JP3832485B2 - Computer system and secondary storage device - Google Patents

Description

  The present invention relates to a computer system composed of a host computer and a secondary storage device, and more particularly to a technique for optimal file placement in a secondary storage device.

  As a computer system composed of a host computer and a secondary storage device, a hard disk device (hereinafter referred to as “WS”) that employs an operating system (hereinafter referred to as “OS”) known as UNIX is used. Hereinafter, a system in which a “disk device” is connected is known.

  As a file management technique in such a computer system, a technique described in Non-Patent Document 1 is known.

  Hereinafter, this technique will be described.

  FIG. 24 shows the configuration of this computer system.

  In the figure, 1 is a host computer and 2 is a secondary storage device.

  In the host computer 1, 20 represents an OS, and 12 represents an application program (hereinafter referred to as “AP”) executed by a user managed by the OS 20. In other words, the application program executed in the host computer is managed as the process 12 by the OS 20.

  In the OS 20, 13 is a file management / buffer management means for managing a buffer used for file management and data transfer at the time of file access, 15 is a directory or file logical address (address inside the OS), a disk A file management table describing information for associating with a local address (local address) in the apparatus, and 14 describing directory information (information for associating directory and file names with the file management table) The directory management table is shown. Also, 16 is a device driver that converts and controls a file access request from the file management / buffer management means 13 according to the physical characteristics of external devices such as various secondary storage devices, and 30 newly writes a file. File address determining means for determining the local address of the disk device most suitable for writing, 25a a disk management table for managing the disk usage status, and 21 and 22 communication and data between the host computer and the secondary storage device 2 shows interface control means for controlling transfer.

  Here, the logical address is a logical address for managing files in the host computer. The local address (local address in the disk device) is a logical address that is a local address in the disk device for managing data by the disk device (secondary storage device). In some cases, it may be a physical address. For example, in a disk device (SCSI disk) conforming to SCSI (Snall Computer Interface), a logical serial number (LBA: Logical Block Address) is used as a local address. Further, in an IDE (Intelligent Device Electronics) disk device or the like, a physical address that directly represents a physical storage position, such as a head number, a cylinder number, or a sector number, is used as a local address.

  Next, in the secondary storage device 2, 29 is one or more disk devices, 24 is a disk device control means for controlling the disk device 29, 23 receives and analyzes the read and write commands transmitted from the host computer, A command management means for creating a disk command specific to the disk device 29 and sending it to the disk device control means 24 is shown.

  Hereinafter, file access processing executed when the process 12 accesses an existing file will be described.

  When the process 12 performs file access, it issues a file access read and write request (system call) to the OS 20.

  The file management / buffer management means 13 of the OS 20 receives this request and refers to the directory management table 14 to obtain the position of the file management table from the file name. Next, referring to the obtained file management table 15, a logical address where process read (or write) request data exists (or data is registered) is calculated and converted to a local address. In addition, a buffer area necessary for data transfer is secured.

  Next, the OS 20 makes an access request to the device driver 16 using the calculated local address. In response to this access request, the device driver 16 creates a command in a form that matches the physical characteristics of the secondary storage device 2 in which the file is stored (or stores the file). The data is sent to the next storage device 2.

  In the secondary storage device 2, the command management means 23 receives a command via the interface control means 22 and analyzes this command. Based on the analysis result of the command such as the access type, the access start address, and the data length, a disk command unique to the disk device 29 is created, and the disk control device 24 is requested to execute the command. The disk control device 24 controls the disk device 29 based on this command and executes an appropriate data transfer process.

  When the process 12 accesses an existing file, it is necessary to determine at which address of the secondary storage device 2 the file is stored.

  Therefore, when a new file is created, the following new registration process must be performed prior to the file access (write) process.

  That is, when a request for writing a new file is received from the process 12, the file management / buffer management means 13 registers a file name in the directory management table 14, and sets a management data area for this file in the file management table 15. And the location is registered in the directory management table 14. Next, the file address determination unit 30 refers to the disk management table 25a, and determines the optimum address based on physical characteristic parameters such as the number of heads, the number of cylinders, and the number of sectors of the disk device set in advance by the user. A local address of an area that does not seek as much as possible and avoids unnecessary rotation waiting is determined by a predetermined arrangement algorithm. Then, the determined local address is registered in the file management table and associated with the logical address.

  With the above new registration process, in the subsequent access, this file access can be performed by a normal file access process performed by referring to both the directory management table 14 and the file management table 15.

Design and implementation of Unix 4.3BSD, published by Maruzen, 1991

  According to the technique described in Non-Patent Document 1, as described above, in order to determine the optimum address for storing a newly created file, physical parameters of a secondary storage device such as a disk device are set in advance. I have to leave.

  On the other hand, with recent demands for higher capacity, higher performance, and higher reliability, instead of simply connecting one or more disk devices, disks are placed on the array and data is stored in each disk device. Various forms of secondary storage devices, such as disk array devices that are divided and arranged, have become more sophisticated and complicated.

  Therefore, the physical parameters of the secondary storage device are not only the parameters of a single disk, but also the arrangement of redundant data calculated from the data of multiple disk devices in order to realize the array configuration, data distribution method, and high reliability Etc., depending on the architecture of the secondary storage device. In general, since the host computer and the secondary storage device are distributed independently, the manufacturer cannot set the parameters of the secondary storage device in the host computer in advance.

  For this reason, the parameters of the secondary storage device are set by the user, but it is difficult for the user to set all the parameters of the various secondary storage devices in the host computer. Often, this is not possible. As a result, it becomes difficult to optimally place files on the secondary storage device having a complicated configuration in the host computer, and the performance of the secondary storage device cannot be used effectively.

  SUMMARY OF THE INVENTION An object of the present invention is to provide a computer system that can realize an optimal arrangement of files in a secondary storage device without setting parameters of the secondary storage device in a host computer.

In order to achieve the object, the present invention includes, for example, a computer and one or more secondary storage devices connected to the computer, and the secondary storage device is a low-speed memory in the secondary storage device. A computer system for managing data stored by a local address which is a local address, wherein the secondary storage device includes a storage means for storing a file composed of one or a plurality of logical blocks, and a local When the computer requests the computer to access the logical block that is already stored in the storage unit and is requested by the computer by designating an address, and to store the new logical block in the storage unit A means for determining a local address for storing the logical block requested to be stored by a predetermined procedure, and a logical block requested to be stored in the determined local address. Means for storing a click, determined Russia - and means for notifying the local address to the computer,
The computer includes a file management table for managing a local address in which each logical block constituting a file stored in the secondary storage device is stored in association with each logical block, and a secondary address already stored. When accessing the logical block stored in the storage device, the local address of the logical block to be accessed is obtained by referring to the file management table, and the local address is designated to the secondary storage device. Requests storage of a logical block from the secondary storage device without specifying a local address when storing a new logical block in the secondary storage device and means for requesting access to the logical block. And a local address notified from the secondary storage device is registered in the file management table in association with a new logical block for which storage is requested. To provide a computer system and having a means.

