JPH11305945A - Data storage controller and data storage control method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To quarantee a read rate and to minimize the decrease in the availability of a disk without degrading data quality by deciding whether or not to add restoration data to the data based on the read rate requested for inputted data, the read rate from a storage means and required retrial time. SOLUTION: A disk read rate information part 4 decides a read rate requested for respective data and a disk zone speed information part 5 calculates a specified transfer speed provided in a writing disk zone. A retrial time information part 6 calculates a retrial time required for error retrial, and a comparison part 7 compares the read rate + the retrial speed with the transfer speed. Since an error processing selection part 3 decides whether to store the data in the disk after adding the restoration data or without adding them based on the compared result of the comparison part 7, the need of adding the restoration data each time is eliminated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data storage control device and a data storage control method for temporarily storing a large amount of data such as images and sounds and transferring the data to an output device in real time.

[0002]

2. Description of the Related Art Conventionally, in a device for handling images, sounds, and the like, a disk array device using a plurality of disk devices (hereinafter, referred to as disks) has been used as means for accumulating a large amount of data and realizing high-speed input / output. (Hereinafter referred to as a disk array). In these applications, real-time properties are required, such as sequentially switching files (planes) at a predetermined timing and transferring a predetermined amount of data.

[0003] However, in general, in a disk, an automatic retry process in which re-execution is repeated a certain number of times when a reading error occurs is set, and there is a problem that the real-time property cannot be guaranteed by the time required for the retry process. Specifically, the time required for one re-execution is one rotation waiting time of the disk, and it takes 8.3 msec at maximum for a disk of 7200 rpm. On other disks, since re-execution is performed up to 30 times, about 250 ms
ec. The frequency of occurrence of the retry process is once for reading 10 ¹⁰ bits from the disk. The read error rate when performing retry processing is 1/1
0 ¹⁴ bits, error rate when no retrying process is its approximately 10,000 times.

[0004] For example, when a printer that outputs 8 Mbytes per plane in one second is considered, theoretically, it occurs about once every 140 sheets, and the disk transfer speed drops by 30% at worst in the plane where retry processing has occurred. However, the plane cannot be switched at a predetermined timing. To avoid this,
In general, a plurality of full plane size output buffers are required (at least two planes are required for continuous output), and output is started after data of one plane or more is accumulated.

[0005]

However, in the above-mentioned prior art, in order to increase the reliability, a memory for many planes is required, which leads to an increase in cost. Therefore, as a process for the above-described error itself, a method is known in which a RAID device (levels 3 to 5) adds parity data at the time of writing, and recovers error data by using parity data + other normal disk data when an error occurs. However, there is no real-time property,
It is not suitable for applications such as the printer described above.

As a method for solving this real-time property, for example, Japanese Patent Application Laid-Open No. Hei 8-202504 discloses a method in which parity data is added to another disk and the parity data is read in parallel at the time of reading, so that parity recovery can be performed when an error occurs. A technique for performing in real time is disclosed. In the case of this method, there is a problem that one or more extra disks are required for parallel processing, which leads to an increase in cost. Further, since parity data is always added, for example, when n disks are used, the usage rate is (n−1) / n.
Therefore, there is a problem that the use efficiency of the disk is reduced.

In addition, instead of the parity data, for example, Japanese Patent Application Laid-Open No. 8-194585 discloses a technique using a highly compressed data disk. In this case, the use efficiency is improved as compared with the above-described technique, but since the error occurrence plane is restored by the highly compressed data, deterioration in quality (image quality or the like) occurs. It is difficult to use it for applications such as printers where output remains. On the other hand, in the case of an application in which the amount of data to be handled is further increased, the reading speed of the disk (for example, 10 Mbytes / sec or less) is not enough for the required transfer speed (several tens of Mbytes / sec).
A method of compressing data (using irreversible compression means such as JPEG for image data), guaranteeing the worst compression ratio, and expanding and outputting in real time is also used.

The compression rate of the compressed data changes depending on the arrangement of bits (a picture in the case of a picture) of each file, and the amount of data generated for each file changes, so that the required transfer speed also differs for each file. . In the disk itself for storing the compressed data, the data transfer speed is set to 1.
There is a difference of 5 to 2 times. Therefore, depending on the data transfer speed required for output and the zone of the disk to which data is written, there is a problem that, even when the retry time is included, data can be read within a predetermined time or not in time. is there.

