JP2002342038A - Disk unit that controls the execution order of commands - Google Patents

Abstract

PROBLEM TO BE SOLVED: To shorten access time and to improve the performance of a disk device by controlling the execution order of commands with both of read commands and write commands as objects. SOLUTION: At the time of read and write, read data and write request data are read to a cache 104 and managed. On condition that the write request data on the cache are not to be subscribed by the read data read by the read command, the execution order of the read command and the write command is controlled so as to shorten the access time.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention controls the execution order of commands so as to reduce the access time required when writing write request data to a disk and reading read request data from the disk, and improves the performance of the disk device. Related to the technology for improving.

[0002]

2. Description of the Related Art In a hard disk, a cache is provided, write request data transferred from a host is temporarily stored in a cache, and the cache is managed. The difference between the write request data write speed and the disk write request data write speed (slow compared to the write request data transfer speed from the host) is absorbed.

Specifically, write request data of a received write command is transferred to a cache for management. When data is written to a continuous area on a disk by a series of continuous write commands, the data is written in a single write process. To execute the write command.

Further, based on management information of write request data transferred to a cache, as in a “high-speed write cache disk device” described in Japanese Patent Application Laid-Open No. H11-265262,
There is a method of shortening the execution time of a series of write commands by controlling the execution order of the write commands so as to minimize the access time when processing a series of write commands.

[0005]

In order to reduce the execution time of a command, it is important to reduce the time required to access a specific area on a disk specified by the command. To shorten the time required to access a specific area on the disk specified by the command,
It is effective to execute the command after controlling the execution order of the commands so as to minimize the access time.

In the "high-speed write cache disk device" described in Japanese Patent Application Laid-Open No. H11-265262, the order of command execution is controlled in order to reduce the access time to a specific area of the disk. Control is limited to write commands.

Therefore, when it becomes necessary to alternately process the write command and the read command, there is a problem that the access time cannot be reduced.

An object of the present invention is to control a command execution order for a read command and a write command,
The purpose is to shorten the access time and improve the performance of the disk device.

[0009]

In order to achieve the above object, a disk device according to the present invention is designed to prevent a write request data in a cache from being overwritten by data read by a read command. The command execution order is controlled so as to shorten the access time for the command and the read command.

That is, the present invention provides, as a first aspect,
A disk device that reads and manages read data and write request data into a cache at the time of reading and writing, and performs access by assuming that write request data in the cache is not overwritten by data read by a read command. A disk device characterized by having means for determining the order of command execution so that time is reduced.

[0011] The command execution order control means includes:
Writing of data to the disk can be stopped or put on standby, and read processing can be given priority over write processing by allowing restart or start at an appropriate timing.

Further, when the disk device receives a read command while data is being written from the cache to the disk, the disk device interrupts the write process and executes the received read command at a timing for shortening access. Can be.

Further, in order to achieve the above object, the disk device of the present invention waits for a certain time to issue the next command when controlling the execution order of the commands, so that the way of issuing the next command is checked. The order of command execution can be determined.

That is, the disk device of the present invention
In the first mode, when a write command is received in a state where there is no unprocessed write command and when a new write command is received while the write processing is stopped, the current head position and the unprocessed The process starts immediately only when the distance to the address of the write command is within the preset fixed distance. Otherwise, waits for the next command to be issued for a fixed time and checks the way the next command is issued. The command execution order control means further comprises means for determining the execution order of
Provided is a means for starting or resuming the processing of a standby or stopped write command at a timing to minimize the access time.

[0015]

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

FIG. 1 shows an example of a disk drive to which the present invention is applied. A disk device to which the present invention is applied includes a program ROM 101 for mounting a prefetch control program, and a control processor 10 for reading and executing a control program on the ROM.
2. A magnetic disk controller 103 for controlling data transfer between the host and the cache, and between the cache and the disk, and a cache 104 for temporarily writing read data / write request data (the cache includes data read from the disk. Cache data management table (for read data) 105 for managing information, and cache data management table (for write data) 10 for managing information on write request data transferred from the host
6), when reading and writing data, a servo control unit 107 for controlling movement of the head to a designated position, and a voice coil motor (VCM) 108 for moving the head in accordance with an instruction of the servo control unit. A motor driver 109 for controlling the rotation of the disk, a selector 110 for selecting only a specified head signal from magnetic signals read from the head, and converting analog data sent from the selector into digital data or from a magnetic disk controller. It comprises a signal processing unit 111 for converting the transmitted digital data into analog data.

Next, the operation in the read command processing will be described.
Description will be made based on the configuration of FIG.