As described above, according to the present invention, it is possible to provide a computer system capable of realizing the optimal arrangement of files in a secondary storage device without setting the parameters of the secondary storage device in the host computer. it can.

  Hereinafter, embodiments of the computer system according to the present invention will be described.

First, a hardware configuration example of a computer system according to the present embodiment is shown in FIG.

  As shown in the figure, the computer system according to the present embodiment includes a host computer 1 and a secondary storage device 2. The host computer 1 is a CPU 10 that executes an OS or an application program, a primary storage device 11 that is used by the CPU 10, and an I / O that controls an input / output request (I / O request) to the secondary storage device 2. It comprises a control unit 9 and an interface 8 to the secondary storage device 2.

  On the other hand, the secondary storage device 2 includes an interface 7 that transmits and receives I / O requests from the host computer 1, a control device 5 that controls data transfer and internal control of the secondary storage device 2, and a disk device that stores data. 4 and a buffer device for absorbing the difference in data transfer speed between the disk device 4 and the host computer 1.

  However, the hardware configuration of the computer system according to the present embodiment may be other configurations similar to the configuration shown in FIG.

  Next, a method for managing the logical location of the file and the location of the file on the secondary storage device in the computer system according to the present embodiment will be described.

  As shown in FIG. 2, the logical positions of files are managed hierarchically by a tree structure constructed using directories. This is a method adopted by the OS such as UNIX described above.

  Here, the file is stored for each block, which is a collection of data of a predetermined length, in the real storage area of the secondary storage device. The local location of the file on the secondary storage device is the logical address at which each block constituting the file is stored, and the local address on the secondary storage device corresponding to this logical address. It is managed by managing. The local location on the secondary storage device is managed by the secondary storage device using the local address of the secondary storage device. A local address of the secondary storage device is called a local address. As described above, the local address may be a physical address that designates a physical location on the secondary storage device, or may be a logical address that designates a logical location on the secondary storage device. .

  However, the logical location of the file and the location of the file on the secondary storage device may be managed by a method according to such a management method.

  When a new file is stored in the secondary storage device 2, as a part of the above-described new registration process, each block into which the file is divided is mapped to a local address of the secondary storage device according to an optimization algorithm. It is necessary to perform processing. However, it is complicated such that a plurality of disk devices are connected in the secondary storage device 2 or the secondary storage device 2 distributes data to a plurality of disk devices according to a so-called RAID (Redundant Array of Inexpensive Disks) architecture. In the case of a secondary storage device having a configuration, it is extremely difficult to input an optimization algorithm corresponding to the type of secondary storage device to be used and various parameters to the host computer 1. For this reason, normally, the secondary storage device 2 looks like a conventional simple secondary storage device, and a conventional algorithm is used. However, in this case, not only the potential capability of the secondary storage device cannot be brought out, but in the worst case, only a lower performance than in the conventional case can be expected.

  The first embodiment of the present invention will be described below.

  FIG. 3 shows the configuration of the computer system according to the first embodiment.

  As shown in the figure, the computer system according to the first embodiment has a host computer 1 and a secondary storage device 2.

  In the host computer 1, 20 represents an OS, and 12 represents an application program process managed by the OS 20. An application program executed in the host computer is managed as a process 12 by the OS 20.

  In the OS 20, reference numeral 13 denotes file management / buffer management means for managing a buffer used for file management and data transfer at the time of file access, and 15 describes a correspondence between a logical address of a directory or file and a local address. A file management table 14 indicates a directory management table describing directory information. Also, 16 is a device driver that converts and controls file access requests from the file management / buffer management means 13 according to the physical characteristics of external devices such as various secondary storage devices, and 18 that the file has been opened. Open file information notifying means for notifying the secondary storage device 2, 19 is a new file write notification means for notifying the secondary storage device 2 that a new file is to be written, and 17 is determined by the secondary storage device 2. A new file address registration means for receiving a new file storage address and registering it in the file management table 15 is shown. In the host computer 1, reference numeral 21 denotes an interface for controlling communication and data transfer with the secondary storage device. -Shows the control means.

  The file management table 15 stores file management information provided for each file. The file management information provided for each file is specified through the address of the file management information specified by the file identification number. In the UNIX, this file management information is called i-node, and the file identification number is called i-node number. Each file management information is configured as shown in FIG.

  That is, each file management information has entries of a mode 151, an owner 152, a time stamp 153, a size 154, a block number 155, a reference count 156, a reference pointer 157, and an update flag 158.

  The mode 151 stores information on the type of file and the current access mode (read / write, etc.) of the file. The owner 152 accesses the owner of the file and the file. Stores information for identifying the authorized user group. Further, the time stamp 153 stores the time when the file was last accessed, the time when the own file management table was updated, and the like. The size 154 stores the number of bytes of the file, and the block number 154 stores the number of blocks of the file. The reference counter 156 stores the number of times the file has been referred to, and the reference pointer 177 stores the logical address of each block of the file in association with the local address of the secondary storage device. Hereinafter, the logical address of the block is referred to as a logical block address, and the local address of the secondary storage device of the block is referred to as a local block address.

  Next, the directory management table 14 also stores directory management information provided for each directory, and each directory management information specifies file management information corresponding to a file belonging to a corresponding directory in the file management table 15. A file identification number and number information for specifying directory management information corresponding to a directory belonging to the corresponding directory (or a pointer indicating the location of the directory management table) are stored.

  By configuring the directory management table 14 and the file management table 15 as described above, a file named file5 is designated by the directories dir1, dir2, and dir3 and the file 5 is designated as a file of each block constituting the file 5 as follows. A Cull Block address can be obtained. That is, the directory management table 14 of dir1, dir2, and dir3 is sequentially scanned, and the file identification number of the file management information corresponding to file5 of the file management table 15 is obtained from the directory management information of dir3, and the file management information of file5 Thus, the local block address of each block constituting the file 5 is obtained.

  On the other hand, in the secondary storage device 2, 29 is one or more disk devices, 24 is a disk device control means for controlling the disk device 29, and 23 receives and analyzes the read and write commands transmitted from the host computer. Command management means for creating a disk command specific to the disk device 29 and sending it to the disk device control means 24, 22 for interface control means for controlling communication and data transfer with the host computer 1, and 27 for secondary A local file management table 39 stores file management information of files stored in the disk device 29 of the storage device 2, and receives information indicating that a file sent from the host computer 1 has been opened, and receives local file management information. An open file information reception registering means for registering in the table 27, 26 is the sending of the host computer 1 New file address determining means for receiving a write of a new file and determining the storage address of the new file, and 28 is a new file address notification for notifying the host computer 1 of the address determined by the new file address determining means 26 Means 25b is a disk management table for managing parameters for determining the configuration of the disk device in the secondary storage device and the usage status of the disk device.

