JPH0354625A - Cluster type magnetic disk device - Google Patents

Abstract

PURPOSE:To input/output plural files successively by satisfying fast transfer speed by requested by host equipment, and comprising the device of the minimum number of magnetic disk devices corresponding to the capacity of a data base to store data. CONSTITUTION:The device is comprised of (n) data memory devices 15 having the magnetic disk device 1, an input/output buffer 3 with the capacity of one cylinder of the magnetic disk device 1 connected to each of the memory devices, respectively, and a multi-disk controller 4 which controls them. Assuming data transfer speed from the magnetic disk device to the input/output buffer as tByte/sec, and the rotating speed of the magnetic disk device as Rrps, the device is comprised of the minimum number of data memory devices which satisfies conditions I and II as equation II in the case of being less than equation I when the minimum seek time of the magnetic disk device exceeds a time M/ Tsec to transfer the data of MByte of one input/output buffer to the host equipment. In such a way, it is possible to input/output the plural files large in memory capacity successively and at high speed in spite of the size of the file.

Description

[Detailed description of the invention] [Industrial application field]

The present invention relates to an external storage device, including a collective magnetic disk device with a large storage capacity and capable of writing and reading in a short time, a collective magnetic disk device suitable for continuous writing and reading of multiple files, and This invention relates to a data storage structure of a collective magnetic disk device.

[Conventional technology]

In recent years, secondary information such as literature information and patent information (bibliographic information)
Large-scale database services that include not only primary information (text) but also primary information (text) are becoming increasingly important. Conventionally, methods using keywords and classification codes have been used to search for information in such databases. However, this method only narrows down the results from a few dozen to a few hundred, and there is an efficiency problem in that the searcher must directly read the text to confirm the content at the final stage. Furthermore, since the classification system itself changes over time, the problem arises that keywords and classification codes must be constantly updated. Furthermore, since adding keywords (referred to as indexing) takes time, new documents are registered in batches in large quantities. Therefore, there has been a problem that registered information that can be searched always has a certain period of delay. As one method for dealing with these problems, a full-text search system is being considered that allows a searcher to directly refer to the text of a document and search for content based on free keywords. On the other hand, several character string search devices aiming at such a full-text search system have been proposed. A typical structure thereof is shown in FIG. 2, and its contents will be explained first. In the string search device 101, the search control means 102 controls the entire search device, receives search requests sent from the host computer, analyzes them, and sends the search to the string matching means 105 and the compound condition discrimination stage 104. Send as information. The search control means 1-02 also controls the disk control means 103 to send the character string data stored in the character string storage means 106 to the character string collation means. The character string matching means 105 checks whether there is anything in the input character data that matches the search request, and if there is any matching, the character string matching means 105 passes information identifying the character string to the complex condition determining means 1. Output to 04. The compound condition determining means 104 checks whether compound conditions such as mutual positional relationship specified in the search request are satisfied based on the character string identification information. If the compound conditions are satisfied, pointer information to the corresponding document and text data of the document content are returned to the host computer as search results. Character string search device 1. A magnetic disk device capable of storing a large amount of data is required as the character string storage means 106, which is a component of 0.01. General magnetic disk drives have the problem of not being able to input and output data at high speeds, and multi-head magnetic disk drives that can input and output data at high speeds are extremely expensive. Therefore, a collective type magnetic disk device has been considered in which a plurality of inexpensive general small magnetic disks are connected to increase the data input/output speed. One of them is the Japanese Patent Publication No. 6
0-1. There is an "Image data division storage device J" described in Publication No. 17326. This device has multiple magnetic disk devices, and has the same number of magnetic disk controllers, input/output buffers, and external devices as the magnetic disk devices. The master controller divides data input from an external device into pieces that are less than the capacity of the input/output bus sofa, and sequentially transfers the divided data to each magnetic disk controller. The magnetic disk controller writes to the corresponding magnetic disk device.The mask controller causes the magnetic disk controller of the magnetic disk device that is not writing to perform a seek operation, so that two of the plurality of magnetic disk devices that store data are stored. This is an attempt to apparently eliminate the seek time after the second one and shorten the time for writing and reading data. (Problem to be solved by the invention) An important element in the string storage means of a string search device. teeth,
Large storage capacity, ability to input and output multiple files continuously at high speed regardless of file size,
There were three points: low cost, and it was necessary to develop a collective magnetic disk device that satisfied these factors. Conventional technology attempts to shorten data writing and reading time by simply eliminating the access time of seek time, and it is difficult to reduce the number of magnetic disk drives needed to meet the data transfer speed required by external equipment. There was a problem in terms of cost performance as there was no consideration given to whether it should be used in the configuration. In addition, the conventional technology has the effect of reducing access time when a file with a large data size such as image data spans multiple magnetic disk drives, but when the data size does not span multiple magnetic disk drives, When writing or reading small files,
Unable to hide seek time, only one magnetic disk'
There was a problem that the access time was the same as jArl. In addition, the conventional technology continuously writes multiple files,
No consideration is given to the point of reading, and write and read commands from the host device can only be processed for one file, and when accessing multiple files, it is necessary to repeatedly perform one process. However, there is a problem in that the overhead time required for this becomes long. Further, as one of the overhead times, there is a process of searching storage location information of the magnetic disk device from a file identification code for specifying a file to be accessed from a host device. In conventional general magnetic disk drives, the file identification code is expressed as a file name that is a string of character codes such as ASCII code.