The host transmits read command data to the disk via a bus connected to the disk.
FIG. 2 shows a configuration of a command issued at the time of read processing. The read command includes a code 21 indicating the read command, a logical block address (hereinafter, LBA) 22 of the data to be read, and a host request data transfer number 23 indicating the size of the read request data.

When the disk device receives the read command, it uses the cache data management table (both for read data and write data) in the cache memory to search whether the requested data exists in the cache. FIG. 4 shows the configuration of the cache data management table (common for read data and write data). The cache data management table includes a start LBA 41 of data written to the cache, a start address on the cache 42, and a data size 43. At the time of a hit search, the cache data management tables (both for read data and for write data) 105 and 106 are used to check whether or not all of the read request data exists in the cache. The start address of the host request data on the cache is calculated from the cache start address of the data on the cache, and the host request data is read from the start address and transferred to the host. If the host request data does not exist in the cache, the program ROM 101
Using the logical address translation program implemented in
The LBA is converted to a physical address indicating an arrangement on the disk, and then the position of the physical address is detected by using the servo control unit 107 and the signal control unit 111 is controlled to transfer the data specified by the host to the magnetic disk. Control device 103
Transfer to The magnetic disk control device writes the transferred data to the cache 104 and, at the same time, transfers the requested data to the host via the bus 112 connected to the magnetic disk device, and further stores information about the data written to the cache as cache data. It is registered in the management table (for read data) 105.

The operation in the write command processing will be described based on the configuration shown in FIG.

The host transmits write command data to the disk via a bus connected to the disk.
FIG. 3 shows a configuration of a command issued at the time of a write process. The write command includes a code 31 indicating the write command, and an LB indicating the position of the write request data on the disk.
A32, a host request data transfer number 33 indicating the size of the write request data, and data 34 to be written to the disk.

Upon receiving the write command, the magnetic disk control device 103 writes the write request data from a preset write request data writeable position on the cache. However, the write request data in the cache must be held in the cache until it is written to the disk. Therefore, the cache is full of unwritten write request data, and there is no space for writing new write request data. In this case, it is necessary to stop the data transfer from the host to the cache, wait until the write request data in the cache is written to the disk, and a free area is created in the cache. Further, since the write-enabled position of the write request data on the cache is set before receiving the write command, the magnetic disk control device 103 can transfer the write request data to the cache at the same time as receiving the write command. . Magnetic disk controller 1
When all the write request data is transferred to the cache, information about the write request data in the cache is stored in the cache data management table (for write data) 1.
Register at 06.

When data on the disk, which is the transfer source of the read data on the cache, is updated by the write command, the data on the cache data management table (for read data) 105 is guaranteed in order to guarantee data consistency. Of the read data updated by the write data.

When a plurality of write commands update the same area even in an unexecuted write command, only the latest write request data is always stored in the cache data management table (write data ) 106 so as to be registered in the cache data management table (for write data) 106.

Next, based on the information in the cache data management table (for write data) 106, the position specified by the command is detected in the same manner as in the read command processing, and the magnetic disk controller 103 and the signal processor 11
1 to write the write request data written to the cache to the disk. It is not necessary to write the write request data on the cache to the disk in the order in which the write commands are received.

The manner of processing the write command and the read command is as described above. When the disk receives the write command and transfers the data to the cache, the disk issues the command to permit the host to issue the command. A new command can be received while writing is in progress.

During the read processing, the next command is not issued unless the host confirms that the host has acquired the read request data, so that the next command is not received.

Hereinafter, an example of means for controlling the execution order of commands so as to reduce the access time for write commands and read commands in the present invention will be described with reference to the flowcharts of FIGS.

In FIG. 5 and FIG. 6, when not only a command issued by the host but also a plurality of commands can be processed at once by a single access, the word command is used for the command group. used.

FIG. 5 shows a method of processing a read command when a read command is received during a write process.

If a read command is received during the write process, the cache data management table (both for read data and for write data) 105.1 on the cache
06 to retrieve host request data (step 50).
1) If the host-requested data by the read command completely exists in the cache (hereinafter, this is referred to as a hit), the read-request data requested by the host is cached while the data writing to the cache is continued. To the host (step 502).

Further, it is checked whether the host request data by the received read command partially exists in the cache by searching the cache data management table (for write data) 106 (hereinafter, this is called a half hit) ( Step 503), if it is determined that the read request data of the received read command partially exists in the unwritten write request data, all the write request data not yet written to the disk on the cache is written to the disk (step 503). 504) Read processing is performed (step 508).