The present invention has been made in view of the above circumstances, and guarantees the reading speed of a file per unit time even when a disk error occurs, and reduces the disk usage efficiency without deteriorating data quality. It is an object of the present invention to provide a data storage control device and a data storage control method capable of minimizing deterioration.

[0010]

In order to solve the above-mentioned problem, according to the first aspect of the present invention, an input means for inputting data from a host device and a storage device having different data reading speeds depending on storage locations. A medium, storing means for storing data input from the input means, output storage means for temporarily holding data read from the storage means, and data input from the input means. Error processing selecting means for determining whether to add restoration data to the data based on a required read speed, a read speed when reading data from the storage means, and a retry time required for retrying the storage means; And storage means for storing the restoration data in the storage means when the error processing selection means determines to add the restoration data. When an error occurs when the data to which the restoration data is not added is read from the storage unit, a retry control unit that performs a retry process on the storage unit, and the data to which the restoration data is added is A recovery unit that recovers data held in the output storage unit according to the restoration data when an error occurs during reading from the storage unit. .

In order to solve the above-mentioned problem, in the invention according to the sixth aspect, a step of acquiring a read speed required for input data, and a storage means having a different read speed depending on a storage location of the data. Obtaining a storage position at which the data is stored, calculating a read speed at the time of reading from the storage unit based on the storage position at which the data is stored, and causing an error at the time of reading. Calculating a retry time required for retrying, and determining whether to add restoration data to the data based on the read speed required for the data, the read speed, and the retry time. When it is determined that the restoration data is added, the restoration data is generated and stored in the storage unit. A step of performing a retry process for the storage unit when an error occurs when reading data to which the restoration data is not added from the storage unit, and rewriting the data to which the restoration data is added. Restoring the data according to the restoration data if an error occurs when the data is read from the storage means.

According to the present invention, when the read speed differs depending on the disk zone in which data is stored, the transfer speed differs depending on the disk zone, the transfer speed required for the data to be handled per unit time, and The retry times are compared, and the processing at the time of occurrence of the error is selected between retry processing of the disk (when it is determined that the retry processing can be performed in time) or restoration using restored data (parity). As a result, it is not necessary to add restoration data every time, and even when a disk error occurs, the file read speed per unit time is guaranteed, and the reduction in disk usage efficiency is minimized without deteriorating data quality. It is possible to keep it to a minimum.

[0013]

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

A. Basic Configuration of Embodiment FIG. 1 is a block diagram showing a basic configuration of a data storage control device according to an embodiment of the present invention. In the figure, a disk array unit 1 includes N (N ≧ 1) disks D1 to DN.
And holds data for one output. The disk control unit 2 includes, for example, a SCSI controller, and controls a disk sequence and data transfer between the CPU and the memory and between the disk and the output buffer in accordance with a command from the CPU 11. The error processing selection unit 3
It is determined whether or not restoration data (parity) is to be added. The error processing selection unit 3 includes a disk read speed information unit 4, a disk zone speed information unit 5, a retry time information unit 6, and a comparison unit 7.

The disk read speed information section 4 determines a read speed required for each data. The read speed required for each piece of data is A / B (bytes / second) from the data amount A (bytes) written to each disk and the number of files (planes) B per unit time processed by the output device.
Becomes Since the value of the uncompressed data is fixedly determined by the configuration of the output device, the required reading speed may be tabulated. On the other hand, in the case of compressed data, the data amount is different each time, and therefore, the calculation is performed at the time of data input.

The disk zone speed information section 5 calculates the transfer speed in the specification of the disk zone to be written.
However, since the transfer speed of the disk zone is a fixed value depending on the disk, in the present embodiment, it may be incorporated as a program in advance. Here, FIG.
(B) is a conceptual diagram for explaining a difference in transfer speed between disk zones, and disk zones # Z0 to # Z0.
FIG. 4 is a conceptual diagram showing an example of correspondence between transfer speeds s1 to sn of Zn and disk addresses AdrZ1 to AdrZn. In FIG. 2A, each disk can be divided into a plurality of concentric zones according to the transfer speed. Here, from the outer circumference to the inner circumference of the disc,
Zones # Z0 to #Zn, the transfer speed of zone # Z0 is s0 (Mbyte / s), the transfer speed of i-th zone #Zi is si (Mbyte / s), and the n-th transfer speed is sn (M
Bytes / s), the relationship between the zone transfer rates is s
0>si> sn.