This file name requires searching the file management information stored in the file management information area of the magnetic disk device to find the physical storage location, which poses a problem in that the processing time required is large. SUMMARY OF THE INVENTION An object of the present invention is to provide an inexpensive collective magnetic disk device that has a large storage capacity and can input and output multiple files continuously at high speed regardless of the file size. [Means for Solving the Invention] In order to achieve the above object, a collective magnetic disk device employing the following means is provided with a plurality of data storage devices each having a magnetic disk device, and data to be input/output to the data storage device. It consists of an input/output buffer that temporarily stores data, and a multi-disk controller that controls the data storage device and the input/output buffer. Furthermore, the data storage device is configured by l magnetic disk devices each having a magnetic disk controller, or
Alternatively, it is configured by a plurality of magnetic disk devices each having a magnetic disk controller and a multiplexer for selecting the magnetic disk devices. Furthermore, the input/output buffer has a capacity equivalent to at least one cylinder of a magnetic disk device for each of the data storage devices, and is constituted by semiconductor memory on one or two sides. Note that the memory can also be realized using a high-speed storage element such as an optical memory other than a semiconductor storage element. The multi-disk controller, which controls data storage devices and input/output buffers, consists of a communication memory using semiconductor storage elements that stores requests from higher-level devices, a multiplex controller that controls data transfer, and a multi-disk controller that controls data transfer. It consists of a physical information table using semiconductor memory elements to search for physical storage locations, and a master controller to control them. Note that the communication memory and physical information table can also be realized using high-speed storage elements such as optical memory other than semiconductor storage elements. Note that the mask controller uses a microcomputer to control each component. Furthermore, the multi-disk controller uses the logical classification, which is an identification code unique to the logical classification of files that have been logically classified into hierarchical groups, as well as a unique number within the logical classification, as a file identifier. In this case, the file ID is used. In addition, in a multi-disk controller, the file ID
The mask controller has a structure definition table in its memory that stores management information for determining the physical storage location of a file in the magnetic disk device according to the logical classification ID in the file. The multiplex controller controls data transfer between the host device and the input/output buffer, and includes a multiplexer that selects the data bus for the input/output buffer, a DMA controller that transfers data without the intervention of a mask controller, and a DMA controller that controls data transfer between the host device and the input/output buffer. It consists of a start address registration table that stores the start address of the input/output puffer in the range, and an end address registration table that stores the end address of the range.

[Effect]

The operation of the above technical means will be described below.6 When there are n data storage devices and the transfer data of the above magnetic disk devices in the data storage devices does not span between tracks and does not perform a seek operation, the transfer data is transferred from the magnetic disk device to the input/output buffer. The data transfer rate is [Byte/s]
ec], the capacity of one cylinder of the disk device is M [B
ytel, the minimum seek time of the magnetic disk device is s [
seel. The rotational speed of the magnetic disk device is R [rp
s], the output buffer capacity is the same as the capacity of one cylinder of the magnetic disk device M [Byte], the data transfer rate from the collective magnetic disk device to the host device T [Byte/sec is as follows. conditions must be met. Minimum seek time s of magnetic disk device [secl is 1
The time (M/T) to transfer M [Bytes] of data from the above input/output buffers to the host device [If larger than secl, the data transfer time from the data storage device to the output buffer 8 is the minimum seek time of the magnetic disk device s [s
ecl and the maximum rotational waiting time of the magnetic disk device (1
/ R ) [see] and the data transfer time from the data storage device to the input/output buffer (M/t) [se
c], which is the time (nM/T) to transfer the data of all input/output buffers to the upper device [se
c]. This can be expressed as a mathematical formula, and the number n of data storage devices can be expressed as in the following formula. In addition, the minimum seek time of the magnetic disk device is 5 [sec
] is less than the time (M/T) [seC] for transferring 1 Byte of data to the host device of one input/output buffer, even if the magnetic disk device finishes the seek operation, the magnetic disk device Data cannot be transferred from the data storage device to the input/output buffer because the input/output buffer that is attempting to transfer data is currently transferring data to the device mentioned above. Therefore, it is necessary to wait until the data transfer from the input/output buffer to the higher-level device is completed. Therefore, the data transfer time from the data storage device to the input/output buffer is the data transfer time from one input/output buffer to the host device (M/T) [secl] and the maximum rotational waiting time (1/R) of the magnetic disk device. ) [sec] and the data transfer time from the data storage device to the input/output buffer (M/
t) [The total time of secl is the time to transfer data of all input/output buffers to the upper device (nM/T)
It only needs to operate within [sec]. This can be expressed as a mathematical formula, and the number n of data storage devices can be expressed as in the following formula. By constructing a collective magnetic disk device with the minimum number of data storage devices that satisfy these conditional expressions, it is possible to provide a magnetic disk IT with good cost performance that satisfies the data transfer speed required by host equipment. . The data storage device stores data files. By configuring the data storage device as a magnetic disk device having a magnetic disk controller, the magnetic disk controller controls writing and reading data to and from the magnetic disk, and the processing of the multi-disk controller is reduced. In addition, the data storage device can be a plurality of magnetic disk devices,