If the read request data of the read command received while writing the write request data from the cache to the disk does not exist at all in the cache, the address of the read request data by the read command on the disk is It is checked whether the position of the read request data is within a certain distance from the current head position by checking whether the position of the head currently writing is within a predetermined distance range (step 505). Even when the read request data is read into the cache, it is checked whether unwritten write request data in the cache is overwritten by the read request data by the read command (step 506). When the write command is completed, stop the write process. Step 507), performs a read process according to the read command (step 508).

If the physical address of the read request data according to the received read command is not within a certain distance from the position of the head currently writing the write request data, the write command processing currently being executed is continued. (Step 510) It is checked whether or not there is an unprocessed write command even after finishing the currently executed write command processing (Step 511). If there is an unprocessed write command, the unprocessed write command is newly added. The write command is executed (step 512), and the process from step 505 is repeated for the write command. If there is no unprocessed write command after the end of the currently executed write command,
The received read command is processed (step 508).

If unread write request data in the cache is overwritten by the read request data by the read command, the currently executed write command processing is continued (step 509). The processing from step 505 is performed on the write command newly started after the processing is completed.

Hereinafter, a method of processing a write command will be described with reference to FIG. The method of processing a read command during a write process has been described with reference to FIG.

When the disk receives the write command,
To automatically transfer data to the cache from the pre-set write request data writable position,
Hereinafter, a write processing method other than transfer of write request data to the cache will be described.

In the disk device according to the present invention, upon receiving a write command, it is first checked whether there is an unprocessed write command (step 602). If there is no unprocessed write command, the write process for the received write command is immediately performed. , And waits for the next command to be issued (step 603). However, it waits for the issuance of the next command for a preset fixed time, and if the next command is issued within the fixed time, it is checked whether the command next to the received write command is a write ( Step 601) The processing after step 601 is performed on the next command after the write command without performing the write processing of the write command. If the next command is not issued within the fixed time, the write command is started (step 604).

It is checked whether or not there is an unprocessed write command. If there is an unprocessed write command, it is checked whether or not the write process is waiting or stopped (step 60).
5). If the write process is waiting or stopped, it is checked whether there is an area in the cache for writing the write request data of the write command in the cache (step 606), and the write request data of the write command is stored in the cache. If there is no area for writing, the write processing is started or restarted (step 60).
4).

Further, it is checked whether or not the write processing is waiting or stopped and there is an area in the cache for writing the write request data of the write command to the cache (step 606). If there is an area, it is checked whether any of the write commands on standby or stopped accesses an area within a certain distance from the current head position (step 60).
7). At this time, if any of the waiting or stopped write commands accesses an area within a certain distance from the current head position, the processing of the write command for accessing the area within the certain distance is started (step S1). 60
8). Also, if there are no write commands in standby or stop that access the area within a certain distance from the current head position, the number of commands in standby or stop can be set to standby or stop in advance. It is checked whether the number of commands has reached (step 609). If the number of write commands on standby or stopped exceeds the allowable number of standby or stop commands, write processing is started or restarted (step 604). If the number of waiting or stopped write commands is less than the number of standby or stop commands,
It waits for the next command to be issued (step 603), and performs the processing from step 603.

If a write command is received during the continuation of the write process, it is checked whether or not the cache has an area for writing the write request data of the received write command (step 610). Wait for an area to be created (step 611), and continue the write processing (step 612). If there is an area in the cache where the write request data of the received write command is to be written, the write process is continued (step 612).

Next, an example of a command pattern for shortening the access time by the disk device of the present invention will be described with reference to FIGS. The vertical axis indicates the cylinder address indicating the position on the disk of the request data of the issued command. The horizontal axis indicates the order of command execution. The symbol R indicates a read command, and the symbol W indicates a write command. The numbers indicate the order in which commands are issued for each of the read and write command types.

In the example shown in FIGS. 7 and 8, the read command of R1 is received in the state where no command has been received before, the write command of W1 is received after the read command processing of R1, and the read command of R2 is further received. Receiving read command, R2
7 is a processing example of a command when a write command of W2 is received after the read command processing of FIG.

7 and 8 show the access time on the assumption that the seek time increases in proportion to the seek distance.

FIG. 7 shows an example in which the above-mentioned series of commands is processed by a disk device to which the present invention is not applied. In the embodiment of FIG. 7, upon receiving W1 after processing R1, seek is started immediately and write processing for W1 is started. Upon receiving R2 during the processing of W1, it waits for the end of the write processing of W1, and then starts the processing of R2.
Also, after completing the processing of R2, as soon as W2 is received, W2
Is started, and when the processing of W2 ends, all the processing ends. The access time in FIG.
7 shows the total access time required for processing the command group from to W2. The access time in the disk device to which the present invention is not applied is determined by the seek time T1 (701) generated when W1 is processed after the processing of R1 and W1.
Seek time T generated when R2 is processed after the processing of
2 (702), the total time of T3 (703) generated when W2 was processed after the processing of R2 was completed.

