6 or by configuring each storage device as shown in FIG. It is also possible to allocate a plurality to each of the above. See also FIG.
It is also possible to collectively handle a plurality of storage devices as one data storage area as shown in FIG. The number of data storage areas in the storage device and the size of each area are arbitrary, and the number of data storage areas between the storage devices and the size of the areas may be different from each other.

Storage device 8 managed by the data management device 10
Has a function of storing data, as long as it can read or rewrite data at an arbitrary position in the storage device 8, a semiconductor memory, a hard disk, a floppy disk, It is possible to use a magnetic tape or the like.

The data management device 10 can be configured as an independent device, or can be realized as software or hardware as a function of the computer.

The operation of the data management device 10 shown in FIG. 1 will be described below.

The data access request generated from the processing request executed in the computer group 9 including one or more computers is received by the access request receiving unit 1. The access request is made up of data necessary for data access and data necessary for generating correlation information. The access request reception unit 1 passes the received data access request to the correlation information management unit 3 and the data access unit 7.

The data access unit 7 reads or updates data from the storage device 8 based on the received access request, and returns the result to the data access request source (any computer of 9).

The correlation information management unit 3 creates the correlation information from the newly received access request, compares it with the correlation information stored in the correlation information storage table 2, and if the correlation information is already registered, the correlation information is stored. Information is updated, and if it is unregistered correlation information, the correlation information is registered in the correlation information storage table 2. At that time, those that have not been updated for a certain period of time or those that do not meet the predetermined conditions are deleted from the correlation information storage table 2. Correlation information management unit 3
Repeats such updating of the correlation information each time there is an access request, and passes the correlation information to the data arrangement determination unit 4 when the correlation information is sufficiently obtained.

The data arrangement determination unit 4 determines a new data arrangement based on the received correlation information and the data storage information stored in the data storage information storage unit 5. After deciding the new data arrangement, the data arrangement determining unit 4 transfers it to the data arrangement changing unit 6 and the data storage information storage unit 5.

When changing the arrangement, the data arrangement changing unit 5 first notifies the data access unit 7 of which data location and how. Upon receiving this notification, the data access unit 7 temporarily prohibits update requests for those data. Next, the data arrangement changing unit 6
Replaces the data based on the received data arrangement,
When the replacement is completed, the fact is notified to the data storage information storage unit 5 and the data access unit 7.

When the data storage information storage unit 5 receives from the data layout update unit 6 that the data replacement is completed, the data storage information storage unit 5 updates the data storage information based on the data layout information received from the data layout determination unit 4 in advance. . Further, when the data access unit 7 receives the notification that the data exchange is completed, the data access unit 7 changes the management information for the data access due to the change of the data arrangement as necessary, and then, Enables update requests.

Next, another configuration example of the data management apparatus of the present invention is shown in FIG. The data management apparatus of FIG. 2 basically has the same components as the data management apparatus of FIG. 1, but since the functions and connection relationships of some components are different, emphasis is placed on characteristic points. The description will be given below.

The data access request generated from the processing request is received by the access request receiving unit 1. The access request reception unit 1 passes the received data access request to the correlation information management unit 3 and the data access unit 7.

The data access unit 7 is the same as the embodiment shown in FIG. 1 up to the point of reading or updating data based on the received access request and returning the result to the source of the data access request. However, the difference is that the correlation information management unit 3 uses both the newly received access request and the data storage information stored in the data storage information storage table 2 when creating the correlation information. Other configurations are the same as those in FIG.

Next, another example of the configuration of the data management apparatus of the present invention is shown in FIG. The data management device of FIG. 3 basically has the same components as the data management device of FIG. 1, but since the functions and connection relationships of some components are different, emphasis is placed on characteristic points. The description will be given below.

The data access request generated from the processing request is received by the access request receiving unit 1. The access request reception unit 1 passes the received data access request to the correlation information management unit 3 and the data access unit 7. In the configuration shown in FIGS. 1 and 2, the data access unit 7 independently manages information about where in the storage device 8 the data is when the data is accessed, and the data is stored using the data. The position of the data in the device 8 was checked and accessed. On the other hand, this configuration is different in that the data access unit 7 does not have information on where in the storage device 8 the data is.

That is, when the data access unit 7 receives the access request, it searches the data storage information in the data storage information storage unit 5 for where the data is stored, then accesses the data, and returns the result as the data. Return to the source of the access request. The correlation information management unit 3 creates the correlation information from the newly received access request without using the data storage information, compares it with the correlation information storage table 2, and updates the correlation information if it is already registered. If not registered, it is registered in the correlation information storage table 2. At that time, those that have not been updated for a certain period of time or those that do not meet the predetermined conditions are deleted from the correlation information storage table 2. The correlation information management unit 3 updates the correlation information in this way, and passes the correlation information to the data arrangement determination unit 4 when it is determined that the correlation information is sufficiently obtained. Data placement determination unit 4
Determines a new data arrangement based on the received correlation information and the data storage information stored in the data storage information storage unit 5. The data arrangement determining unit 4 is the same as that in FIG. 1 up to the point of passing the new data arrangement to the data arrangement changing unit 6 and the data storage information storage unit 5 when the new data arrangement is determined. However, in this configuration, when changing the data arrangement, the data arrangement changing unit 6 first notifies the data storage information storage unit 5 of which data location and how. Upon receiving this notification, the data management information storage unit 5
Remembers that the data is currently inaccessible. Next, the data arrangement updating unit 6
Replaces the data based on the received data arrangement,
When the replacement is completed, the fact is notified to the data storage information storage unit 5. When the data storage information storage unit 5 receives from the data arrangement change unit that the replacement of the data has been completed, the data storage information storage unit 5 stores that the data is in the accessible state again, and updates the data storage information for them.