The storage capacity can be increased by configuring a multiplexer that selectively connects the data path of the magnetic disk device to the data bus of the input/output buffer. The input/output buffer temporarily stores data to be input/output to the data storage device. In the case of writing, data is transferred one after another from the host device to the input/output buffer at a speed faster than the writing speed of the magnetic disk device in the data storage device, and after the data transfer is completed, the input/output buffer is transferred to the magnetic disk device. Data is written at the writing speed of the device. In the case of reading, each magnetic disk device reads data to the input/output buffer at the read speed of the magnetic disk device, and the input/output buffer after reading is transferred to the upper level at a speed faster than the read speed of the magnetic disk device. Transfers data to the device. Thereby, data can be input/output to the host device at a faster speed than the writing/reading speed of the magnetic disk device. Furthermore, by having two input/output buffers for each data storage device, while the input/output buffer on the first side is transferring data with the host device, the input/output buffer on the second side is connected to the data storage device. You can write to and read from. This can reduce the time the magnetic disk device waits for the data transfer operation until the data transfer with the host device is completed, and writing and reading can be performed in a short time. The conditional expression for providing a magnetic disk device with good cost performance that satisfies the data transfer speed required by the host device at this time is expressed by the first expression. The multi-disk controller controls data storage devices and input/output buffers in response to data file write and read requests from higher-level devices. A communication memory using a semiconductor memory element that can store multiple file IDs of files to be written or read is
Overhead time in receiving commands from host devices and processing completion reports is reduced, allowing continuous writing and reading of data files in a short time. A physical information table using a semiconductor memory element that can be accessed in a short time can quickly determine the physical storage location of a magnetic disk device from a logical file ID, and therefore reduces the overhead required to read data files. Time becomes short. In addition, whereas files stored in magnetic disk devices have traditionally been identified by file names consisting of variable-length character code strings, file IDs are made up of fixed-length numerical codes. , can be expressed using small-sized code, simplifying the process of specifying data files to be written or read and searching for physical storage locations, and reducing the overhead time required. Furthermore, when storing data files, by physically storing logically related files close to each other, seek time and access time can be shortened. A multiplexer in the multiplex controller selects the data path of the input/output buffer. The start address registration table and end address registration table store several start addresses and end addresses that specify the range where the necessary data is stored among the data stored in the input/output buffer. The DMA controller transfers input/output buffer data in the range specified by the start address registration table and the end address registration table to the host device at high speed without the intervention of a mask controller. If there are multiple files to be read on the same cylinder of the magnetic disk device, set the size of the file to be read to f.
1 [Byte v f 2 [Byte], the size of the file that does not need to be read during that time is k [Byte]
] , the read speed of the magnetic disk device [B
yte/sec]. When the rotational speed of the magnetic disk device is R [rps] and the average seek time of the magnetic disk device is S [sec], the average rotational waiting time is (1
/2R) [sec], and the condition that the time to read at a time is shorter than the time to read one by one can be expressed as follows. This formula can be easily written as the following formula. f When this conditional expression is satisfied, the multiplex controller temporarily reads files that do not need to be read into the input/output buffer, and transfers only the necessary portions, excluding unnecessary file portions, when transferring the files to the host device. This allows the magnetic disk to read a plurality of files in one read process, thereby shortening the access time that occurs in the read process.

[Embodiment 1] Embodiment 1, which is an embodiment in which the present invention is applied to a character string search device, will be described below. FIG. 3 shows the configuration of a collective magnetic disk device using the present invention, which includes n data storage devices 15 each having a magnetic disk device 1, and one of the magnetic disk devices 1 connected to each of the data storage devices 15. It consists of an input/output buffer 3 having a capacity equivalent to a cylinder, and a multi-disk controller 4 that controls the data storage device 115 and the input/output buffer 3. Here, the data storage device W15 consists of a device and a plurality of magnetic disk drives 1, and the input/output buffer 3 consists of one memory surface of the magnetic disk drive 1 having a capacity of one cylinder. The multi-disk controller 4 includes a communication memory 5 that can directly set the file ID of the file to be accessed from the host device 7, a multiplex controller 8 that controls the high-speed data bus 10, and a physical storage destination of the magnetic disk device based on the file ID. It consists of a physical information table 6, which is a conversion table for obtaining information, and a mask controller 9, which controls them. The host device 7 consists of a host controller that gives commands to the collective magnetic disk device, and a character string search device that detects a specified character string from input data and outputs the detected information. Before constructing a database constituting data files in this collective magnetic disk device, a database structure definition process is performed. In this collective magnetic disk device, as a means of arranging logically related files so that their physical storage locations are close to each other, we first assign physical cylinders to logical classification IDs with a hierarchical structure.
