JPH07239808A - Distributed data management method - Google Patents

Abstract

(57) [Summary] [Purpose] The purpose of the distributed data management system is to improve file transfer efficiency by minimum required file transfer in page units. In the server computer 10, an on-demand page transfer means 10b that realizes a minimum required file transfer in page units, a block page transfer means 10c that realizes a minimum required number of file transfers by dividing a plurality of pages into blocks, and an actual file. Shadow page transfer means 10d for converting the contents of the page into low-resolution image data to realize high-speed file transfer, a delayed page transfer means 10f and a delayed page writing means 10g for efficiently controlling writing to the server file in page units. It consists of and.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a distributed data management system for transferring files between a client computer and a server computer connected on a distributed network.

[0002]

2. Description of the Related Art FIG. 1 is a diagram showing a system configuration of a distributed data management system. FIG. 6 shows a block diagram of the conventional system. In FIG. 6, 10 is a CPU,
A server computer composed of a memory and a large-capacity secondary storage device, 20 and 30 are client computers requesting various services to the server computer, and connected to the server computer 10 via data transmission paths 40 and 50. ing. Here, the transmission line 40 is a local area network, and the transmission line 50 is a wide area network which is a wide area communication network. Conventionally, when a file is transferred between computers connected to a network via a data transmission path, one file is the minimum unit of file transfer, and access to a specific page in the file is performed after transferring the entire file. .

For example, as shown in FIG. 6, when a specific page in a file 11 on the server computer 10 is taken out to the client computer 20, first, the server computer 10 is selected.
The file transfer means 10h and the transmission means 10a of
All data for one file is transferred to the client computer 20 through the transmission line 40. In the client computer 20, the transmission means 20a and the file transfer means 20c are used to
All data for a file is received as a file 22.
After that, the page access means 20d took out specific page data from the data for one file and processed it.

Further, even when a specific page in the file 11 of the server computer 10 is replaced with a specific page in the file 22 on the client computer 20, the transfer of the entire one file including the specific page is performed by each means shown in FIG. The entire file was replaced by executing.

[0005]

Since the conventional file transfer method is configured as described above, in the normal operation of file data, the data actually required is a part of the entire file. Despite the large number of files, one file must be transferred as the minimum unit of data transfer, and the entire file including the necessary part must be transferred. Therefore, the traffic volume of the transmission path increases, and the data transmission efficiency on the network deteriorates. There was also a problem that the contents of the file could not be referenced at high speed.

The present invention has been made in order to solve the above-mentioned problems, and the amount of data transfer and the transfer frequency between computers configured on a distributed network are suppressed as much as possible to improve the response time. The purpose is to provide an efficient distributed data management method.

[0007]

DISCLOSURE OF THE INVENTION A distributed data management system according to a first invention is, in a server computer, a real file store management means for managing real file data on the server computer in page units, and a file from a client computer. A demand page transfer means for performing data transfer in page units including the access request portion in response to an access request, and the client computer is provided with page data transfer requesting means for the server computer. .

In the distributed data management system according to the second invention, in the server computer according to the first invention, a plurality of pages including page data requested to be accessed by the client computer are transmitted in a transmission buffer size of a data transmission path. Block page transfer means for performing block transfer in batch based on the above.

A distributed data management system according to the third invention is, in the server computer according to the first invention, shadow page file storage means for converting real file data on the server computer into low resolution image data,
A shadow file management means for managing the shadow page file data and a shadow page transfer means for transferring the shadow file on the server computer in page units are provided.

A distributed data management system according to the fourth invention is a server computer according to the first invention, the second invention or the third invention, wherein a cache file comprising page data management elements on a client computer. In response to a write request from the store means and the client computer, a delayed page transfer means for setting the write target file name, page number and storage destination information of the write page data in the cache file store means, and the cache file store means A delay page writing means is activated for a write request from the client computer at a certain timing and writes out as page data as real file data.

[0011]

According to the present invention, only required pages are transferred, and pages for reference can be transferred as shadow pages composed of low resolution data, so that the file transfer amount can be suppressed. You can
Further, since pages before and after the designated page can be transferred in advance as block pages, the number of file transfers can be suppressed. Further, since the page transfer and the file write are independently operated by the delayed writing of the server file, the apparent response time to the page update processing can be shortened.

