JPH0612388A - Access system for decentralized file - Google Patents

Abstract

PURPOSE:To speed up processing by reducing the load on a line and decreasing the frequency of input to and output from an external storage device in the decentralized file system. CONSTITUTION:One block in a file of another computer on a network is inputted from a terminal device 16. If, however, the block is not existed in the memory 15 of an access request source computer 11, the contents of the block are sent from a computer where the desired file to be accessed is existed and the contents while sent back to the terminal device 16 are entered into the memory 15. When a request for input to the block is made thereafter, the contents of the memory 15 are only sent back to the terminal device 16 to complete the access processing.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a distributed file system in which, for example, when a file is copied to another computer or when a database of another computer is accessed, the load on the line is reduced, and further, the other computer. The present invention relates to an access processing method capable of high-speed processing by reducing the number of database accesses of the system.

[0002]

2. Description of the Related Art In a distributed file system, when a file of another computer on a network is accessed, in the conventional technique, generally, every time an access request is made by a user program, the access request source computer makes a change to a computer where the file exists. A data transmission request was made and the transmitted data was returned to the user program. In this method, if the access is frequent, the load on the line is increased, and the access speed seen from the user is reduced. Therefore, Japanese Patent Laid-Open No. 3-
In Japanese Patent No. 263940, when an input request is made to a file of a remote host computer, the file is copied to the external storage device of the relay machine, and thereafter, the input to the file is copied to the external storage device of the relay machine. It was to access a certain file.

[0003]

DISCLOSURE OF THE INVENTION Problems to be Solved by the Invention
In the technique of Japanese Patent Publication No. 940, when the input of the first one block is performed, the relay machine needs the operation of relaying the content of the one block and the operation of copying the file to the external storage device as in the conventional case. With this copy, the line load increases as the file size increases, and the load on the relay machine also increases because the file is copied to the external storage device.

It is an object of the present invention to reduce the time required for this copying and to provide a means for accessing at higher speed.

[0005]

The means of the present invention will be described with reference to FIG. FIG. 1 is a block diagram of the present invention. 11
Is an access request source computer, 12 is a computer in which a file to be accessed exists, and 13 is a transmission line.

According to the present invention, when a certain block of the file 14 shown in FIG. 1 is desired to be input from the terminal device and the block does not exist in the storage device (memory) 15 of the access request source computer 11, the file to be accessed exists. The computer 12 sends the contents of the block to the terminal device 16 and at the same time stores the contents in the memory 15. When an input request is made to the block thereafter, the access processing is ended simply by returning the contents of the memory 15 to the terminal device 16. If there is an update request, the computer 12 in which the file 14 desired to be accessed exists according to the information transmitted from the access request source computer 11.
Update the block with 4. Then, the number representing the position of the block in the file is broadcast to all the computers on the network, and the computer receiving the block invalidates the memory contents when the block of the number is placed in the memory 15. To do.

[0007]

As described above, when there is an access request,
The data that flows on the line is only the contents of the target block. Therefore, the data of the unnecessary portion of the file does not flow on the line, which reduces the load on the line. Further, since the data is not copied to the external storage device but is simply stored in the memory, the input to the block can be realized by simply accessing the memory.

Further, when an update request is made, the file number is not copied and only the block number is reported, so the load on the line can be greatly reduced.

[0009]

EXAMPLES Examples of the present invention will be described below. 2 shows an example of a multi-computer system, in which a plurality of computers 21, 22, 23, ... Are connected via a transmission line 210 and external storage devices 24, 25, 26 for mutually storing data. .. exist in files A, B, C ... Further, storage devices 27, 28, 29 for temporarily storing the access contents of the files in the respective computers.
・ There exists.

FIG. 3 shows the contents of the storage devices 27, 28 and 29. The file management table has an OPS number, which is the number of the computer to which the file belongs, a device number in which the file is stored, a file name, and an address (address) of a buffer holding the contents of the file.
Consists of. The contents of the file are divided and held in block units, and the block management table stores the block numbers and
It consists of a block status indicating whether the contents of the block in the storage device are valid. The search for the block is immediately performed by using the hash method using the block number.

When accessing files of different computers, an access request is made through the transmission line 210,
The computer in which the file exists accesses the external storage device, and the access contents are sent back through the transmission path 210. The sent access contents are stored in the buffers in the storage devices 27, 28 and 29 corresponding to the hash value obtained from the block number. When the data stored in the buffer is accessed after being stored, the data can be obtained by simply transferring the data from the storage devices 27, 28, 29 in the own computer. On this occasion,
No data is transmitted or received through the transmission path 210 even though the file of another computer is accessed. Also, no access to the external storage device is required.

The access contents stored in the storage device of the own computer can be used only for the file input access. This is because in the case of output access, it is necessary to update the contents of the access target file existing in the external storage device of another computer. Further, when the file is updated, the content of the block stored in the storage device is the pre-update data, so that the transmission path 210
Requests (broadcasts) all computers to change the block status to a value indicating that the contents of the block are invalid. In order to make a request, an OPS number, a file name, and a block number to be invalidated are necessary, and they are notified via the transmission line 210. However, since the computer that has performed the output access stores the updated data in the storage device when the access is completed, the contents of this block are not invalidated.

According to the above processing method, when a block of a file of another computer that has been accessed once is input for the second time and thereafter, transmission / reception via a transmission line and access to an external storage device are not required. As a business that uses this function, when handling a file with a large number of blockings that requires frequent access to different records in the same block, you may create new data in the file and then search for that data. In this case, when it is necessary to access the same data in the file many times a day, it is considered that the speed-up of access is particularly effective.

[0014]

As described above, according to the present invention,
It is possible to bring about an effect of speeding up while normally accessing without any special processing such as copying a file to a repeater for speeding up of access and putting a load on the line.

Further, the processing for invalidating the block stored in the storage device at the time of updating the data does not impose a load on the line because the amount of transmitted data is small, and the existing autonomous distributed function broadcast can be used, so that the overhead is small. It has been processed.

[Brief description of drawings]

FIG. 1 is a block diagram of the present invention.

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

FIG. 3 is a diagram showing a management table of a storage device of the present invention.

[Explanation of symbols]

11 ... Access request source computer, 12 ... Computer in which file to be accessed exists, 13 ... Transmission line, 14 ... File, 15 ... Storage device, 16 ... Terminal device, 21, 22, 23 ... Computer, 24, 25, 26 ... External storage device, 27, 28, 29 ... Storage device, 210 ... Transmission line, 31 ... File management table, 32 ... OPS number, 33 ... Device number, 34 ... File name, 35 ... Buffer address, 36 ... Block management table, 37 ... Block number, 38 ... Block status, 39 ... Lock contents.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Shigeki Hirasawa 1099 Ozenji, Aso-ku, Kawasaki-shi, Kanagawa Stock Company Hitachi Systems Development Laboratory

Claims (2)

[Claims] 1. A distributed file system comprising a plurality of computers, wherein one computer accesses a file of another computer, and when a file of another computer is accessed on a memory of an access request source computer. The distribution is characterized in that when the user wants to access the same portion of the file next time, he / she can access the same part of the file by referring to the memory of the own computer without accessing the other computer. File access method. 2. When the distributed file is updated, all the computers on the network are notified of the update, and the contents of the updated part are recorded in the memory of the computer that receives the update. In this case, by invalidating the contents, the distributed file access method is characterized in that the same file can be updated at the same time from a computer on the network.