It is allocated according to the following. When accessing multiple files at once, logically related files are often targeted. Therefore, by placing the storage locations close together, the distance that the magnetic head moves between the cylinders of the magnetic disk drive is shortened, and the seek time, which is a part of the access time, is shortened. In order to allocate physical cylinders according to logical classification IDs that have a hierarchical structure, the higher-level device 7 stores database structure definition information in the communication memory 5, which is composed of a set of logical classification IDs and storage capacity required by the file classification. After storing, a database structure definition command is issued to the multi-disk controller 4. Upon receiving the structure definition command, the mask controller 9 in the multi-disk controller 4 stores in the memory in the mask controller 9 how the physical position corresponds to the logical classification based on the structure definition information of the database set in the communication memory 5. A structure definition table having the structure shown in FIG. 4 above is created. Figure 4 is an example of a two-layer structure with two categories in each layer.The entire magnetic disk drive is combined into one magnetic disk drive, and the storage location for each category is determined by the cylinder position, and the storage capacity is determined by the cylinder position. It is shown in numbers. In addition, in the database structure definition process, the mask controller 9 in the multi-disk controller 4 stores the physical location of the storage destination of the file to be written for each logical classification on the memory in the mask controller 9 as shown in FIG. Create a storage location pointer table that indicates the physical location of the storage location of the file to be written, as shown. When the structure definition is completed, the storage position pointer table indicates the first cylinder, first track, first sector, and first position within the sector of each logical classification set in the structure definition. In FIG. 5, storage position pointer information when files are stored according to the classification shown in FIG. 4 is stored. Next, the construction of the database will be explained. This collective magnetic disk device creates management information using file IDs as a means of specifying files to be accessed using file IDs (consisting of a logical classification ID and a unique number within the logical classification). . After storing, in the communication memory 5, file information consisting of a plurality of sets of file IDs and file sizes of files to be written, the host device 7 issues a write command to the multi-disk controller 4. The multi-disk controller 4 that has received the write command executes processing according to the flow shown in FIG. The mask controller 9 in the multi-disk controller 4 reads the file ID in the file information from the communication memory 5, and reads the storage location where the file indicated by the file ID is to be stored from the storage location pointer table. Once the storage location is determined, the remaining capacity that can be written to that physical cylinder is determined. If the file size given by the file size of the file information is smaller than the remaining capacity, the file size is stored in the physical information table 6 whose entry is the file ID as shown in FIG. ! Write the (disk number, cylinder number, track number, sector number, position within the sector), file size, and number of disks. The number of disk spans indicates how many magnetic disk devices 1 the file spans, and the file being processed could not be written to one cylinder of the device or multiple magnetic disk devices. In this case, the file will be divided and the remaining files will be written to the next disc. If this file is an unwritten file after being divided, this value is counted up. Entries in the physical information table 6 are indicated by file IDs given in file information. After writing to the physical information table, advance the storage location pointer by the file size. If the file size and remaining capacity are equal, when the cylinders of l magnetic disk devices 1 are full, the write process to that magnetic disk device 1 is performed. If the file size is larger than the remaining capacity, compare the magnetic capacity and the division standard size. The division standard size is a value that is set in the structure definition process. Even though the remaining capacity of the cylinder is very small, if a file is stored across magnetic disk units 1, it will take two drives to read the file. The magnetic disk device 1 must be controlled, and the processing overhead increases. Therefore, a certain standard is set, and if the remaining capacity is smaller than the standard value, data is written from the beginning of the cylinder of the next magnetic disk device fll. If the remaining capacity is equal to or larger than the division reference size, the storage location and file size are stored in the physical information table 6, and then the files are divided into files that can be written in the remaining capacity and files that are left unwritten. The storage physical location and file size are written in the physical information table 6. The magnetic disk device 1 that has created the physical information that fills the l cylinder performs a write process. The unwritten file returns through the loop and becomes the next file to be processed. If the remaining capacity is smaller than the division standard size, advance the storage position pointer table to the beginning of the next cylinder, then return to the loop and continue processing with the file to be processed as the next file to be processed. At this time, the magnetic disk device that has created the physical information that fills one cylinder performs a write process. In the write process, the mask controller 9 issues a seek command to the magnetic disk device 1 and starts a seek operation.