Therefore, the disk zone speed information section 5
Is the disk address Adr, as shown in FIG.
Z1 to AdrZn are stored in association with disk zones # Z0 to #Zn (or incorporated in a program), and if it is known at which address on the disk data is to be written, that is the disk zone. I understand. If the disk zone is known, the transfer speed in the disk zone can be known.

Next, the retry time information section 6 calculates a retry time required for the error retry. However, the maximum time Tr required for this retry is also fixedly determined by the specifications of the disk, and may be incorporated in advance as a program. In the case of a disk having a rotation speed of Mrpm and the maximum number of retries in specifications is P, the maximum time Tr required for retry is (1 ÷ (M × 60)) × P. Comparison section 7
Compares the read speed + retry speed with the transfer speed. The error processing selection unit 3 determines, based on the comparison result of the comparison unit 7, whether to add the data for restoration and store the data on the disk or not. Note that the error processing selection unit 3 can also be realized by program control executed by the CPU 11.

The restoration data may be generated and added to one of the plurality of disks in the disk array unit 1, or may be generated and added to each of the disks D1 to DN. The former is parity and the latter is restoration data such as actual data. Further, regarding the parity generation, in the present field, real-time property at the time of writing to the disks D1 to DN is not important, so that the realizing means may be either software or hardware. The memory 8 functions as a working area and a buffer for input data when the CPU 11 executes a program.

The output buffer 9 has a size corresponding to the sum of the buffer of the input / output speed difference and the retry time, and temporarily stores the data read from the disk array unit 1 and then outputs the data (not shown). Output to the device. The sector data repairing section 10 has a function of repairing an error sector basically in real time, though the configuration in the repairing section differs depending on whether the parity is added one by one for a plurality of disks or for each disk individually. . The details of the sector data restoration unit 10 will be described later. CPU1
1 controls the entire apparatus. Further, the input interface unit 12 receives data from a higher-level host device (not shown). .

B. First Embodiment Next, FIG. 3 is a block diagram showing a configuration of a data storage control device according to a first embodiment of the present invention. FIG.
The same reference numerals are given to the portions corresponding to and the description is omitted. In the first embodiment, the disk array unit 1 is a single disk D1, and the parity described above is used for the disk D1.
This is the minimum configuration when adding to. Hereinafter, the output device is controlled at B sheets per second, and the maximum retry time at the time of a read error is assumed to be Tr.

Data from the host device is input to the memory 8 from the input interface unit 12. At this time, the input data amount (compression rate information when compressed data is handled) A is obtained from the host device together with the data input. The required read speed S1 (= A / B) of the disk D1 is obtained from the input data amount A, and then the speed S2 = S1 / (1-Tr) obtained by adding a margin of the retry time Tr to the required read speed S1. ). Since the disk zone (address) to be written is determined by the file management program of the disk control unit 2, the read speed S3 of the disk D1 is determined by the disk zone speed information unit 5 according to the disk zone (address).

The error processing selection unit 3 compares the read speed S2 including the retry time with the actual read speed S3 by the comparing unit 7, and when S2 ≦ S3,
Even when a retry process is executed when reading data from the disk D1, the required reading speed can be guaranteed, so that the data can be read from the disk D1 without adding restoration data.
1 is stored. On the other hand, if S2> S3, the required read speed cannot be guaranteed, so the data for restoration is added and the data is stored on the disk D1. When adding restoration data, the restoration data is written to another area on the disk. When reading data, set the disk D1 to an error immediate response setting (in the case of SCSI, Mod
e Select command is issued), and when an error occurs, the fact is notified to the CPU 11 so that retry processing is not performed. By the way, S
The time required for error report + sense data transfer on the CSI bus is about 2 to 3 msec.