Further, as another configuration of the data management device, FIG. 4 which is a combination of the configuration of FIG. 2 and the configuration of FIG.
A configuration such as is also possible.

Further, in the configuration shown in FIGS. 1 to 4,
The data arrangement determination unit 4 and the data access unit 7 are different in that they share data storage information,
Each is configured to be able to access the data independently. However, as shown in FIG. 5, when the data arrangement determination unit 4 reads out or updates data, it is possible to always pass the data access unit 7, and the configuration shown in FIGS. A configuration in which the configuration of is modified in this way is also possible.

Here, in the above-described embodiment, access to the data is temporarily prohibited when updating the arrangement of the data. However, by creating a copy of the data when changing the layout of the data, making it accessible even when the layout is changed, and by configuring the layout to reflect the updated amount after the layout is changed, It is also possible not to prohibit access.

Further, as shown in FIG. 9, the data arrangement changing unit 6 is not provided, and after the data arrangement deciding unit 4 decides the data arrangement, the data arrangement presenting unit 16 for showing the arrangement to the system administrator. May be provided so that the data management device 10 itself does not change the data arrangement.

It should be noted that the data management device shown in FIGS. 4 and 5, the data management device obtained by modifying the configuration of FIGS. 1 to 4 as shown in FIG. 5, and the configuration and operation of the data management device shown in FIG. The details are self-explanatory, so
The description here is omitted.

Here, when using a plurality of storage devices for storing data, it is possible to provide a data management device for each storage device as shown in FIG.
As shown in FIG. 11, it is also possible to provide a data management device for every plurality. Alternatively, as shown in FIG. 12, it is also possible to configure a single data management device. Further, when there are a plurality of computers that manage the storage device, it is possible to provide a data management device for each computer as shown in FIG. 13, and it is also possible to provide a data management device for each computer as shown in FIG. It is also possible to provide a management device.

Next, FIG. 15 shows an example of the configuration of the correlation information management section 3 inside the data management apparatus 10. Hereinafter, each component of FIG. 15 will be described. As shown in the figure, the correlation information management unit 3 of this embodiment includes a correlation information generation unit 31, a series of process information storage table 32, an access history storage table 33, a correlation information table management unit 34, and a significant information selection unit 36. I have it.

First, the access request sent from the data access request receiving unit 1 is received by the correlation information generating unit 31. The access request consists of data necessary for data access and data necessary for generating correlation information. For example, an identifier of a computer, a process id of a processing request, an identifier for distinguishing processing within a process, and data are stored. Storage device identifier, logical or physical address indicating the position of data in the storage device, data identifier uniquely assigned to data, identifier of library for data access, access time, other data access on the program It consists of parameters for identifying and accessing the data.

The operation of the correlation information generator 31 will be described with reference to FIG.

Upon receipt of the access request (step S1), the correlation information generator 31 determines which process is registered in the series process information storage table 32 (step S2). Do with a series of processes have what kind of data access request string advance in series processing information storage table 32 is registered.

The correlation information generating unit 31 searches the access history storage table 33 based on the result of the next identification to see if a history of the series of processes remains (step S3).
The access history storage table 33 stores a history of data identifiers accessed by a series of processes in the past. If the corresponding entry does not exist in the table, the correlation information generation unit 3
1 secures a new entry and records the identifier (steps S4 to S7). If the entry exists, new access is added to the entry (step S8).

Next, the correlation information generator 31 determines whether or not the series of processing is completed by the current access (step S8). To determine whether it is a series of processing,
For example, the access sequence executed by a process having the same process id on the same computer can be determined as a series of processes by using the computer identifier in the access request and the process id that executed the process request. When it is determined that the series of processing is still continued, the next input is awaited in step S1, but when it is determined that the series of processing is finished, the access history information regarding the series of processing is displayed from the access history table. It is read out, correlation information is generated from it (step S10), and this is passed to the correlation information table management unit 34 (step S11).

Here, the series of processes means processes performed from a predetermined start time to a predetermined end time. For example, it is possible to treat a process as a series of processes from its creation to its termination, or when that process performs different processes according to input, each process can be defined as a series of processes. I can. Alternatively, it is also possible to divide the operation of the process into a plurality of parts and define them as a series of processes. Furthermore, it is possible to define a process executed as a transaction as a series of processes. Further, it is possible to define the same except for specific processing. Alternatively, it is possible to make all the processes executed between a predetermined start time and end time a series of processes.

The correlation information generator 31 can create the correlation information only from the history of access requests relating to a series of processes. Since a series of processing accesses a plurality of data, the access history is composed of a plurality of access requests, in other words, a plurality of accessed data identifiers. For example, in a certain process, each identifier is a, b, c
Assuming that the three pieces of data are accessed, the correlation information generated by the correlation information generating unit 31 can be, for example, the following set of identifiers and the number of times of access to the set.