[0012]

【Example】

Example 1. An embodiment of the present invention will be described below with reference to the drawings. 2 is a block diagram of a distributed data management system,
Reference numeral 11 is a real file store that manages real files in page units, 12 is a shadow file store that converts page contents of real files into low-resolution image data and manages them with shadow page data, and 13 is a page for updating real files. It is a cache file store that manages update source pages so that they can be collectively performed in units. Also,
FIG. 3A is a diagram showing the logical structure of the real file store 11. In the figure, 110 is a file header, 1
11 is a page management unit, 112 is an access management unit, 113
Is an actual file page obtained by dividing data, which is a file for one page, into page units. FIG. 3B is a diagram showing the logical structure of the access control unit 112 which constitutes the actual file store 11. In the figure, reference numeral 1121 is an access page number indicating which page in the file has an access request, and 1122 is access mode information indicating whether the access request is in the shared (S) mode or the exclusive (X) mode. Reference numeral 1123 is an identification number (client ID) for identifying the client computer that issued the file access request.

FIG. 4 is a diagram showing the logical structure of the shadow file store 12. In the figure, 121 is the file name of the real file, 122 is the page number of the real file, 1
Reference numeral 23 is the number of accesses indicating the access frequency of the shadow page, 124 is a shadow page pointer indicating the storage position of the shadow page, and 125 is shadow page data which is a shadow file for one page.

FIG. 5 is a diagram showing the logical structure of the cache file store 13. In the figure, 131 is the file name of the real file, 132 is the page number of the real file,
133 is an identification number for identifying the client computer that has made a write request, 134 is stamp information indicating the time when an access request was made from the client computer, 1
Reference numeral 35 is an address pointer indicating the address of the write source page data 21.

Next, the operation of file transfer in this system will be described for each case of on-demand page transfer, block page transfer, shadow page transfer, and delayed page transfer. First, the on-demand page transfer will be described with reference to FIGS. 2 and 3. <Step 1> When an “on-demand page transfer request” is issued from the client computer 20 to the server computer 10, the on-demand page transfer means 10b checks the file header 110 of the real file store 11 and confirms that the file is a corresponding file. After the confirmation, the page management unit 111 is referred to obtain the storage location of the page for which the transfer is requested. <Step 2> Next, the access page number 1121 of the access control unit 112 is checked to see if there is the same page number as the transfer request page number. <Step 3> If there is no same page number, the page number for which the transfer request is made is made to the access page number 1121 and the access mode 1122 is used in the shared (S) or exclusive (X) mode. Further, the identification number of the client computer is set in the client ID 1123, and the request page data 113 is transmitted by the transmission means 10a.
The file is transferred to the client computer 20 via the. <Step 4> At the end of file transfer,
The access page number 1121, access mode 1122 set in <step 3> corresponding to the transfer request page,
The information of the client ID 1123 is added to the access control unit 1
Delete from 12 entries. <Step 5> On the other hand, if the same page number as the transfer request page number already exists in the access control unit 112 in the above <Step 3>, it is considered that the page has already received a transfer request from another client computer. Therefore, the access mode 1122 is checked, and if the access from the client computer is the shared (S) mode, the file transfer is performed as described above. <Step 6> On the other hand, if the access mode 1122 is the exclusive (X) mode, a page transfer request rejection response is returned to the client computer 20 that made the request.

Example 2. Next, the block page transfer will be described with reference to FIGS. The block page transfer means 10c is called by the on-demand page transfer means 10b, and operates when "prefetch" is designated as an option of the page transfer request from the client computer 20 in the on-demand page transfer. <Step 1> The block page transfer means 10c compares the size of the transfer request page with the transfer buffer size (not shown) of the transmission means 10a, and if there is a free transfer buffer, the page management unit 111 stores the next page. Find position and page size. The size of the next page is compared again with the size of the transfer buffer margin, and the same processing is repeated until the transfer buffer has no space. <Step 2> When the transfer buffer becomes full, predetermined information is set in the access control unit 112 as in the case of on-demand page transfer, and a file is transferred to the client computer 20 via the transmission means 10a. The block page transfer is effective when transferring a file composed of a large number of pages and each page size is small.