  The configuration of the local file management table 27 and the file management information stored in the local file management table 27 are the same as the file management table 15 and file management information of the host computer 1, and file management of the same file is performed. Information is specified by the same file identification number.

  The file access operation of the computer system according to the first embodiment will be described below.

  First, the file management buffer management means 13 previously loads the file management information stored in the local file management table 27 of the secondary storage device 2 to the file management table 15 of the host computer 1. Keep it. This loading is performed at the initialization of the host computer, when a new storage medium is mounted on the secondary storage device, or at the start of a file open process described below.

  When an open system call is issued from the AP process 12, the OS 20 performs a file open process shown in FIG.

That is, as shown in the figure, when an open system call is issued (step 101), the file management / buffer management means 13 in the OS 20 analyzes this system call and determines the file access mode (step 102). . If the file is opened in the write mode, it is further determined whether or not the file is a new file (step 103). If the file is a new file, the directory management table 14 stores the file in the directory where the file is created. The file name is registered in the corresponding directory management information, and an area for storing the file management information for the new file is secured in the file management table 15 (step 104). Further, the file identification number of the secured file management information is registered in the directory management information corresponding to the directory in which the file is created in the directory management table 14. Then, the file identification number is notified to the process 12 (step 105).

  If it is not a new file, the file identification number of the file is obtained from the directory management table 14, and this is notified to the process 12 (step 105).

  Next, the file management / buffer management means 13 informs the open file information notifying means 18 of the local address and file name of the file management information specified by the file identification number unique to this file. The open file information notifying means 18 notifies the secondary storage device 2 of these pieces of information via the device driver 16 and the interface control means 21 and 22 (step 106).

  The open file information reception registration means 39 of the secondary storage device 2 receives these information, and if it is for a new file, the new file management information area in the local file management table 27. And register the received information.

  Through the above processing, the management information of the opened file is registered in both the file management table 15 of the host computer 1 and the local file management table 27 of the secondary storage device 2.

  Next, when the file open process ends, the file read or write process is executed.

  FIG. 6 shows a file read process procedure in the OS 20, FIG. 8 shows a file write process procedure in the OS 20, and FIG. 9 shows a file read process and write process procedure in the secondary storage device.

  First, file read processing will be described.

  As shown in FIG. 6, when the host computer 1 issues a file read system call (read ()) from the AP (process 12) (step 201), the file management / buffer management means 13 of the OS 20 performs directory management. With reference to the table 14, the file identification number of the file (indicating the position of the file management information in the file management table 15) is obtained (step 202). Next, the file management / buffer management means 13 obtains information of file management information in the file management table 15 designated by the acquired file identification number.

  Here, the AP requests to read or write a file in byte units called an offset by the file read system call or the file write system call. FIG. 7 shows a case where the AP requests to read an n-byte offset from the file pointer fp provided for the file fd to the user buffer buffer. As shown in the figure, this offset is data straddling logical block addresses 1 and 2, that is, two local blocks (local block addresses # 11020 and # 11028) corresponding to logical blocks 1 and 2. ) And need to lead from. A local block is a block on the secondary storage device 2 corresponding to a logical block.

Therefore, the file management / buffer management means 13 next obtains the logical block addresses 1 and 2 where the requested offset data exists from the obtained file management information (step 203). Then, a system buffer is secured as a temporary work area necessary for reading these two logical blocks (step 204). Next, the first logical block address 1 is selected (step 205), converted to the local block address # 11020 (step 206), and the data block of the local block address is transferred to the device driver 16. Issue a request. The device driver 16 receives this request, generates a read command specific to the secondary storage device, and sends it to the secondary storage device 2 via the interface control means 21 (step 207).

  As for the read command, the command management means 23 of the secondary storage device 2 receives the read command via the interface control means 22. The secondary storage device 2 that has received the command first analyzes this command as shown in FIG. 9 (step 402). Since it can be seen that this is a read command (step 403), management information such as the number of accesses described later in the local file management table 27 of the secondary storage device 2 is updated as necessary (step 411), and the disk Data to be obtained from the device 29 is read and transferred to the host computer 1 via the interface control means 22 (step 412). Then, if all the requested data can be correctly transferred, the command management unit 23 reports the end to the host computer via the interface control unit 22 (step 408).

  This end report is received by the device driver 16 of the higher-level device 1.

  Returning to FIG. 6, when the command end notification is received from the secondary storage device 2 (step 208), the device driver 16 notifies the file management / buffer management means 13 of this. The file management / buffer management means 13 confirms whether or not reading of all logical blocks to be transferred has been completed (step 209), and if there are still untransferred logical blocks, selects the subsequent logical block ( Step 212) The above processing is repeated. If the transfer of the data of all logical blocks to the system buffer is completed, the offset data requested by the AP is copied to the user buffer (step 210), the system call completion notification is returned to the AP, and the file read processing is performed. To complete.

  Next, the file write process will be described with a focus on differences from the read process described above.

  File write processing is roughly divided into the following two processes. One is a new block registration process, and the other is an existing block update process. The former occurs when writing a new file or adding data to the end of an existing file, and the latter occurs when writing data from the middle of an already registered logical block or when an existing file is recorded (one person). Occurs when updating in units of data.

  As shown in FIG. 8, when a write system call is issued from the AP (step 301), the file management / buffer management means 13 of the OS 20 acquires a file identification number from the directory management table 14 (step 302). As described above, a logical block address to be transferred is obtained from the offset (step 303), and a necessary system buffer is secured (step 304). Here, the file management table 15 is referred to. If the local block address corresponding to the logical block address is not registered, it is determined that this logical block is a new logical block (step 305).

  If it is determined that the block is a new logical block, the file management / buffer management means 13 copies the write data from the user buffer to the system buffer (step 307), and selects the first logical block. On the other hand, the new file write notification means 19 generates a new block write request for requesting writing of the selected logical block. From this, the device driver 16 generates and issues a new block write command unique to the secondary storage device 2 (step 312).

  This new block write command includes a file identification number and a logical block address.

  The issued new block write command is received (step 401) and analyzed (step 402) by the command management means 23 of the secondary storage device 2, as shown in FIG. When the command is recognized as a new block write command as a result of the analysis (steps 403 and 404), the command management unit 23 passes the file unique identification number and the logical block address to the new file address determination unit 261. The new file address determining means 26 refers to the local file management table 27 and the disk management table 25b by the file unique identification number, and determines the local block address for storing this block in the disk device 29 of the secondary storage device 2. This is determined in consideration of configuration parameters, physical parameters, and the like (step 405).