(A, b, 1) (b, c, 1) (a, c, 1) That is, from the access history consisting of a set of n data identifiers, a combination of nC2 identifiers and their respective combinations Correlation information can be obtained based on the number of accesses to.
Alternatively, a set of identifiers of data, which is an access history in a series of processes, and the number of accesses, {a, b,
It is also possible to directly use c, 1} as the correlation information.

The data identifier is a physical or logical address of data, or a tuple id in a relational database, a record id in an index sequential file, or the like assigned to data to logically distinguish the data. But good.

In the above example, the correlation information is generated only from the access history, whereas the correlation information generation unit 31 creates the correlation information from the access history and the data storage information obtained from the data storage information storage unit 5. Is also possible. In this case, the storage area storing the data is divided into blocks consisting of a plurality of data, a unique block number is assigned to each block, and the block number is used as an identifier for the data contained in the block. . As a block, it is possible to handle a block that is divided by the range of its value, a tuple in a relational database, a record in an index sequential file, etc. that are handled collectively during data management as a block, or A logical area that has nothing to do with the content of data such as a page can be treated as a block. For example, it is assumed that a certain series of processing accesses data having identifiers a, b, and c, and the block numbers in which the respective data are stored are 10, 5, and 8, respectively. As the correlation information generated at this time, the following set of (block number, block number, access count) can be used.

(10,5,1), (10,8,1), (5,8,1) In this way, by converting the data identifier into a number assigned to a block composed of a plurality of data, , The number of identifiers that can be generated is small. Therefore, since the number of possible combinations is reduced, the management by the correlation information table management unit 34 becomes easy.

The access history storage table 33 uses, for example, a tree structure in which an identification number assigned to a series of processes is used as a key, and an identifier of data accessed by the linked list structure can be stored in the leaf portion thereof. Is. Further, the access history storage table 33 can be configured by using, for example, a hash table and a linked list structure instead of the tree structure. Further, in order to store the data accessed by each process, it is possible to use a tree structure using an address as a key, in addition to the linked list.

The correlation information table management unit 34 receives the correlation information generated by the correlation information generation unit 31. FIG. 17 shows the operation of the correlation information table management unit 34.

When the correlation information table management unit 34 receives the correlation information from the correlation information generation unit 31 (step S21), the correlation information table management unit 34 checks whether or not it is already stored in the correlation information storage table 2 (step S21). Steps S22 to S2
3). If any correlation information is not stored in the correlation information storage table 2, it is registered in the correlation information storage table 2 (steps S24 to S27). If already stored, the correlation information in the correlation information storage table 2 is updated (step S28).

After performing the above processing for each element of the correlation information received from the correlation information generation unit 31, the correlation information table management unit 34 determines whether the correlation information has been acquired enough to create a new arrangement. Is determined (step S2
9). If it is determined that the correlation information is sufficient, the significant information selection unit 36 is notified that sufficient correlation information has been acquired, and the current contents of the correlation information storage table 2 are copied to another storage area. Alternatively, the contents of the correlation information table storage 2 at that time are transmitted to the significant information selection unit 36. If it is determined that it has not been acquired sufficiently, nothing is done.

Here, whether or not the correlation information is sufficiently acquired can be determined as follows. For example, it can be determined that sufficient correlation information has been acquired when a predetermined time has elapsed since the start of acquiring correlation information. In this way, in a system that performs similar routine processing every day, it is possible to obtain information tailored to the time zone in which processing is concentrated. Alternatively, the number of times the correlation information is acquired may be set in advance, and it may be considered that the correlation information is sufficiently acquired when the number of times the correlation information is collected exceeds the value. It is also possible to configure the system administrator to prompt the system administrator to input the number of times in advance. Alternatively, regarding the correlation information, it is also possible to count significant ones according to a criterion used in a significant information selecting unit 36 described later, and to judge that the correlation information is sufficiently acquired when the number reaches a certain number or more. Furthermore, it is also possible to present the correlation information to the system administrator at any time and to have the system administrator instruct the system administrator to end the acquisition of the correlation information.

When the significant information selection unit 36 receives the notification from the correlation information table management unit 34 that the correlation information has been sufficiently acquired, the correlation information table is sent from the correlation information table when it is sent, and the correlation information table is sent when the correlation information table is not sent. The correlation information table copied by the information table management unit 34 is read out. After that, a significant one is selected from among them, and the selected information is passed to the data arrangement determining unit 4. Alternatively, all of the read results can be passed to the data arrangement determination unit 4 without selecting a significant one.

As a method of selecting a significant one, for example, a means for sorting in the order of access frequency in the correlation information is added, and a predetermined number is selected from the upper ones of the results sorted in the order of access frequency. Can be significant. Alternatively, it is possible to determine that the number of times of access that is equal to or higher than a predetermined ratio as compared with the highest number of access frequencies is significant. Alternatively, it is also possible to make only those that are accessing more than a predetermined ratio to the total number of accesses significant.
Alternatively, it is also possible to examine the distribution of the number of accesses and select based on the average value or standard deviation thereof.