On the other hand, FIG. 8 shows an example in which the above-mentioned command group from R1 to W2 is processed by the disk device to which the present invention is applied. After the processing of R1, when W1 is received, there is no unprocessed write command. Therefore, the write processing for W1 does not start immediately, but waits for the issuance of the next command for a certain time.

When R2 is received within a certain waiting time (assuming that the host issues an R2 command within the certain time),
Since R2 is within a certain area from the head position after the R1 processing (the access position by R2 is within a predetermined distance from the head position after the end of R1), R2 is executed without executing W1. Execute.

When W2 is received after the processing of R2 is completed,
Both W1 and W2 are not within a certain distance from the head position after execution of R2 (W1 and W2 are both
(It is assumed that it is not within a certain distance from the head position after executing R2)
Therefore, it waits for the next command for a certain period of time.

After that, since the next command is not issued even after waiting for a certain time, W1 and W
2 is written to the disk.

The access time 'in FIG. 8 indicates the total access time required for processing the series of commands from R1 to W2.

The access time 'in the disk device to which the present invention is applied is determined by the seek time T1' (801) generated when the processing for R2 is started after the R1 processing is completed, and when the W1 and W2 are processed after the R2 processing is completed. Is the total time of the seek time T2 '(802) that occurred in the second time.

[0052]

As described above, according to the disk device of the present invention, the access time is shortened by controlling the execution order of commands for both read and write commands. Performance can be improved.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of a magnetic disk drive to which the present invention is applied.

FIG. 2 is a diagram showing a configuration of a read command issued by a host.

FIG. 3 is a diagram illustrating a configuration of a write command issued by a host.

FIG. 4 is a diagram showing a cache data management table (common for read data and write data) for managing data on the cache;

FIG. 5 is a flowchart illustrating a read command processing method when a read command is received in a state where there is an unprocessed write command;

FIG. 6 is a flowchart showing processing when a read command is received in the presence of an unprocessed write command and when a write command is received;

FIG. 7 is a diagram illustrating a command execution example in a disk device to which the present invention is not applied.

FIG. 8 is a diagram showing a command execution example in a disk device to which the present invention is applied.

[Explanation of symbols]

101: Program ROM, 102: Control processor, 1
03: magnetic disk controller, 104: cache memory, 105: cache data management table (for read data), 106: cache data management table (for write data), 107: servo control unit, 108: voice coil motor (VCM) , 109 ... motor driver, 1
10 ... selector, 111 ... signal processing unit.

 ────────────────────────────────────────────────── ─── Continued from the front page (72) Inventor Manabu Nishikawa 2880 Kozu, Odawara-shi, Kanagawa F-term in the Storage Division, Hitachi, Ltd. (Reference) 5B005 JJ11 MM11 NN73 5B065 BA01 CA15 CE11 CH01 CH15 ZA15 5D044 BC01 CC05 DE02 DE12 DE17 DE23 DE29 DE38 DE48 EF03 FG10 HH07 HL01 JJ03

Claims (5)

[Claims] 1. A disk, a head, and a cache memory are provided, and read request data and write request data are read and managed in a cache at the time of reading and writing, and the write request data in the cache is read data read by a read command. A disk device having means for determining the execution order of a read command and a write command so that an access time is shortened on the premise that overwriting is not performed by the disk drive. 2. The disk device according to claim 1, wherein the command execution order control means can stop or wait for writing of data to the disk, and can restart or start at an appropriate timing. A disk device characterized by giving priority to read processing over write processing. 3. The disk device according to claim 2, wherein when a read command is received while writing the write request data from the cache to the disk, the received read command is written at a timing to shorten the access. A disk device characterized in that it can be executed by interrupting processing. 4. The disk device according to claim 2, wherein a write command is received in a state where there is no unprocessed write command, and a new write command is received while the write processing is temporarily stopped. , The process is started immediately only when the distance between the current head position and the address of the unprocessed write command is within a preset fixed distance, otherwise, the next command is issued for a fixed time until the next command is issued. A disk unit that has a means to determine the order of command execution based on how the command is executed. 5. The disk drive according to claim 4, wherein the processing of the write command that is on standby or stopped can be started or restarted at a timing for minimizing the access time. .