  A procedure for determining such an optimal local block address will be described later.

  Next, the determined local block address is registered in the local file management table 27 in association with the logical block address, and the local block is in use in the disk management table 25b. Set.

  Next, data transfer is started, data transfer is received from the host computer 1, and this is written to the disk device 29 according to the determined local block address (step 406). Next, the new file address notifying means 28 transmits the determined local address to the host computer 2 (step 407). When the determination of the local block address of this block and the data transfer are completed, an end notification is sent to the host computer 1 (step 408).

  The transmitted local block address is received by the new file address registration means 15 via the interface control means 21 and the device driver 16 in the host device.

  Returning to FIG. 8, when the new file address registration unit 15 of the host computer 1 receives the local block address from the secondary storage device 2 via the interface control unit 21 and the device driver 16, the file management / buffer management unit 13. Request registration of this local block address. The file management / buffer management means 13 registers this local block address in association with the corresponding logical address of the file management information of the corresponding file in the file management table 15 (step 314). Further, when the file management / buffer management means 13 receives the block transfer end notification from the secondary storage device 2 (step 315), the file management / buffer management means 13 ends the write processing of the block. If writing of all logical blocks has not been completed, the next relevant logical block is selected, and the above processing is repeated until the writing processing of all logical blocks is completed (step 318).

  On the other hand, if it is determined in step 305 that the block to be written is not a new logical block, that is, the write processing to the existing logical block is as follows.

  The basic difference between this process and the write process of a new logical block is that the write process to an existing logical block may be updated from the middle of the block, and that -The point where the cull block address is mapped.

  First, when updating from the middle of a logical block, it is necessary to read this logical block once. That is, the block is read from the secondary storage device to the system buffer, new data is transferred from the user buffer to the system buffer, the block of the system buffer is updated, and then this block is written to the secondary storage device. There is.

  Therefore, when writing to an existing block, read processing is first performed (step 306). The processing in step 306 is equivalent to the processing from the symbol a to b shown in FIG.

  Next, since the mapping of the local block address to the existing block has already been completed, when the write command is issued to the secondary storage device 2, the local block address is designated on the host computer 1 side. Therefore, the same logical block to local block address conversion process as in the read process is performed (step 310), and then an update write command to the secondary storage device is generated and issued (step 311).

  On the other hand, in the secondary storage device 2, in accordance with the received update write command, the block data transferred to the designated local block address is written (step 410 in FIG. 9). At this time, management information such as the number of accesses described later in the local file management table 27 of the secondary storage device 2 is updated as necessary (step 409).

  As described above, according to the present system, when a file is stored in the secondary storage device, the secondary storage device determines the local address to be stored. The file can be stored at the optimum position that matches the parameter, the physical parameter, and the file access characteristic determined by the AP characteristic of the host computer 1, and the effect of increasing the file speed is great.

  Incidentally, the host computer 1 has a conventional secondary storage device 2 that does not have the ability to determine a local block address of a new file in addition to the secondary storage device 2 described in the first embodiment (see FIG. 24) is also connectable. That is, it is desirable to configure the host computer 1 so that the conventional secondary storage device shown above can be used interchangeably.

  Therefore, as a second embodiment of the present invention, a computer system that can use such a conventional secondary storage device interchangeably will be described. FIG. 10 shows a configuration of a computer system according to the second embodiment.

  In the computer system according to the second embodiment, the disk management table 25a, the file address determining means 30, the file address determining means 30 and the new file address determining means 26 are added to the host computer 1 according to the first embodiment. Address determining means selecting means 31 for selecting whether to use is added.

  The disk management table 25a manages the disk usage status. The file address determining means 30 determines a local block address of a disk device that is optimal for writing when a logical block is newly written.

  Next, the operation of the computer system according to the second embodiment will be described.

  First, at the time of initializing the system, the file management buffer management means 13 of the host computer 1 negotiates with the command management means 23 of the secondary storage device 2 via the device driver 16, and the secondary storage device 2 Is determined whether it has the ability to determine the local block address of the new file, and the determination result is notified to the address determining means selecting means 31.

  The address determination means selection means 31 invalidates the file address determination means 30 and the disk management table 25a of the host computer 1 if the secondary storage device 2 has the ability to determine the local block address of the new file. Let's say. Thereafter, the same operation as in the first embodiment is performed.

  On the other hand, if the secondary storage device 2 does not have the ability to determine the local block address of the new file, the address determination means selection means 31 will change the open file information notification means 18 and the new file write notification means 19. It is invalidated and the same operation as the conventional computer system described above is performed.

  Further, when a plurality of secondary storage devices are connected and used at the same time, the requested secondary storage device is a local block of a new file for each access request from the AP to the secondary storage device. Depending on whether or not there is an ability to determine an address, one of the set of the open file information notification means 18 and the new file write notification means 19 and the combination of the file address determination means 30 and the disk management table 25a is invalidated. And the same operation as the operation of the first embodiment or the conventional operation is performed.

  Even if the secondary storage device 2 has the ability to determine the local block address of the new file, the open file information notification means 18 and the new file write notification means 19 are invalidated and the secondary storage device By invalidating the new file address determining unit 26, the new file address notifying unit 28, etc., the same operation as the conventional one can be performed.

  As described above, according to the second embodiment, the conventional secondary storage device can also be used in the computer system of the first embodiment. In addition, it is possible to connect multiple secondary storage devices and attach a new secondary storage device to a secondary storage device that has been used in the past, increasing the degree of freedom and expandability of the computer system configuration. be able to.

  The third embodiment of the present invention will be described below.

  In the third embodiment, unlike the first embodiment, the host computer 1 does not manage the local block address of the file at all.

  FIG. 11 shows a configuration of a computer system according to the third embodiment.

  As shown in the figure, the host computer 1 includes a directory management table 14 for managing directories and file names and file identification numbers, a buffer management means 13a for managing buffers used for data transfer, a device driver 16, and an interface. Control means 21.

  The secondary storage device 2 also includes an interface control unit 22, a file management / command management unit 23a that performs file management and command management, and a new that maps a local block address to a logical block of a new file. An address determination means 26, a local file management table 27 for registering file management information, a disk management table 25b for managing the disk usage status, and describing configuration parameters and physical parameters of the disk device 29, A disk control means 24 and a plurality of disk devices are provided.

  The upper computer 1 and the secondary storage device 2 support a command for notifying the opening of the file from the upper computer 1 to the secondary storage device 2 and a command for requesting transfer of specified data in the file together with the file identification number. Have Further, the host computer 1 and the secondary storage device 2 can transfer data to each other in a form (block unit transfer or byte unit transfer) in accordance with the request. The interface control means 21 and 22 control such data transfer and command transmission / reception.

  The operation of the computer system according to the third embodiment will be described below.