In the configuration described above, the significant information selection unit 3
Even after 6 has selected significant information, it continues to acquire correlation information in addition to the previously acquired data. On the other hand, the correlation information table management unit 34 judges that sufficient information has been acquired and saves the correlation information table at that time, or the significant information selection unit 3
It is also possible to discard the data up to that point after transmitting the contents of the correlation information table to No. 6 and newly acquire the correlation information. With this configuration, the arrangement can be dynamically changed according to the change in processing request.

The correlation information storage table 2 is made up of a plurality of entries for storing the correlation information, and each entry stores a set of identifiers of accessed data and its occurrence frequency.
It is also possible to configure each entry to consist of a set of numbers uniquely assigned to a block and the frequency of occurrence instead of the data identifier as described above. As the occurrence frequency, the occurrence frequency number itself can be stored, or a value obtained by normalizing the occurrence frequency, or a function value of any function having the occurrence frequency as its argument can be substituted. I can.

When all the entries managed by the correlation information storage table 2 are already in use, the correlation information table management section 34 needs to secure a new free area. For this purpose, for example, an area for storing the last updated time can be added to all entries, and the entry updated at the oldest time can be removed from the correlation information storage table 2. Alternatively, it is possible to secure a free area by adding a means for sorting the entries of the correlation information storage table 2 according to the number of access frequencies thereof and removing the one having the smallest number of access frequencies as a result of sorting. Alternatively, each entry of the correlation information storage table 2 is classified into a significant one and an insignificant one on the basis of the access frequency, and the above processing is executed only between those not significant and the correlation information storage table 2 is executed. You can also select candidates to be deleted from. As a method of selecting a significant item, the same process as the selection method in the significant information selecting unit 36 described above can be executed.

Next, the operation of the data arrangement determining unit 4 will be described with reference to FIG.
Will be explained. When the correlation information is received from the significant information selection unit 36, the data arrangement determination unit 4 first determines in which storage device the data is stored from the data storage information storage unit 5, or how much capacity each storage device has. The data storage information such as After that, the current data arrangement is evaluated using a predetermined evaluation function from the data storage information and the correlation information. Then, assuming that it is the current data arrangement, the data arrangement and its evaluation value are stored (step S31). Next, the arrangement of the data is changed based on the correlation information (step S37), an evaluation value for the arrangement is obtained (step S32), and if it gives a better evaluation value than the current arrangement (step S32). S
33), the data arrangement and the evaluation value are recorded as the current one (step S34). This evaluation is continued, and if a data arrangement that satisfies a predetermined condition is obtained (steps S35 and S36), this cycle is stopped and the arrangement at that time is transmitted to the data arrangement changing unit 6 ( Step S38).

Here, the data arrangement determination unit 4 evaluates the data arrangement as follows. For example, N data, M
When there are data storage areas, NM variables Xij
(1 ≦ i ≦ N, 1 ≦ j ≦ M, Xij = 0 or 1) can be used to indicate that the i-th data is stored in the j-th data storage area. When the i-th data and the k-th data are often used together, it is desirable that these data exist in the same data storage area as much as possible. So, for example,

[0077]

[Equation 1] The data allocation can be evaluated by using the function. In this function, the evaluation value increases by Cik when the i-th data and the k-th data are stored in the same j-th data storage area. However, when this evaluation function is used, it is guaranteed that the data will be placed in the same data storage area, but it is not known in which data storage area the data will be stored. For example, if possible, the i-th data and j
If you do not want to move from the data storage area where the th data was, use the following evaluation function.

[0078]

[Equation 2] Here, V (Xij) is a function that represents the penalty due to the i-th data being moved from the data storage area where it was originally stored, and Dik is a proportional constant that determines the balance between the penalty and the first term. is there. V (
As Xij), if the number of the data storage area where the i-th data was originally stored is written as Di, for example, the Kronecker delta function δ (
i, j)

[0079]

[Equation 3] The evaluation function for evaluating the data arrangement is not limited to the quadratic expression, and higher-order expressions can be used. Also, another method may be used for expressing variables. Furthermore, it is also possible to use an evaluation function considering the access frequency and the load of the storage device at the data allocation destination. Alternatively, it is also possible to describe them as constraints for the evaluation function.

When the data arrangement is evaluated by using the above evaluation function, the evaluation value of the evaluation function may not change even if a certain data is arranged in either the data storage area a or b.

At this time, the data arrangement determining unit 4 randomly selects and arranges the arrangement destination. Alternatively, with respect to such data, the operator can be presented with a candidate for a placement destination and designated to which one. Further, it is possible to uniquely assign a number to the data storage area and store the number in the smaller or larger number. Alternatively, the concept of distance using the number may be introduced, and one having a shorter distance between the data storage area in which the current data is stored and the data storage area of the relocation destination may be set as a candidate. As the distance, for example, the difference between the numbers can be used as the distance between the two data storage areas. In addition, a means for evaluating the degree of association between data is added,
It is also possible to evaluate the degree of association between the data to be moved now and the data stored in each data storage area, and select the one with the most data with a high degree of association as the destination. As the degree of association, for example, the number of subtrees that include both leaves that indicate the position of data when creating data indexes in a tree structure, or in a relational database, etc., data has the same relation. Whether it is included or not can be used. Furthermore, it is also possible to add a means for checking the load of each data storage device and give priority to the lighter load at the relocation destination.