Next, a file transfer request is issued to the higher-level device 7, and the mask controller 9 requests the higher-level device 7 to transfer the file, and also controls the multiplex controller 8 to switch the data path and process the transferred file. The file is transferred to the input/output buffer 3 specified by the physical information. When the seek operation is completed and the file transfer is completed, the master controller 9 issues a write command to the magnetic disk device 1, and the magnetic disk device 1 executes the write operation. The above operations are repeated to construct the database. FIG. 8 shows the temporal relationship of write processing, from the host device 7 to 1-1, 2-1, . . . as shown in the figure.
... H 1.1-2+ 2 2t... The data that is transferred one after another is sent to the input/output buffers 3-1.3-2, . . . by the multiplex controller 8 in the multi-disk controller 4. ...3-n,
3-1. 3-2, . . . At this time, for example, the magnetic disk device 1-1 transfers data 1-
1, the seek is started by a command from the mask controller 9. When the data transfer 1-1 is completed, the mask controller 9 transfers the data to the magnetic disk device 1-1.
Issue a write command to. After the magnetic disk device 1-1 waits for rotation until it reaches a designated write position, it begins writing data 1-1 in the input/output buffer 3-1 to a predetermined cylinder, track, or sector. During this time, other magnetic disk devices will also perform similar processing as shown in the figure. As is clear from Figure 8 and the above explanation,
Each magnetic disk device can write files in parallel and continuously, making it possible to build a database in a short time. Next, file read processing will be explained. In addition, when there are multiple files to be read on the same cylinder of the same magnetic disk device, the files that do not need to be read between the files to be read are also read into the input/output buffer, and the files that do not need to be read are saved when transferring to the host device. The method for deleting will be explained. After storing file information consisting of a plurality of file IDs of files to be read in the communication memory 5, the host device 7 issues a read command to the multi-disk controller 4. The multi-disk controller 4 that receives the read command executes processing according to the flow shown in FIG. The mask controller 9 in the multi-disk controller 4 determines the file ID of the file to be read first from the communication memory 5.
is read out, and the physical information table 6 is searched for the physical information in which the file is stored based on the file ID. Let this file be the destination file and the physical information be the physical information of the destination file. Next, the file ID of the file to be read next is read from the communication memory 5, and the file ID
The physical information table 6 is searched for the physical information in which the file is stored. After this file,
Set the physical information as the physical information of the subsequent file. Check whether the first file and the next file exist in the same cylinder from the obtained physical information, check whether they exist in the same cylinder, and if they exist in the same cylinder, there is no need to read unspecified files between the first file and the next file. Find out if there is a group of files, and if so, find the total size of the group of files. When the size of one file that does not need to be read is small, the physical information is combined so that the first file and the next file can be read with a single read command. Next, the process returns to the loop using the synthesized physical information as the physical information of the previous file, reads the next file ID from the communication memory 5, and performs similar processing using that file as the next file. If the destination file and subsequent file do not exist in the same cylinder, or if the size of the file that does not need to be read is large,
Executes the process of reading the destination file from the magnetic disk device. The physical information of the next file returns through the loop as the physical information of the previous file, and is transferred from the communication memory 5 to the next file I.
D is read and the same process is performed using it as a subsequent file. This operation is repeated until all specified files are read. The process of reading the destination file from the magnetic disk device is as follows:
First, the mask controller 9 issues a seek command to the magnetic disk controller 2-i of the magnetic disk device 1-i indicated by the physical information of the previous file to move the magnetic head to the physical position indicated by the physical information, and 1
- i starts a seek operation. When the seek operation is completed, if the input/output buffer 3-i is in a state where data can be written, the mask controller 9 issues a read command to the magnetic disk controller 2-i, and writes the input/output buffer 3-i to the magnetic disk device. Start storing the file read from 1-i. When the storage is completed, the mask controller 9 controls the multiplex controller 8 to start transferring data from the input/output buffer 3-i to the host device 7. As shown in FIG. 10, the multiplex controller 8 connects a multiplexer 201 that selects and connects the data paths of the input/output buffers 3-1 to 3-n to the data path of the host device 7, and the selected i-th input/output buffer. A DMA controller 202 that outputs data from 3-i to the host device 7 without the intervention of the mask controller 9, and a start address and an end address for specifying the transfer range of the input/output buffer 3-i are stored in the DMA controller 202. It consists of a start address registration table 203 and an end address registration table 204. The mask controller 9 sets the start address where the file to be transferred of the crowd buffer 3-i exists in the start address registration table 203 and the end address in the end address registration table 204, and then transfers the data from other input/output buffers 3 to the host device. If data has not been transferred to DMA controller 202, a start command is issued to DMA controller 202. DM
The A controller 202 has the start address registration table 20