  The process 12 issues system calls such as opening a file and reading / writing a file to the OS 20. The OS 20 receives this, and the buffer management means 13a secures a buffer necessary for data transfer, and generates and issues an open or read / write command to the secondary storage device 2 via the device driver. The command includes the offset included in the system call. Further, when issuing this command, the directory management table 14 is referred to, the file identification number of this file is acquired, and transmitted to the secondary storage device 2 together with the command. When an unregistered file is opened, a new file identification number is assigned, and the assigned file identification number is transmitted to the secondary storage device 2 together with a command.

  In the secondary storage device 2, the file management / command management unit 23 a receives and analyzes the command, and if this command is a command requesting registration of a new logical block, the storage address is stored in the new file address determination unit 26. Ask for a decision. The new file address determination means 26 refers to the disk management table 25b and the local file management table 27, and stores this new logical block in consideration of the configuration parameters, physical parameters, AP access characteristics, etc. of the disk device 29. A local block address is determined and registered in the local file management table 27. Details of determining the local block address will be described later.

  The file management / command management means 23a refers to the local file management table 27 when the analyzed command is a command requesting reading or writing of an existing logical block, and the local block of the logical block to be accessed. The address is obtained, and the data specified by the local block address and the offset in the command is accessed. For the read command, the read data is transferred to the host computer, and for the write command, the transferred data is written. Data transfer between the host computer 1 and the secondary storage device 2 is performed in units requested by the AP.

  As described above, according to the third embodiment, the host computer does not need to consider the configuration of the secondary storage device, physical characteristics, etc. at all, and only manages the file name and the file identification number. Accordingly, various secondary storage devices and other external devices can be connected to the host computer for use.

  The fourth embodiment of the present invention will be described below.

  FIG. 12 shows a configuration of a computer system according to the fourth embodiment.

  The computer system according to the fourth embodiment is obtained by connecting a plurality of host computers and secondary storage devices 2 according to the third embodiment.

  Each host computer 1 and each secondary storage device 2 are connected by a common interface. This interface is, for example, a bus type interface known as a SCSI bus.

  The operation of the computer system according to the fourth embodiment is almost the same as that of the third embodiment. However, since the management information of each file is distributed and arranged for each secondary storage device, each host computer 1 manages the secondary storage device storing each file. This can be realized when each host computer obtains information on the file stored from each secondary storage device at an appropriate timing. In addition, the secondary storage device monitors and arbitrates double assignment of file identification numbers to new files so that different host computers do not assign the same file identification numbers to different files on the same secondary storage device. . This is because, for example, the secondary storage device requests the host computer to reassign the file identification number when double assignment of the file identification number occurs at the time of opening a new file from the host device. Can be realized.

  Now, the sequence of command and data transfer through the bus is executed as shown in FIG. 13, for example. That is, first, arbitration is executed to logically connect the specific higher order computer 1 and the specific secondary storage device 2, and the command is transferred from the higher order computer 1 to the secondary storage device. Next, the host computer transmits message 1 to release the bus to another device during the preprocessing period (disk device seek period, etc.) for the command of the secondary storage device 2, and once sends the bus. To release. When the preprocessing is completed, the secondary storage device 2 transmits a message 2 and reconnects the host computer and the bus. Next, the data to be read / written is transferred between the higher order computer 1 and the secondary storage device 2, and finally the message 3 indicates the normal end or abnormal end of the command from the secondary storage device. And release the bus.

  Such a bus can also be used in the first and second embodiments. When used also in the first and second embodiments, the local block address storing the new logical block by the message 3 is also transferred from the secondary storage device 2 to the host computer 1.

  As described above, according to the fourth embodiment, each host computer 1 does not need to hold information other than the file name, its file identification number, and the identification of the secondary storage device storing the file. A plurality of host computers and a plurality of secondary storage devices can be easily connected. In addition, since the file management information is held only by the secondary storage device that stores the file, the secondary storage device can easily perform control to maintain the coherency of the files for a plurality of higher-level devices. Can do.

  Now, the details of the procedure for determining the local block address of each logical block of the file in the secondary storage device will be described.

  The procedure described below can be applied to any of the first to fourth embodiments.

  In FIG. 14, 24 is a disk device control means, and 29 is a disk device group composed of a plurality of units. The disk device control means 24 manages the disk device group 29 as a RAID 5 type disk array. The RAID 5 type disk array is a disk array architecture for the purpose of high-speed disk access and high reliability. In this architecture, data is divided into units called stripes and distributed to each disk device as shown in the figure. As a result, in the stripe unit access, each disk device can be operated completely independently, and a transaction performance that is double the number of disks can be obtained.

  Further, a parity group is formed by horizontally arranged data stripe groups (for example, D00, D01, D02), and an exclusive OR of these data is calculated and stored as parity (for example, P0) on one disk device. . This parity is distributed and arranged in different disk devices for each parity group. Thereby, even if one disk device fails, the data of the failed disk can be restored using the parity.

  In such a RAID 5 type disk array, in order to optimally arrange each logical block, configuration parameters such as a stripe size, the number of disks, and a parity arrangement method are required. However, since these parameters are different for each type of secondary storage device, it is difficult to manage them in a host computer.

  On the other hand, if these parameters are not taken into account, for example, when the stripe size and the logical block size are the same size, the logical block is not arranged properly at the top of one stripe, that is, the stripe boundary, and is arranged across two stripes. Will happen. In this case, when one logical block is accessed, it is necessary to access two stripes, that is, two disk devices at the same time, so that the performance is remarkably deteriorated.

  Therefore, in the present embodiment, the secondary storage device 2 side determines the arrangement of logical blocks based on its own configuration parameters. Hereinafter, problems caused by the arrangement of logical blocks performed without considering the configuration parameters, and examples of the arrangement of logical blocks performed by the secondary storage device 2 based on its own configuration parameters in order to avoid this problem Indicates.

  First, a first example will be described with reference to FIG.

  FIG. 15a shows a case where the data is stored using two stripes of the D00 stripe of the disk 0 and the D01 stripe of the disk 1 as a result of determining the arrangement of the logical block without considering the stripe boundary. ing. In this case, since two disk units are used for one request from the host computer, two disks 2 and 3 can be used for processing other requests in parallel with this. Become.

  Therefore, the secondary storage device of this embodiment determines the arrangement so that the top of the logical block transferred from the host computer is aligned with the stripe boundary based on its configuration parameters, as shown in FIG. 15b. The data is stored only in the D01 stripe of disk 1. In this case, when the host computer issues a write request for a new logical block with 8 KB fixed, a maximum of four requests can be processed simultaneously, and the processing throughput can be improved.

  Next, a second example will be described with reference to FIG.