After formulating in this way, the best data arrangement can be determined by, for example, obtaining the evaluation values of all combinations regarding the data arrangement. Alternatively, after the above formulation, this problem can be transformed into a linear programming problem and solved using a linear programming method. Alternatively, this problem can be solved using a simulated annealing method. Furthermore, when the arrangement is not necessarily the best arrangement, and it is sufficient that the arrangement is sufficiently close to the best arrangement, it is possible to predetermine conditions and select the arrangement when the arrangement satisfying the conditions is obtained.

The data arrangement changing unit 6 changes the arrangement of data based on the arrangement received from the data arrangement determining unit 4. The data arrangement changing unit 6 needs to disable updating of the data before changing the arrangement of the data. This is because if there is an update to the data during transfer from the original data storage area to the change destination data storage area, the data consistency may not be maintained. Therefore, first, in order to put the corresponding data in the unwritable state, the device that manages the data lock, for example, the data access unit 7 in the configuration example of FIG. After that, the data is moved to the transfer destination data storage area, and after the transfer, the lock for the data is released. Alternatively, you can make a copy of the data and make changes to the copy during the move so that the changes in the move are reflected in the original data after the move is completed. In this case, change the data arrangement. It is possible to access the data inside.

Further, when transferring data, the loads on all storage devices are not always equal. Transferring data from a heavily loaded storage device takes that much time. In addition, since the weight of the load changes from moment to moment, when considering the possibility that the load on the storage device will be lighter later, when transferring data, the transfer from the storage device with a heavy load is postponed. It is better to Therefore, it is possible to add a means for checking the load of each data storage device and to preferentially execute the relocation to the data storage area in the storage device having the light load.

Next, an embodiment in which the present invention is applied to a database using indexed sequential files stored on a plurality of disks will be described.

FIG. 19 shows the configuration of the data management apparatus of this embodiment.

In this embodiment, a processing request for data access is executed on a plurality of computers connected via a network, and the data is distributed and stored in the disks attached to the respective computers. The system is composed of one master computer 40 and another computer 43. FIG. 20 shows a configuration example of the master data management device 41 connected to the master computer 40 and the data management device 44 connected to the other computers 43.

The data management device 44 connected to the computer 43 other than the master computer has the access request receiving unit 1, the correlation information generating unit 31, the data arrangement changing unit 6, the access history storage table 33, and the access history storage table 33 shown in FIGS. It comprises a data storage information storage unit 5, a series of processing information storage tables 32, a data access unit 7, and a means 61 for receiving an access request from another computer. Access request receiver 6 from others
When the data to be accessed does not exist in the own computer, 1 is requested to be accessed by another data management device, and serves to receive the request and pass it to the data access unit 7.

In the master data management device 41 connected to the master computer 40, a correlation information table management unit 34, a data arrangement determination unit 4, a significant information selection unit 36, and a correlation information storage table 2 are added to those described above. Exists.

Each processing request uses the library function provided by the data management device to access the database. The library function has read () used when reading data, write () used when writing, and c used when performing commit processing.
ommit () and so on. A unique data number is assigned to the data stored in the database by the system, and the data can be identified by specifying this number. Each processing request specifies this identifier when accessing the data.

When these library functions are called by a processing request, first, the process id of the calling process is acquired from the operating system. The process id is an identification number uniquely assigned to each process by the operating system. Then, the process id, the data identifier of the access destination, the flag indicating which library has been called, and the parameter for distinguishing them when a plurality of series of processes are defined on one process as described later. , And sends parameters required for access processing to the access request receiving unit 1. As a parameter for distinguishing a series of processes, a numerical value that is uniquely determined for each process executed on the process can be assigned and used. The access request receiving unit 1 transmits a process id, a data identifier, a parameter for distinguishing a series of processes, and a flag indicating a library to the correlation information generating unit, and sends the data identifier of the access destination and the parameter required for the process. The data is transmitted to the data access unit 7. The data access unit searches the index in the data storage information storage unit 5 as shown in FIG. 21A using the identifier of the received data as a key. The index has a tree structure that uses the data identifier as a key. In the leaf part of the tree structure, which computer stores that data, and when that data exists in its own computer, in which file and at which position. A flag indicating whether the data is stored or the data is locked is stored. If it is found as a result of searching the index that the own computer has the data, the data access unit 7 accesses the data based on the search result and sends the result to the library function. If necessary at the time of data access, it is possible to prohibit the access of other processes by setting an information lock flag in the index for that data. If it is stored in another computer, a data access request is sent to the data management device on the other computer. When the result is received, the access to the data is completed by returning it to the library function.

When the correlation information generating unit 31 receives the process id, the data identifier, the flag indicating the library and the parameter for distinguishing the series of processes from the data access unit 7, the series processing information storage table 32 is based on the process id.
To search. The series processing information storage table 32 is shown in FIG.
As shown in, each process defines what kind of process should be regarded as a series of processes. In (a) of FIG. 22, the library function called from the begin () to the commit () of the process with the process id of 10 is defined as a series of processes. In addition, two processes are defined for the process whose process id is 20, and they are distinguished according to the value of the parameter, and the library function called from each of them is defined as a series of processes from begin () to commit (). ing. In this example, the parameter is only one integer type variable that takes a positive value, but it is possible to set multiple parameters. The series process information storage table also has a part for defining default processes as shown in FIG. 22B, and in this embodiment, the library function init () called at the start of the process to the library function exit () called at the end of the process. ) Is defined as a series of processes, so those whose process id is not listed in this table follow this default definition. A unique number can be assigned to each series of processes. Since a plurality of series of processes may be defined in one process, it is possible to uniquely assign a number to a series of processes with a set of numbers (process id, serial number of processes in process). In the case of the default process, there is one process in that process, so the sequence process number is (process id, 1).