3 and the end address registration table 204, only the data in the specified range is transferred at the transfer speed requested by the host device 7 without the intervention of the mask controller 9. When the physical information is combined and read into the input/output buffer 3-i so that the first file and the next file can be read with a single read command, the first address registration table 203 and the end address registration table 204 are required. Set multiple addresses so that all files are transferred, and perform the same process. In order to read the first file and the next file with a single read command, the process of combining physical information is performed when the following conditions are satisfied. The size of the first file is f 1 [Byte], the size of the next file is f 2 [Byte] + the total size of the file group that does not need to be read is k [Byte], and the seek operation from the magnetic disk device 1 to the input/output buffer 3 is The effective transfer rate does not include [Byte/se
e], and the rotational speed is R [rps]. When the average seek time is s [sec], the average rotation waiting time is (1/
2R), and the condition that the time to read at a time is shorter than the time to read one by one is as follows, and can be written as shown in the third equation. t k ≦ (3) The temporal relationship of the 2R file read processing is as follows:
The transfer rate required by
e], the transfer rate from each magnetic disk device 1 to the input/output buffer 3 is [Byte/sec]. If the minimum seek time of each magnetic disk device l is s [sec] and the rotation speed is R [rps], then the minimum seek time s [s]
ec] is longer than the time (M/T) for transferring the file on the i-th input/output buffer 3-i to the higher-level device 7, as shown in FIG. 11. In order to satisfy the transfer speed required by the host device 7, the time required for the i-th magnetic disk unit 11-i to read the file to the input/output buffer 3-i (s + 1 / R + M /
t) is within the time (nM/T) for transferring all the files on the input/output buffers 3 to the host device 7. Here, the seek time was set as the minimum seek time in order to read consecutive cylinders. Also,
Regardless of the position of the magnetic head at the time a read command is issued to the magnetic disk unit W1, the rotation waiting time is set to the maximum value (1/R) so as to satisfy the transfer speed required by the host device 7. did. This relationship can be expressed mathematically as shown in the first equation. Furthermore, the temporal relationship of file read processing when the minimum seek time 5 [sec] is less than the time (M/T) for transferring the file on the i-th input/output buffer 3-i to the host device 7 is as follows: The result is as shown in FIG. In this case, even after the seek operation is completed, the input/output buffer 3-i is still transferring the file to the host device 7, so a read command cannot be issued to the i-th magnetic disk device 1i1-i. ．． Therefore, when the transfer of the file in the input/output buffer 3-i to the host device 7 is completed, a read command is issued to the i-th magnetic disk device 1-i. Therefore, in order to satisfy the transfer speed required by the host device 7, the time required for the i-th magnetic disk device M1-i to read the file to the input/output buffer 3-i (M/T
+ 1 /R +M/t) is the time (nM/t) to transfer the files on all input/output buffers 3 to the host device
It is good if it is within T). This relationship can be expressed mathematically as shown in the second equation. It is possible to determine how many magnetic disk drives 1 should be combined in order to satisfy the tMR speed, and to construct a collective magnetic disk drive with the minimum number of magnetic disk drives 1 that satisfy the first formula. It is the most cost effective option. For example, the capacity of one truck is 20k (kg) [B
The magnetic disk device 1 is composed of 6 tracks of 120K [Bytel] and has a capacity of 1 cylinder, and the transfer rate required by the host device 7 is 2M (mega) [Byte/sec]. The effective transfer rate from each magnetic disk device 1 to the input/output buffer 3, excluding seek operations, is I M [Byte/se
c], the minimum seek time of each magnetic disk device 1 is 10 m
(millimeter) If the rotational speed is 50 [rps], the first equation becomes as follows, and the minimum n that satisfies this equation is 4. Figure 13 shows the time relationship during reading from a collective magnetic disk drive configured with three magnetic disk units W1, and Figure 14 shows the time relationship during readout from a collective magnetic disk drive consisting of four magnetic disk units W1. Time Relationship: FIG. 15 shows the time relationship during reading in a collective magnetic disk device composed of five magnetic disk devices 1. When the three magnetic disk devices 1 shown in FIG. A time a occurs during which data cannot be transferred from the input/output buffer 3 to the higher-level device 7 because the transfer time is not met, and the transfer speed from the input-output buffer 3 to the higher-level device 7 is approximately 1. 6 M[Byte/see], and the transfer speed required by the upper-level device cannot be satisfied. Furthermore, when configured with the five magnetic disk devices 1 shown in FIG. 15, although the transfer speed required by the host device 7 is satisfied, the four magnetic disk devices i! shown in FIG. Compared to the case where the main magnetic disk unit N1 does not perform processing, the time b is longer when the main magnetic disk unit N1 does not perform processing, and the usage efficiency of the magnetic disk unit is poor. Therefore, it can be said that the case where the four magnetic disk drives 1 that meet the minimum value n that satisfies the first equation are used is the most cost-effective collective magnetic disk drive. Embodiment 2, which is another embodiment in which the present invention is applied to a character string search device, will be explained using FIG. When reading only a specified file, the collective magnetic disk device described in Embodiment 1 can read only a specified file, as long as the specified file exists on average in magnetic disk devices W1-1 to 1-n. The host device 7