  FIG. 16a shows a 48 KB new logical block placed in consideration of only the stripe boundary. Here, it is necessary to generate a parity in the RAID5 type disk array. The required parities are P0, P1, and P2, which can be generated according to the following generation formula, assuming that “*” represents an exclusive OR.

P0New = (D01Old * D02Old) * (D01New * D02New) * P0Old
P1New = D10New * D11New * D31New
P2New = D20Old * D20New * P2Old
Here, the data with subscript Old (for example, D01 Old, D02 Old, P0 Old,...) Is data already written on the disk, and the data with subscript New (D01 New, D02 New,. The data to be written. Therefore, it is necessary to read D01 Old, D02 Old, and P0 Old from the disks 1, 2, and 3 before writing D01 New and D02 New to generate P0, and to generate P2, D20 Old from disks 0 and 1 before writing D20 New. , P2Old must be read. A read prior to this write (read-modify-write process, hereinafter referred to as “RMW process”) requires a great deal of processing time.

  Therefore, the secondary storage device according to the present embodiment determines the arrangement according to the following procedure based on its own configuration parameters.

  That is, the new file address determining means 26 refers to the local file management table 27 and determines the approximate address. When adding a logical block to a file, the approximate address is the address where this file is currently stored. If the logical block of a new file is stored, the approximate address of the disk area currently being used is stored. Address. Selecting the approximate address of the new logical block in this way places the data that is supposed to be related to the disk space data that is currently in use close to the disk space data that is currently in use. This is because the time required for moving the head of the disk is less probable.

  Next, when the approximate address is determined, the new file address determining means 26 refers to the disk management table 25 to determine the detailed address.

  That is, just 48 KB, that is, 6 stripes in the vicinity of the approximate address, starting from the stripe of disk 0, an area where new logical blocks can be placed in order is determined as an address for storing a new file. In this example, it can be known by referring to the stripe management table that a total of six consecutive data stripes excluding the parity stripe of the stripe # n + 1 of the disk 3 from the stripe #n of the disk 0 are unused.

  As shown in FIG. 17, the disk management table further includes a stripe management table and a sector management table.

  Here, the stripe management table 1700 has an entry for registering, for each stripe, whether the data stripe is a parity stripe, whether it is used, unused, or partially used. Whether the stripe is a parity stripe or whether the entire stripe is in use, unused, or partially used can be known for each disk. A sector management table 1710 is provided for managing the sector usage status for each stripe of each disk. The sector management table 1710 shown in FIG. 17 is for a stripe size of 8 KB. In this case, one stripe is composed of 16 sectors of 512 B. With such a configuration, it can be determined whether each sector is in use or unused.

  For example, the stripe # n + 2 (added with a circle in the figure) of the disk # 0 has the attribute “010”, which indicates that the data stripe is partially used. Therefore, referring to the sector management table of the same stripe, it can be seen that sector numbers 0 to 7 in the stripe are in use and 8 to 15 are unused. From this, it is determined that the new logical block transferred from the host computer is stored in a continuous area from D00 to D13 as shown in FIG. 16b.

  Furthermore, in this case, there are two parity stripes, P0 and P1, that need to be generated, and each is generated by the following generation formula.

P0New = D00New * D01New * D02New
P1New = D10New * D11New * D13New
From this generation equation, it can be seen that parity can be obtained without reading data prior to writing.

  Assuming that the RMW process takes 1.7 times longer on average than the W process, the RMW process is unnecessary compared with the case of storing as shown in FIG. Then, the new logical block is stored.

  As described above, according to the present embodiment, in the secondary storage device, the data transfer length and other related files are related to the parity arrangement method of the RAID type disk array, the stripe size, the number of disks, and the like. Optimal file storage considering the positional relationship of

  In the above, an example of the optimal logical block arrangement in the RAID 5 type disk array has been shown. However, regardless of the configuration of the disk device (group), its configuration parameters, physical characteristics, logical block size, AP An optimal logical block layout that matches the access characteristics can be realized. Similarly, an optimal arrangement can be realized for a disk device other than the disk array, or an optical disk, a tape device, or a semiconductor storage device.

  The fifth embodiment of the present invention will be described below.

  The fifth embodiment relates to a secondary storage device 2 that can be used in the computer system according to each of the embodiments.

  The main difference between the secondary storage device 2 according to the fifth embodiment and the secondary storage device (see FIG. 3) shown in the first embodiment is the secondary storage device 2 according to the fifth embodiment. Has a plurality of disk device groups.

  FIG. 18 shows the configuration of the secondary storage device according to the fifth embodiment.

  In the figure, 36a1, 36a2, 36a3 are disk device groups, 24a1, 24a2, 24a3 are disk device control means for controlling each disk device group, 34 is a disk device group selection means for selecting which disk device group is used, 35 is a disk device group switching means for switching the disk device group, 25b1, 25b2, and 25b3 are disk management tables for managing the configuration parameters, physical characteristics, and usage status of each disk group, and 1800 is a disk group update control means. The other parts are the same as the parts indicated by the same reference numerals in FIG.

  The outline of the operation of the secondary storage device according to the fifth embodiment will be described below.

  The command management means 23 of the secondary storage device 2 receives an access request from the host computer 1, and if the request is a request to write a new logical block, notifies the new file address determination means 26 of this, The new file address determination means 26 refers to the local file management table 27 and the respective disk management tables 25b1, 25b2, and 25b3, selects an optimum disk device group, and maps the local block address. Then, this is notified to the disk device group selecting means 34, and the disk device group selecting means 34 receives this and operates the disk device group switching means 35 to select the selected disk device group. The subsequent processing is the same as in the first embodiment.

  The selection of the disk device group performed by the new file address determining means 26 is performed in consideration of the transfer length, the characteristics of the access frequency of the file itself, the characteristics of the disk device group, and the like. A more detailed example will be described later.

  The disk device group according to the fifth embodiment can be configured specifically as shown in FIG. 19, for example.

  In the example shown in FIG. 19, the disk device group 1 is a RAID 1 type disk array device (or mirror disk device), the disk device group 2 is a RAID 3 type disk array device, and the disk device group 3 is a RAID 5 type disk array device.

  In this case, it is appropriate to store a frequently accessed file or a file requiring high performance in the RAID 1 type disk device group 1, and store a large-capacity sequential file such as image data in the RAID 3 type disk device group 2. The random file for transaction such as database access is preferably stored in the RAID 5 type disk device group 3.

  Therefore, the new file address determining unit 26 selects a disk device group as follows.

  First, the new file address determination means 26 selects the disk device group storing the existing logical block when the command from the host computer requests to update the existing logical block of the existing file. To do.

  On the other hand, if the command from the host computer requests registration of a new logical block, the new file address determination means 26 first determines which disk device group this logical block is registered in as shown in FIG. Determine according to the procedure shown in.