The correlation information generator 31 can search the series of process information storage tables 32 to determine which series of processes the current access is part of, and obtain the process number.
Next, the access history storage table 33 is searched using the processing number as a key.

FIG. 23 shows a configuration example of the access history storage table 33.
Shown in. In this configuration example, the access history storage table 33 has a tree structure with the process id as a key, and each leaf part has a plurality of linked list structures corresponding to the serial numbers assigned to the respective processes in the process. And stores the identifier of the accessed data in the linked list.
FIG. 22 shows that the history regarding the process ids 2, 5, 10, and 20 is stored.

When the access history storage table 33 is searched and there is no entry for the processing, a flag indicating the library received from the access request receiving unit as to whether the call is from the library for starting a series of processing and the series information storage table 3
2 is compared with a flag indicating the start library of the entry, and if this access starts a series of processes, an entry corresponding to this process is newly secured and the data identifier is stored in the linked list. On the other hand, if the two flags are different, nothing is done. When the access history storage table 33 is searched and there is an entry for the processing, the identifier of the data accessed this time is first added to the linked list. After that, the flag indicating the library received from the access request receiving unit 1 is compared with the flag indicating the end library of the series process information storage table 33, and when they are different, nothing is done. If they match, a series of processing is completed by this access, so that in order to generate the correlation information, all the identifiers of the data accessed by the processing in the past are read and the entry is deleted.

The correlation information generator 31 creates the correlation information as follows. Assuming that the identifiers of the data accessed in the series of processes just completed are 14, 17, and 39, the correlation information generating unit 31 generates the following information (data id, data id, access count). .

(14, 17, 1) (14, 39, 1) (17, 39, 1) Then, a set of these correlation information is transmitted to the correlation information table management unit 34 in the master computer 40.

Correlation information table management unit 3 on the master computer 40
FIG. 24 shows an example of the configuration of No. 4. In the present embodiment, the correlation information table management unit 34 includes a correlation information receiving unit 62 that receives the correlation information from the correlation information generating unit 31 on each computer, and a correlation that searches whether the received result is in the correlation information table 2. An information table search unit 63, a correlation information table updating unit 64 that updates each correlation information of the correlation information table 2, a correlation information number storage unit 65 that stores how many correlation information items are registered in the correlation information table 2, and a correlation. The deletion entry selection unit 66 that selects the correlation information to be deleted when the information table 2 is full, the correlation information table sorting unit 67 that sorts each entry of the correlation information table 2 by the access count, and the number of times the correlation information is acquired. An acquisition count registration unit 68 that is registered in advance, an acquisition count storage unit 69 that stores the number of times correlation information has been acquired so far, and an empty entry that selects an empty entry when the correlation information table is empty Entry selector 70 consists of correlation information table insertion portion 71 for inserting new data in the correlation information table.

Next, the operation of the correlation information table management section 34 will be described with reference to FIG.
This will be described using 5. The correlation information sent from the correlation information generating unit 31 is received by the correlation information receiving unit 62 (step S41). Since the correlation information is composed of a plurality of pieces of information as described above, the correlation information table search unit 63 is used to search for each element as to whether or not it is already stored in the correlation information storage table 2 (step S42, S4
3). As a result of the search, if the information is not stored in the correlation information storage table 2, the correlation information table insertion unit 71 is instructed to newly insert it in the table (step S44), and if it is stored, the correlation information table is updated. The section 64 is instructed to update the correlation information of the entry (step S45).
After that, the correlation information acquisition count stored in the acquisition count storage unit 69 is increased by 1 (step S46). The number of acquisition times registration unit 68 and the correlation information acquisition number storage unit 69
(Step S47), and if the predetermined number is reached, it is assumed that sufficient correlation information has been acquired, and the current correlation information table 2 is read out (step S47).
49), the contents are sorted using the correlation information table sorting unit 67 (step S50), and the significant information selecting unit 36 is notified together with the fact that the correlation information is sufficiently acquired (step S51). Then, all the entries in the correlation information table 2 are deleted, and the correlation information is newly collected. If the number of times is less than the predetermined number, the process for the next element is continued. The above processing is performed for all the elements included in the correlation information received from the correlation information generation unit 31 (step S4).
8).

The correlation information table insertion unit 71 first reads the number of data items currently stored in the correlation information table 2 from the correlation information number storage unit 65 and checks whether the correlation information table 2 is already full. If there is a vacancy, a vacant entry selection section 70
To select an empty entry, and when there is no empty entry, the deletion entry selection unit 66 is caused to select an entry to be deleted from the correlation information table 2. Then, the correlation information is stored in the selected entry.

When receiving a command to update the correlation information of a certain entry, the correlation information table updating unit 64 increases the access count of the entry storing the relevant correlation information by one.