Eight data transfer speeds can be increased. However, if the specified file exists only in the device and the plurality of magnetic disk devices 1-i, reading from the device and the plurality of magnetic disk devices 1-i will be performed continuously. In this case, data transfer from the host device 78 must be performed in two stages, in which data is read from the magnetic disk device 1-i to the input/output buffer 3-i and then transferred from the input/output buffer 3-i to the host device 7. As a result, a situation occurs in which data transfer is degraded. In this way, if the designated files exist unevenly in the magnetic disk device 1, a situation may arise in which the data transfer speed of the host device 78 cannot be effectively increased. Therefore, in the second embodiment, by preventing files from being stored unbalanced, all the magnetic disk devices 1 are always operated for reading, and the data transfer speed to the host device 7 is increased. Furthermore, in this embodiment, the number of magnetic disk devices is increased in order to further increase the storage capacity. FIG. 1 shows the configuration of a collective magnetic disk device using the present invention, and the difference from FIG. 3 is that the magnetic disk device 1
It has two sides of input/output buffer 3 with the same capacity as one cylinder of
The file read from the magnetic disk device 1 can be stored in the input/output buffer 3b on the second side while the file is being transferred to the magnetic disk device 1. In addition, one data storage device M15 is configured by m magnetic disk devices 1i1-i-1 to 1-i-m and a multiplexer 14, and the total storage capacity of the collective magnetic disk device is divided by the devices and multiple devices. It is (nXm) times the storage capacity of the magnetic disk device. To explain the operation, first, similar to the first embodiment, the database structure definition process is performed, but information identifying the m magnetic disk drives N1 connected to the input/output buffer 3 via the multiplexer 14 is added to the structure definition information. to add. The database is constructed in the same manner as in Example 1, but there are some differences. The difference from the first embodiment is that the file given by the file information is divided into the number of magnetic disk drives to be monitored, and is distributed and stored in all the magnetic disk drives. In addition, the data of the input/output buffer 3 is stored in the m devices given by the physical information and the plurality of magnetic disk devices 1.
-i-j to control and store multiplexer ~l4. The file division method is to calculate the division size by dividing the file size by the number of devices, and from the beginning of the file to the device for each division size and to the multiple magnetic disk devices 1-1-
In some cases, data is stored in order from j to 1-2-j, then 1-3-j, and in other cases, data is stored in a device of a predetermined size, such as 1-2-j, 1-3-j, etc. from the beginning of the file. There are disk devices that store data in order from disk devices 1-1-j to 1-2-j and 1-3-j. If the file size is not divisible by the number of magnetic disk units, invalid data is added to the end so that the file size is a multiple of the number of magnetic disks. Store the file so that the beginning of the file is located at . Next, file reading will be explained. This is also done in the same way as in the first embodiment, but in this configuration, the input/output buffer 3 is
3a and 3b, it is possible to start the process of reading the next file at the time when the file read from each magnetic disk device l is stored in the input/output buffer 3. The time relationship of file read processing is as shown in Figure 6 above, and compared to the first embodiment, there is no waiting time until the state is ready for writing data to the input/output buffer 3, and faster transfer is possible. It becomes possible. The relationship that satisfies the transfer speed required by the host device 7 under the same conditions as in Example 1 is as follows:
device and multiple magnetic disk drives ill-i-
The time to read a file from j to one input/output buffer 3a-i of two input/output buffers 3-i (s +
1/R + M/t) is within the time (n M./T) for transferring files from all other input/output buffers 3b-1 to 3b-n to the host device 7, and this is expressed as a mathematical formula, and this mathematical formula can be easily written as the following formula. Based on this condition, it is possible to determine the number of data storage devices 15 to satisfy the transfer speed required by the host device, as in the first embodiment. Furthermore, when a large storage capacity is required, the data storage device 15 can be configured with m magnetic disk devices and the multiplexer 14, and the storage capacity can be multiplied by m. If a collective magnetic disk device is constructed using the minimum number of magnetic disk devices l, it will have the best cost performance. In the embodiment shown in FIG. 16, the seek operation of each magnetic disk device is started at the time when the data transfer from the input/output buffers 3-1 to 3-n to the host device is completed, but each time the read operation is completed. It is clear that you can do this with Although the above two embodiments have been described using a magnetic disk device, it is clear that the same applies to a storage device in which a storage medium rotates, such as an optical disk device other than a magnetic disk device. [Effects of the Invention] According to the present invention, it is possible to satisfy the high transfer speed required by host equipment, and to configure a collective magnetic disk device with a minimum number of magnetic disk devices according to the capacity of the database to be stored. This has the effect of providing an inexpensive collective magnetic disk that can input and output multiple files continuously at high speed regardless of the file size.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing a second embodiment, which is an embodiment in which the present invention is applied to a character string search device. FIG. 2 is a block diagram showing an example of a string search device. FIG. FIG. 4 is a diagram showing the structure of a structure definition table, FIG. 5 is a diagram showing the structure of a storage position pointer table, and FIG. 7 is a flowchart of the file writing process in the first embodiment. FIG. 8 is a time chart of the file writing process in the collective magnetic disk device of the first embodiment. FIG. 9 is a flowchart of file read processing in Embodiment 1, 10 [i1 is a diagram showing the structure of the multiplex controller, and FIG. 11 is a time chart of file read processing in the collective magnetic disk device of Embodiment 1. Chart diagram, No. 12@ is a time chart diagram of file read processing in the collective magnetic disk device of Example I, and FIG.