  That is, first, it is determined whether the file to which the logical block to be written belongs is new or existing (step 2001). If it is a new file, the characteristics of the file are unknown, so for the time being a RAID 1 type disk array (disk group 1). Is selected (step 2006).

  On the other hand, if the file to which the logical block to be written belongs is an existing file, a determination is made as to the random accessibility of the file (step 2002). This determination is made using the local file management table 27.

That is, as shown in FIG. 21, five subfields for determining access characteristics are provided in the reference count field 156 of each file management information in the local file management table. Here, the first subfield is a field 1561 for storing the address (number of bytes from the beginning of the file) of the request from the host computer referenced last time, and the second subfield is the transfer length (number of bytes) at the previous reference time. , The third subfield is a field 1563 indicating the frequency of random access, the fourth subfield is a field 1564 indicating the frequency of sequential access, and the fifth subfield is the total reference count of this file. Is a field 1565 for storing. The random access count field 1563 and the sequential access count field 1564 are updated by the following process every time a file is accessed. That is,
Expression Current reference address ≤ Previous reference address + Previous transfer length + α
If this expression is satisfied, it is determined that this access is sequential, and 1 is added to the sequential access count subfield 1564. If not established, it is determined as random access, and 1 is added to the random access count subfield 1563. However, “current reference address” is the position (number of bytes) from the top of the file of the write request of the host computer, and α is a constant for determining the sequentiality. Note that if it is desired to make only continuous access only sequential, α may be set to 0. If it is desired to determine the sequentiality with a certain amount of width, α may be set to a certain value. In this way, random / sequentiality is determined each time a file is referenced.

  Now, returning to step 2002, the determination regarding the random accessibility of the existing file, that is, the determination of whether the random accessibility is large or small is performed as follows.

  That is, if A and B are appropriate setting values, the value of the reference count field 1565 of the corresponding file management information is larger than A, and the value of the random access count field 1563 / reference count field 1565. If the value of B is greater than or equal to B, it is determined that the random accessibility is large, the value of the reference count field 1565 is greater than A, and the value of the random access count field 1563 / the value of the reference count field 1565 is If it is less than B, it is determined that the random accessibility is small. If the value of the reference count field 1565 is smaller than A, it is determined that the determination is impossible. This is because if the reference count is smaller than a certain value A, the file has just been created and the sequential / randomness cannot be determined yet.

  Next, as a result of this determination, if the randomness is large, the RAID 5 type disk array (disk group 3) is selected (step 2004). If the determination is impossible, the RAID 1 type disk array (disk group 1) is selected (step 2006). . Here, when the determination is impossible, the RAID1 type disk array (disk group 1) is selected because the file that has just been created should be stored in the RAID1 type disk array (disk group 1). is there.

  In step 2002, it is determined whether or not the random access ratio = random access count / reference count is greater than the constant B (O <B <1), but the sequential access ratio is set in the sequential access count field 1564. The determination may be made based on the value in the value / reference number field 1565, or may be determined in consideration of both.

  If it is determined in step 2002 that the randomness is small, the transfer length is determined. That is, if the data length (transfer length) requested to be written by the command is equal to or greater than a certain value C, it is determined that the transfer length is large, and the RAID 3 type disk array (disk group 2) is selected. When the transfer length is less than a certain value C, the RAID3 type is selected because the RAID3 type is inefficient.

  As described above, an appropriate disk group is selected by determining whether the new file is an existing file, whether the randomness is large or small, and whether the transfer length is large or small. Then, when a new logical block is written to the selected disk group and the selected disk group is different from the disk group in which the existing logical block of the existing file to which the new logical block belongs is stored, select The existing logical block is moved to the disk group and the file management table is updated accordingly. At this time, an update flag 188 indicating whether or not a file is relocated provided in each file management information of the local file management table 27 is set to “updated”.

  In the above description, all new files are written to the RAID 1 type disk array (disk group 1). This is because the access characteristics are unknown. However, the RAID type 1 disk array has the same performance as a single disk except for its high reliability, and is not suitable for accessing a file with a strong randomness or a file with a strong sequence sealing property. Further, since the disk capacity is limited, it is desirable to reselect a disk group suitable for the access characteristics at an appropriate timing and move the file.

  Therefore, the disk group update control unit 1800 periodically issues a request to the command management unit 23 to reselect a disk group to be stored for each file.

  In response to this request, the command management means 23 refers to the update flag 188 of each file management information in the local file management table 27. If the update flag is set to “unupdated”, it is newly created recently. Therefore, the disk device group selection means is activated and tries to reselect the disk group according to the procedure shown in FIG.

  If it is determined that the file has a strong randomness or sequentiality, the disk group as described above is selected, and when a disk group other than the RAID 1 type disk array (disk group 1) is selected, the disk group is selected. And the update flag 188 is set to “updated”.

  If the determination is impossible, the file is placed in the disk group 1 as it is. At this time, the update flag remains “unupdated”. Since an access request to this file may be issued from the host computer during file movement, the update flag is set to “Updating” at this time. In addition, a new file management table is recreated as the file is moved, and when the file movement is completely completed, the file management table is switched to a new one and the old one is deleted. The registration of the logical block after selection of the disk group to the logical block address may be performed as described above.

  Through the above processing, a disk group suitable for access characteristics can be selected.

  In the above example, RAID 1 type, RAID 3 type, and RAID 5 type disk arrays are used as the disk group. However, even if other devices are installed, the disk group suitable for the access characteristics is selected by the same procedure. it can.

  The disk device group according to the fifth embodiment (see FIG. 19) can also be configured as shown in FIG.

  In the example shown in FIG. 22, three disk array devices having different stripe sizes, which are one of the configuration parameters, are defined as disk device groups 1, 2, and 3.

  In this case, registration to the disk device group 1 having a small stripe size for storing relatively small files, and conversely to the disk device group 3 having a large stripe size for storing sufficiently large files is required. Appropriate. Therefore, if a disk device group is selected according to the file length, a disk group suitable for the characteristics of the file can be selected.

  Alternatively, if the disk group is selected such that the stripe length × number of disks × n is the transfer length according to the data transfer length, transfer performance equivalent to the number of disks can be obtained efficiently.

  Further, the disk device group (see FIG. 19) according to the fifth embodiment can also be configured as shown in FIG.

  In the example shown in FIG. 23, a magnetic disk device is used for the disk device group 1, an optical disk device is used for the disk device group 2, and a tape library device is used for the disk device group 3. In this case, it is desirable to register a file with high access frequency in the magnetic disk device, a file with low access frequency with the optical disk device, and a file with little access (such as backup) in the tape library device.