The deletion entry selection unit 66 first sorts the correlation information table 2 by the number of accesses using the correlation information table sorting unit 67. Then, after sorting, one entry is randomly selected from the entries with the lowest access count and is selected as the entry to be deleted.

As shown in FIG. 26, the correlation information storage table 2 is composed of a plurality of entries for storing the correlation information, and each entry is composed of a set of ids of the accessed data and the number of times the set is used together. There is.

When the significant information selecting unit 36 receives from the correlation information table managing unit 34 that the correlation information is sufficiently acquired and the correlation information table is received, the significant information selecting unit 36 selects a significant one from the received information. Since each entry of the correlation information table 2 is sorted in advance by the access frequency in each entry in the correlation information table management unit 34,
An entry having an access count equal to or greater than a predetermined ratio with respect to the maximum access count is selected as significant. Then, only the one obtained as a result is transmitted to the data arrangement determination unit 4.

The operation when the data arrangement determining unit 4 determines the data arrangement will be described with reference to FIG. When significant correlation information is received from the significant information selection unit (step S61),
According to the index and the disk capacity table in the data storage information storage unit 5 shown in FIGS. 21A and 21B, the storage device where the data is stored, the size of each data, or each The data storage information indicating the capacity of the storage device is read out (step S63). Then, the data arrangement is evaluated by using the following evaluation function (step S64). here,
Variable Xij (1 ≦ i ≦ N, 1 ≦ j ≦ M, Xij = 0 or
1) is a variable indicating that the i-th data is stored in the j-th data storage area. Dik is the i-th and k
The th data is the number of accesses used together,
This is obtained from the correlation information. This value is zero for combinations with no correlation information.

[0106]

[Equation 4] Also, as a constraint condition expression, the capacity of the j-th disk is d
Let j be the size of the i-th data and si be the following. These pieces of information are obtained from the data storage information.

[0107]

[Equation 5] Then, first, the value of Q relating to the current arrangement is calculated using equation (4), and the current arrangement and the value of Q are stored. Next, the data arrangement is changed so as to satisfy Expression (5). When a good value is obtained by comparing with the stored value of Q (step S65), the value and the arrangement are stored, but if not, the next arrangement is selected (steps S66 and S67). If a new layout can be selected here (step S68),
The arrangement is changed (step S69), and the calculation of the value of Q is repeated again. The arrangement having the best evaluation value can be selected by repeating such processing until a new arrangement cannot be selected.

Here, in order to select the data arrangement that satisfies the expression (5), the following may be done. First, there are N pieces of data that appear in significant correlation information, and the data storage area is M
Suppose that there is an individual. Then, the arrangement of these data is N
It is conceivable that the M-th power is arranged. First select one from this arrangement. Then, since the other data is not moved and only the data appearing in the correlation information is moved, the arrangement in which the expression (5) is not satisfied may occur. In that case, some of the data in the storage device that does not satisfy the expression (5) that do not appear in the correlation information are selected, and the data that does not satisfy the expression is stored in the storage device as desired. Move to the device. In this way, if the equation (5) can be satisfied for all the storage devices, the arrangement is selected as an arrangement that satisfies the equation (5), and if not, the arrangement is rejected.
In this way, the process is repeated until all the M-th power placements of N are selected.

After selecting the layout in this way, the data layout determining unit 4 instructs the data layout changing unit 6 to change the data layout. The operation in this case will be described below.

In this embodiment, since it is necessary to move the data on a plurality of computers, the data placement determining unit 4 instructs the data placement changing unit 6 on how to move the data. For that purpose, the data arrangement determining unit 4 first transmits information about movement within a plurality of data storage areas in each computer to the data arrangement changing unit 6. Data arrangement changing unit 6
When this information is received, the data is locked to temporarily prohibit access and change the layout of the data. After that, the fact that the change of the data arrangement is completed is transmitted to the data arrangement determining unit 4. Upon receiving the notification that the data arrangement has been changed from all the data arrangement changing units 6, the data arrangement determining unit 4 starts changing the data arrangement between the remaining computers. For that purpose, first, all the data arrangement changing units 6 are notified to prohibit access to all the data. When receiving the notification that access to the data is prohibited from all the data arrangement changing units 6, the data arrangement determining unit 4 first selects one data to be changed, and the data arrangement changing unit 6 on the computer having the data is selected. On the other hand, indicate the destination computer and instruct it to change the data. Upon receiving this instruction, the data arrangement changing unit 6 transmits the data to the designated data arrangement changing unit 6 on the computer. Change destination data arrangement changing unit 6
When the data arrangement determination unit 4 receives the fact that the arrangement change has been completed, the data arrangement determination unit 4 selects the next data to be changed. When the arrangement of all the data is changed in this way, the data arrangement determining unit 4 transmits the new arrangement to the data storage information storage units 5 on all the computers, and when the transmission is completed, all the data arrangement changes are made. The unit 6 is notified to unlock the data.

Next, an embodiment in which the present invention is applied to a data management device for managing read-only data will be described.