FIG. 4 is a conceptual diagram showing the arrangement of each data in the first embodiment. The data for restoration is
It is stored from the disk address Aend. CPU
When 11 issues a command to the disk controller 2, data transfer from the disk D1 to the output buffer 9 starts. Data transfer is performed in units of blocks (for example, 32 Kbytes). If an error occurs during the transfer of the block #L, the check condition is transmitted from the disk D1.
The n status is reported to the CPU 11, and an error is recognized. Subsequently, the sense data is transferred from the disk D1 to the CPU.
11 and the sector address A of the error disk
erd.

The CPU 11 determines the sector address Aer
According to d, the difference Aof from the first sector address =
(Aerd-Asd) is calculated. At this time, all data of the block #L from the disk D1 is output from the output buffer 9
Has been sent to After the transfer of the block #L is completed, the CPU 1
1 is a disk address (Ae) corresponding to the error sector.
nd + Aof), the data sector for restoration is read out, and the address As corresponding to the error sector in the output buffer 11 is read.
The disk controller 2 is instructed to transfer the data to m + (Aof × 512). As a result, only the error portion of the output buffer 9 is repaired. After that, the disk reading is restarted. Incidentally, as described above, the time required for error reporting and sense data transfer on the SCSI bus is 2 to 3 times.
msec and the positioning time is 8.3m
sec (when the restored data is stored in the vicinity, the maximum of one rotation wait (7200 rpm) or less, and the restored data transfer time is:
The transfer time to the output buffer after restoration is 2 to 3 msec is 1 msec or less. Therefore, the total time is about 20 msec including the execution of the control program.

C. Second Embodiment Next, FIG. 5 is a block diagram showing a configuration of a data storage control device according to a second embodiment of the present invention. FIG.
The same reference numerals are given to the portions corresponding to and the description is omitted. The second embodiment is applied to a case where each output system of a full-color printer or the like is temporally separated. The disks D1 to D4 are connected to the disk control units 2a to 2d via a disk bus, respectively, and are arrayed by connecting this set to the system bus for the number of colors (4 for CMYK). Each data of CMYK is one disk Di (i = 1 to 4) for each color.
Is stored in The output buffers 9a to 9d are connected to the system bus by the number of colors.

The second embodiment is the same as the first embodiment except that the error processing selection is determined for each disk. That is, the input data amount An (n
= 1 to 4), the read speed S1n (= An / B) required for each of the disks D1 to D4 is obtained, and the speed S2n = S1n / (1−1) when the read speed S1n is added to the margin of the retry time Tr. Tr). Then, from the S2n and the reading speed S3n of each disk from the disk zone speed information section 5, the error processing selecting section 3 checks each of the disks D1 to D4. As a result, optimal error processing is performed for each of the disks D1 to D4 within the same page.

D. Third Embodiment Next, FIG. 6 is a block diagram showing a configuration of a data storage control device according to a third embodiment of the present invention. FIG.
The same reference numerals are given to the portions corresponding to and the description is omitted. In the figure, disks D1 to D4 are connected to a common disk control unit 2 by a disk bus. In the fourth embodiment, when one restoration data (in this case, parity) is added to a plurality of disks as a whole, a buffer of the sector data restoration unit 10 is used together to restore data in real time.

In the third embodiment, the number of output systems is simultaneous or one (one file can be created). Data is,
The data is equally divided and stored in the disk array unit 1 including the disks D1 to D4. In the illustrated example, the disk array unit 1 includes four disks D1 to D4.
It may be an individual. One output buffer 9 is connected to the system bus. Since the read speed S1 required for the disks D1 to D4 is the same for all disks,
From the input data amount A, S1 = (A / N) / B. The operation up to the operation of selecting the error processing is the same as that of the above-described embodiment. Disks for writing parity use the disks D1 to D4 sequentially.

FIG. 7 is a block diagram showing an example of the configuration of the sector data repair unit 10 according to the third embodiment. In the figure, a data input / output unit 20 is an interface with a system bus. Block buffer 21
Holds data in units of transfer blocks corresponding to the disks D1 to D4. The parity decode (exclusive OR) unit 22 recovers error data from the data from the block buffer 21 and the corresponding parity data read from the disks D1 to D4. Sector buffer (FI
The FO) 23 temporarily holds the restored data.
The buffer control unit 24 controls each of the above units and performs sector data restoration.