  Therefore, the disk group update control unit 1800 issues a request to the command management unit 23 to reselect a disk group to be stored for each file at certain time intervals. In response to this request, the command management means 23 refers to the previous reference time and the reference count field 1565 of the time stamp 153 of each file management information in the local file management table 27, and this file has been accessed recently. It is determined whether or not the number of times of reference is large. If the file has not been accessed for a certain period of time, it is an old file and it is inefficient to occupy the space on the high-speed disk. Therefore, the file is moved to an optical disk, for example.

  Note that, for example, at night, a file stored in another disk device may be copied to the tape library device and backed up.

  As described above, according to the present embodiment, a plurality of disk device groups can be provided in the secondary storage device, and logical block addresses of logical blocks suitable for various access requests and file characteristics can be changed by changing the configuration. Mapping can be realized, and various forms of storage systems and computer systems can be constructed.

  As described above, according to the present embodiment, it is possible to realize an optimal file arrangement that takes into account the configuration parameters and physical parameters unique to the secondary storage device and that matches the access characteristics of the AP of the host computer, and is fast. File access can be realized. This also makes it possible to realize a high-performance computer system. Further, not only the secondary storage device of the present invention but also a conventional secondary storage device can be connected to the host computer, and a flexible computer system can be constructed. In addition, since the host computer need not consider the configuration and physical characteristics of the secondary storage device, various secondary storage devices can be easily connected to the host device. In addition, a distributed file management computer system in which a plurality of host computers and a plurality of secondary storage devices are connected can be easily constructed.

According to the computer system of the present invention, when storing a new logical block in the secondary storage device, the computer does not specify the local address of the secondary storage device, but the secondary storage device. Request the device to store a logical block. On the other hand, the secondary storage device, when requested by the computer to store a new logical block in the storage means, determines a local address for storing the logical block requested to be stored by a predetermined procedure, The logical block requested to be stored is stored in the determined local address, and the determined local address is notified to the computer. The computer registers the notified local address in the file management table in association with the new logical block requested to be stored, and when accessing this logical block thereafter, the file management table. The secondary storage device is requested to access by designating the local address of the logical block to be accessed.

As described above, according to the present invention, the secondary storage device determines the local address for storing the logical blocks constituting the file. Here, since it is easy for an individual secondary storage device to grasp its own configuration and state, the secondary storage device can determine the optimal storage location of the logical block. Therefore, it is possible to realize the optimal arrangement of the files in the secondary storage device without setting the secondary storage device parameters in the host computer.

It is a block diagram which shows the hardware constitutions of the computer system which concerns on the Example of this invention. It is explanatory drawing which showed the concept of the file management in the Example of this invention. It is the block diagram which showed the structure of the computer system which concerns on 1st Example of this invention. It is explanatory drawing which shows the structure of the file management table which concerns on 1st Example of this invention. It is a flowchart which shows the procedure of the file open process which OS which concerns on 1st Example of this invention performs. It is a flowchart which shows the procedure of the file read process which OS which concerns on 1st Example of this invention performs. It is explanatory drawing which showed the relationship between the data to access and a logical block. It is a flowchart which shows the procedure of the file write process which OS which concerns on 1st Example of this invention performs. 3 is a flowchart showing a file read / write processing procedure performed by the secondary storage device according to the first embodiment of the present invention; It is the block diagram which showed the structure of the computer system which concerns on 2nd Example of this invention. It is the block diagram which showed the structure of the computer system which concerns on 3rd Example of this invention. It is the block diagram which showed the structure of the computer system which concerns on 4th Example of this invention. It is a time chart which shows the transfer sequence between the high-order computer and secondary storage which concern on 4th Example of this invention. It is explanatory drawing which shows the stripe of a disk array apparatus. It is explanatory drawing which shows how a local address is allocated to a logical block. It is explanatory drawing which shows how a local address is allocated to a logical block. It is a block diagram which shows the structure of the stripe management table and sector management table which concern on the Example of this invention. It is a block diagram which shows the structure of the secondary storage device based on 5th Example of this invention. It is a block diagram which shows the 1st specific structural example of the secondary storage device based on 5th Example of this invention. It is a flowchart which shows the procedure of the disk group selection process based on 5th Example of this invention. It is explanatory drawing which shows the structure of the local file management table which concerns on 5th Example of this invention. It is a block diagram which shows the 2nd specific structural example of the secondary storage device based on 5th Example of this invention. It is a block diagram which shows the 3rd specific structural example of the secondary storage device based on 5th Example of this invention. It is a block diagram which shows the structure of the conventional computer system.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... High-order computer 2 ... Secondary storage device 12 ... Process 13 ... File management and buffer management means 14 ... Directory management table 15 ... File management table 16 ... Device driver 17 ... new file address determining means 18 ... open file information notifying means 19 ... new file write notifying means 20 ... operating system 21 ... interface control means 22 ... interface control means 23 ... Command management means 24 ... disk device control means 25b ... disk management table 26 ... new file address determination means 27 ... local file management table 28 ... new file address notification means 29 ... Disk unit

Claims (1)

A computer system comprising a computer and a secondary storage device connected to the computer and having a plurality of disks for storing files comprising logical blocks handled by the computer with logical block addresses;
  When the computer newly stores a file in the secondary storage device, the computer determines a file identifier for identifying the file, the file identifier, the logical block, and a logical block address corresponding to the logical block, Means for transmitting to the secondary storage device a first type 1 write command comprising:
  The secondary storage device determines an interface that receives the first-type write command and a local address that indicates a position in the disk where the logical block is stored when the first-type write command is received. Means, first means for storing a correspondence relationship between the file identifier, the logical block address, and the determined local address; and dividing the plurality of disks into a plurality of stripes; Management means for managing the storage status of the logical block for each sector, means for storing the logical block at the determined local address, means for transmitting the determined local address to the computer,
  The computer further stores a second means for storing the correspondence relationship between the received local address, the file identifier, and the block address, and a second means for reading a file stored in the secondary storage device. Means for referring to the local address using the file identifier of the file for the stored correspondence; and means for transmitting a second type of read command having the referenced local address to the secondary storage device. Possess
  The secondary storage device further uses interface means for receiving the second type of read command and a local address included in the second type of read command when the second type of read command is received. Means for reading a logical block constituting the file stored in the plurality of disks, and means for transmitting the read logical block to the computer.
  When the computer further stores a new logical block in a file stored in the plurality of disks of the secondary storage device, the computer further includes a logical identifier corresponding to the file identifier, the new logical block, and the new logical block. Means for transmitting a second type 1 write command having a block address to the secondary storage device,
  The secondary storage device further includes an interface unit that receives the second first type write command, and the file identifier that the second first type write command has a correspondence stored in the first unit. Means for determining a local address storing the file before receiving the second first type write command, and a logical block indicated by the determined local address is stored. A computer system comprising: means for determining a stripe using the management means; and means for storing the new logical block in a sector in which no logical block is stored in the determined stripe.

Images