A read / write storage device can be used to store read-only data, and a read-only storage device that can be written only once when initially storing data can be used. When a readable / writable storage device is used, it can be realized using the configuration described above. On the other hand, when a read-only storage device such as a CD-ROM is used, it can be realized in the same manner as the above method until the data rearrangement method is determined. You cannot write to read-only storage. In that case, as shown in FIG. 28, a read-only storage device 8 in which data is first stored.
2 and a readable / writable storage device 84 for newly storing data, and a read storage device identification unit 86 for identifying from which storage device the data should be currently read, and is newly arranged according to the characteristics of the system. When reworking, the data can be stored in the readable / writable storage device 84. In this case, when the data is accessed or the data is rearranged, the data access unit 7 and the data arrangement changing unit 6 inquire the read storage device identification unit 86 from which storage device the data should be read.

Alternatively, as shown in FIG. 29, a means for newly creating a read-only storage device 88 such as a CD-ROM is provided, and when the data is newly rearranged according to the characteristics of the system, the data is stored. It is also possible to store the read-only data in the read-only storage device 82 again and read it from the newly created read-only data storage device 82.

Since this is a read-only data management device, the data will not be updated when the data allocation is changed to a new storage device. Therefore, the data allocation is changed while making a copy to a new area. However, it is also possible to access the original data while changing the data arrangement and to access the new area after copying all the data.

It is also possible to divide the contents of the read-only storage device storing the read-only data into a plurality of storage devices and store the contents, or it is also possible to store a plurality of read-only storage devices. It is also possible to rearrange the data of 1) in one storage device.

By applying the present invention to the read-only data management device as described above, the data arrangement can be optimized in accordance with the difference in the access pattern for the data stored in the system, and the read speed can be increased. Can be executed.

The present invention is not limited to the above-mentioned embodiments, but can be carried out in various modified forms without departing from the scope of the invention.

[0118]

As described above, according to the data management method of the present invention, when the data stored in a plurality of data storage areas are frequently referenced / updated together in a series of processes, the correlation between them is high. Can be automatically determined, and the data arrangement can be automatically changed so that such data is stored in the same data storage area.

Further, since the system administrator does not need to analyze the processing on the system to determine the data allocation because the information necessary for the data allocation is automatically acquired, the execution time of each process is shortened. It will be possible to easily bring out the performance of the.

[Brief description of drawings]

FIG. 1 is a diagram showing an example of a configuration of a data management device according to the present invention.

FIG. 2 is a diagram showing another configuration example of the data management device according to the present invention.

FIG. 3 is a diagram showing still another configuration example of the data management device according to the present invention.

FIG. 4 is a diagram showing still another configuration example of the data management device according to the present invention.

FIG. 5 is a diagram showing still another configuration example of the data management device according to the present invention.

FIG. 6 is a diagram for explaining an example of how to allocate a data storage area to a storage device.

FIG. 7 is a diagram for explaining another example of how to allocate a data storage area to a storage device.

FIG. 8 is a diagram for explaining still another example of how to allocate a data storage area to a storage device.

FIG. 9 is a diagram showing still another configuration example of the data management device according to the present invention.

FIG. 10 is a diagram showing a configuration example in which a data management device manages a plurality of storage devices.

FIG. 11 is a diagram showing another configuration example when the data management device manages a plurality of storage devices.

FIG. 12 is a diagram showing still another configuration example when the data management device manages a plurality of storage devices.

FIG. 13 is a diagram showing a configuration example in which each computer has a data management device.

FIG. 14 is a diagram showing a configuration example when a data management device is shared by a plurality of computers.

15 is a diagram showing an example of a configuration of a correlation information management unit in FIG.

16 is a flowchart showing an example of the operation of the correlation information generation unit in FIG.

FIG. 17 is a flowchart showing an example of the operation of the correlation information table management unit.

FIG. 18 is a flowchart showing an example of the operation of the data arrangement determination unit shown in FIG.

FIG. 19 is a diagram showing a configuration example of a system when the present invention is applied to a database.

FIG. 20 is a diagram showing a configuration example of a master data management device and a data management device of another computer.

FIG. 21 is a diagram showing a configuration example of an index of the data storage information storage unit and a disk capacity table of the data storage information storage unit.

22 is a diagram showing a configuration example of a series of process information storage table of FIG.

23 is a diagram showing a configuration example of the access history storage table of FIG.

24 is a diagram showing a configuration example of a correlation information table management unit in FIG.

FIG. 25 is a diagram showing an operation example of a correlation information table management unit in FIG.

26 is a diagram showing a configuration example of a correlation information storage table in FIG.

FIG. 27 is a flowchart showing an example of an operation in which a data arrangement determination unit determines data arrangement.

FIG. 28 is a diagram showing a configuration example of a read-only data management device.

FIG. 29 is a diagram showing another configuration example of a read-only data management device.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Access request receiving unit, 2 ... Correlation information storage table, 3 ... Correlation information management unit, 4 ... Data allocation determination unit, 5 ... Data storage information storage unit, 6 ... Data allocation changing unit, 7 ... Data access unit, storage Device 8, computer group 9

Claims (1)

[Claims] 1. An access request receiving step of receiving an access request for data stored in a plurality of different data storage areas, and mutual correlation of data indicating that a plurality of specific data have been accessed together in a series of processes. Correlation information creating step of creating information based on a predetermined number of access requests, data storage information indicating a data storage area in which data is stored, and data arrangement for determining new data arrangement in the data storage area from the correlation information A data management method comprising: a determining step.