A time chart diagram of file read processing in a collective magnetic disk device configured with one magnetic disk device,
FIG. 14 is a time chart of file read processing in a collective magnetic disk device configured with four magnetic disk devices in Example 1, and FIG. 15 is a time chart of file read processing in a collective magnetic disk device configured with five magnetic disk devices in Example 1. FIG. 16 is a time chart of file read processing in the collective magnetic disk device of the second embodiment. Explanation of symbols 1...Magnetic disk device 2...Magnetic disk controller 3...I/O buffer 4...Multi-disk controller 5...Communication memory 6...Physical information table 7...Upper device 8...Multiplex controller 9...Mask controller 10.12...Data bus 11.13...Control bus 14...Multiplexer 15...Data storage device 6...Collective magnetic disk Device 101...Character string search device 102...Search control means 103...Disk control means 104...Compound condition determination means 105...Character matching means 106...Character string storage means 107-111. ...Control means 201...Multiplexer 202...DMA controller 203...Start address registration table 204...End address registration table 13 11th level snoring round 1 poem praise L・j beak defense 1st level chido diagram 5 company'i-v-m ki. +3 garden calculation? φ garden 7-t de figure

Claims (1)

[Claims] 1. A plurality of data storage devices each having a magnetic disk device, an input/output buffer for temporarily storing data to be input/output to the data storage device, and control of the data storage device and the input/output buffer. In a magnetic disk system that has a collective magnetic disk device consisting of a multi-disk controller that performs magnetic disk device 1
The capacity of the cylinder is M [Bytes], and the data transfer rate from the data storage device to the input/output buffer is t [B
yte/sec], the minimum seek time of the magnetic disk device is s [sec], the rotational speed of the magnetic disk device is R [rps], and the capacity of the input/output buffer is the capacity M of the magnetic disk device for the cylinder. [Byte], when the minimum seek time s [sec] of the magnetic disk device is the time (M
/T)[sec] when n≧T{1/t+1/M[s+(1/R)]} Also, the minimum seek time s[sec] of the magnetic disk device is longer than the data M[sec] of the input/output buffer. When the time (M/T) [sec] for transferring [Byte] to the above-mentioned host device is less than n≧1+T(1/t+1/MR)}. type magnetic disk device. 2. In the collective magnetic disk device according to claim 1,
A collective type characterized in that the data storage device is constituted by a plurality of magnetic disk devices each having a magnetic disk controller and a multiplexer that selects one of the plurality of magnetic disk devices. Magnetic disk device. 3. In the collective magnetic disk device according to claim 1,
2 of the above input/output buffers per data storage device.
While the data in the input/output buffer on the first side is being transferred to the host device, the data read from the data storage device is stored in the input/output buffer on the second side, and n≧T{1/t+1/M[s+(1/K)]} is satisfied when the seek time s[sec] is less than or equal to the time to transfer data M[Byte] of the input/output buffer to the higher-level device. A collective magnetic disk device characterized by having n. 4. In the collective magnetic disk device according to claim 1,
A collective magnetic disk device characterized in that the multi-disk controller includes communication memory means for connecting the host device and the multi-disk controller. 5. In the collective magnetic disk device according to claim 4,
A collective magnetic disk device comprising a semiconductor storage element as the communication memory means. 6. The collective magnetic disk device according to claim 1,
A physical controller that interprets a file ID consisting of a logical classification, which is a unique identification code of a logical classification transferred from a higher-level device, and a number unique to a file within the logical classification, and makes it correspond to the physical location of the magnetic disk device. A collective magnetic disk device comprising information table means in the multi-disk controller. 7. The collective magnetic disk device according to claim 6,
A collective magnetic disk device comprising a semiconductor storage element as the physical information table means. 8. The collective magnetic disk device according to claim 1,
When there are multiple files to be read on the same cylinder of the above magnetic disk device, the total capacity of the group of files that will not be read between the first file to be read and the next file to be read is set in [Bytes], and the total capacity of the files to be read from the above data storage device is If the transfer rate to the input/output buffer is [Byte/sec] and the rotation speed of the magnetic disk device is R [rps], then if k≦t/2R is satisfied [Byte], read first. Files that do not need to be read between the file and the next file to be read are also read into the input/output buffer, and when transferred from the input/output buffer to the host device, the unnecessary files are removed. A collective magnetic disk device characterized by having a multi-disk controller having means for transferring data.