In the above embodiment, the method of blocking the next page (the one with the larger page number) has been described. The page for which the transfer request this time is compared with the previous transfer request page, and the previous page (page To the one with a smaller number)
Block page transfer means 10c performs a process of automatically determining whether to block a subsequent page (to a larger page number), or a process of blocking a page before and after a transfer request page. May be performed in. According to this embodiment, the pages before and after the designated page for which the data access request is made can be collectively transferred in advance as one block block, so that the file transfer frequency can be suppressed and the client computer and the server computer can be controlled by this. This has the effect of reducing overhead processing.

Example 3. Next, the shadow page transfer will be described with reference to FIGS. <Step 1> When the client computer 20 issues a “shadow page transfer request” to the server computer 10,
The shadow page transfer means 10d checks the file name 121 and the page number 122 of the shadow file store 12 to check whether there is a requested page. <Step 2> If there is a corresponding page, access count is 12
3 is incremented, and the shadow page pointer 124
The shadow page data 125 is extracted from the file and transferred to the client computer 20 via the transmission means 10a. At this time, the file name 121 and the page number 122 are stored inside the shadow page transfer means 10d. <Step 3> If there is no corresponding page, the shadow file management means 10e is automatically called to create shadow page data of the corresponding page in the shadow file store 12. At this time, the shadow file management means 10e
Shows the file name and page number of the page for which a shadow page transfer request is made in the empty record area of the shadow file store 12, the file name 121 and the page number 12
2, the corresponding page data 113 of the actual file store 11 is read, the content of the page data is converted into low resolution image data, and the low resolution image data is stored in the shadow page data 125. An address indicating the position of the shadow page data 125 is set in the shadow pointer 124, and the access count 123 is set to zero. Subsequent operations are the same as when there is a corresponding page. <Step 4> When the “real page transfer request” is issued from the client computer 20 subsequent to the “shadow page transfer request”, the on-demand page transfer means 10b is determined by the file name and page number stored in the shadow page transfer means 10d. To transfer the actual file. Further, in the above-described embodiment, the shadow page transfer and the real page transfer are performed by the shadow page transfer means 10d and the on-demand page transfer means 10b based on different transfer requests, but based on one transfer request. The shadow page transfer means 10d alone may transfer the shadow page and the real page at the same time.

The shadow file management means 10e
Can be started not only while the system is running, but also before the system is running. That is, the shadow file management means 10e is used by the shadow file store 12
Corresponding to the page 1 of the real file store 11 for the file name 121 and the page number 122 of
13 is read, the contents of the page are converted into low resolution image data and stored in the shadow page data 125.
Then, the address indicating the position of the shadow page data 125 is set in the shadow pointer 124. In this way, the shadow file store 12 can be collectively created by the shadow file management means 10e in advance before using the system.

When a shadow page transfer request is issued from the client computer 20 while the system is operating and there is no corresponding shadow page data in the shadow file store 12, the shadow page transfer means 10d in the server computer 10 forcibly manages the shadow file. Means 1
0e is called and the shadow file of the requested page is immediately created. (Create on-demand shadow page)
Further, the shadow file management means 10e checks the access count 123 of the shadow file store 12,
If the access frequency is zero or the access frequency is less than the specified value, the access frequency is considered to be low and it is automatically deleted. here,
The access frequency check is performed regularly or at a designated time, thereby effectively using the space (storage area) of the shadow file store 12. As described above, according to this embodiment, it is possible to transfer the minimum necessary files in units of pages forming one file, and to reduce the data amount for a page accessed for the purpose of reference only, a low resolution. Since it can be transferred as a shadow page consisting of data, there is an effect that the file transfer amount on the network can be suppressed.

Example 4. Next, the delayed page transfer will be described with reference to FIGS. <Step 1> When the client computer 20 issues a “delayed page transfer request” to the server computer 10, the delayed page transfer means 10f causes the cache file store 13 to operate.
File name 131, page number 132, and client ID 1 for identifying the requesting client computer
Set 33. <Step 2> Next, the address indicating the storage position of the write page data 21 is transferred from the client computer 20 to the server computer 10. Delayed page transfer means 10f
Stores this storage location address in cache file store 1
No. 3 is set in the page pointer 135, the time at that time is set in the time stamp 134, and the completion response of the write request is returned to the client computer 20. <Step 3> The write request accumulated in the cache file store 13 is processed by the delay page writing means 10g which operates under a constant time interval or conditions such as cache file full. The delayed page writing unit 10g checks the file name 131 and the page number 132 of the cache file store 13 and checks whether the same file name and page number are registered in another entry of the cache file store. <Step 4> If the same is not registered, the page data 21 on the client computer 20 is read via the transmission means 10a by the client ID 133 and the page pointer 135, and the corresponding page on the real file store 11 on the server computer. Write in. <Step 5> If there is another file name / page number that is the same, the page data 21 with the latest time of the time stamp 134 is written to the real file store 11 by the same process as in <Step 4>. The processing of <step 4> and <step 5> is repeated until the processing for all the records in the cache file store 13 is completed.
As described above, according to this embodiment, by implementing the delayed write function, the page data transfer process between the client computer and the server computer and the file write process on the server computer operate independently. Since this is done, there is an effect that the response time to the page data update processing request can be shortened.

[0022]

As described above, according to the present invention, it is possible to transfer the minimum required files in units of pages that make up one file, and reduce the data amount for pages accessed only for reference. Since it can be transferred as a shadow page consisting of low resolution data,
This has the effect of reducing the amount of file transfer on the network. In addition, the pages before and after the specified page for which a data access request is made can be collectively transferred in advance as one block page, so the file transfer frequency is suppressed and the overhead processing in the client computer and the server computer is thereby suppressed. There is an effect that can be reduced. Furthermore, by implementing the delayed write function, the page data transfer process between the client computer and the server computer and the file write process on the server computer can be operated independently, so the page data update process This has the effect of shortening the response time to a request.

[Brief description of drawings]

FIG. 1 is a system configuration diagram showing an embodiment of the present invention.

FIG. 2 is a block diagram showing an embodiment of the present invention.

FIG. 3 is a logical structure diagram of a real file store according to an embodiment of the present invention.

FIG. 4 is a logical structure diagram of a shadow file store according to an embodiment of the present invention.

FIG. 5 is a logical structural diagram of a cache file store according to an embodiment of the present invention.

FIG. 6 is a block diagram showing a conventional file transfer method.

[Explanation of symbols]

 10 server computer 10a transmission means 10b on-demand page transfer means 10c block page transfer means 10d shadow page transfer means 10e shadow file management means 10f delayed page transfer means 10g delayed page writing means 11 real file store 12 shadow file store 13 cache file store 20 Client computer 30 Client computer 40 Data transmission line (local area network) 50 Data transmission line (wide area network)

Claims (4)

[Claims] 1. A distributed data management system comprising client computers distributed on a network for requesting a service and a server computer processing a service request from the client computer, wherein the server computer stores real file data on the server computer. The client computer includes a real file store management unit that manages in page units, and a demand page transfer unit that transfers data in page units including the access request location in response to a file access request from the client computer. A distributed data management method characterized by comprising means for requesting page data transfer to a computer. 2. The server computer comprises block page transfer means for batch-transferring a plurality of pages including page data requested to be accessed by the client computer, based on the transmission buffer size of the data transmission path. The distributed data management system according to claim 1, characterized in that. 3. A server computer, a shadow page file storage means for converting real file data on the server computer into low resolution image data, a shadow file management means for managing shadow page file data, and a shadow on the server computer. The distributed data management system according to claim 1, wherein a high speed read request from a client computer is made possible by comprising a shadow page transfer means for transferring a file in page units. 4. The server computer comprises a cache file store means comprising page data management elements on the client computer, and a file name, page number and write page data to be written in the cache file store means in response to a write request from the client computer. A delay page transfer means for setting storage destination information of the cache file, and a delay page writing means for writing the write request from the client computer stored in the cache file store means for each page as real file data, which is activated by a certain trigger. 2. The method according to claim 1 or 2, characterized in that
The data management method according to item 3 or item 3.