FIG. 8 is a block diagram showing an example of the configuration of the buffer control section 24. In the figure, an address decoding unit 25 decodes an address on a system bus, guides data from each of the disks D1 to D4 to a corresponding block buffer 21, and selects an internal register. The block buffer control unit 26 operates in synchronization with the write signal (W) on the system bus, controls the writing / reading of the block buffer 21 from the address set in the block buffer ADR register 29, and controls the parity decoding unit 22 To control writing to the sector buffer 23. The sector byte counter 27 controls the number of bytes for error recovery (for one sector). The output buffer control unit 28 updates the sector buffer 2 to update the error in the output buffer 30.
3 and the transfer to the address set in the output buffer ADR register 30 is controlled. The value of the output buffer ADR register 30 is set by the CPU 11 based on the sense data information from the error disk.

Next, the operation after the start of data transfer from the disk to the output buffer according to the third embodiment will be described.
Here, FIG. 9 is a conceptual diagram showing an arrangement of each data when four disks D1 to D4 according to the third embodiment are used. The parity data is stored in the address A of the disk D2.
end2. The data transfer is performed in units of blocks (32 Kbytes), and the data is sent to the output buffer 9 and stored in the block buffer 21 at the same time. If there is no error during each block transfer, the transfer is continued and the contents of the block buffer 21 are overwritten.

As in the first embodiment, if an error occurs in the disk D4 during the transfer of the block #L, for example, the CPU 11 sets the difference Aof = from the head sector address according to the error sector address Aerd4 of the disk D4. (Aerd4-Asd4) is calculated. After the transfer of the block #L of all disks is completed, the CPU 11 sets the address Ao corresponding to the block buffer ADR register.
fb = Aof − ((L−1) × 32K) is stored in the output buffer ADR register in the corrected start address Asm4 + (Aof × 51) of the output buffer 9 corresponding to the disk D4.
Set 2). Further, it instructs the disk control unit 2 to read one sector of parity data from the disk address Aend2 + Aof corresponding to the error sector of the disk D2 to the sector data recovery unit 10.

The buffer control unit 24 synchronizes with the input of the parity data and operates the normal disks D1, D2, D3.
The corresponding sector is read from the address Aofb of the block buffer 21 in which the data of the block #L is held, and the parity decoding unit 22 restores the sector in real time using the sector and the input parity data, and writes the data in the sector buffer 23. After the restoration is completed, the output buffer control unit 28 reads the restoration sector and outputs
The data is transferred to the address indicated by the R register 30, and only the error portion corresponding to the disk D4 of the output buffer 9 is repaired. After that, the disk reading is restarted.

The third embodiment described above is similar to the second embodiment by adding a buffer capacity for absorbing the time difference between each color output on the printer side (may be on the printer side). Can be applied to full color printers. Further, in the third embodiment, the error processing selection is determined for each disk. However, it is also possible to check all disks and add parity data if one disk satisfies S2n> S3n. Further, by adding the parity to the disk corresponding to the color data having the smallest data amount in the CMYK data, it is possible to reduce the deviation of the disk usage rate in the disk array.

In the above-described embodiment, the seek time can be reduced by writing the restored data from the disk address following the actual data. Also,
When the data to be handled is non-compressed data, the input data amount is constant for each output size. Therefore, when the output device is determined, the required read speeds S1 and S2 are also uniquely determined. Therefore, since the reading speeds S1 and S2 can be handled as fixed values, the reading speeds can be realized without the calculation means for the input data amount as described in the first to third embodiments.

[0037]

As described above, according to the present invention, according to the present invention, the read speed required for input data, the read speed for reading data from the storage means, and the retry required for retry are provided by the error processing selecting means. Based on the time, it is determined whether or not to add the restoration data to the data, and if it is determined that the restoration data is to be added, the restoration data is added to the data, and an error occurs at the time of reading. In this case, if the data has the restoration data added thereto, the restoration means restores the data according to the restoration data. If the data does not have the restoration data added thereto, the retry control means executes the normal restoration. Since retry processing is performed, it is not necessary to add data for restoration every time. Guaranteed readout rate per time, and without degrading the quality of the data has the advantage that it is possible to minimize the reduction in the use efficiency of the disc.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a basic configuration of a data storage control device according to an embodiment of the present invention.

FIG. 2 is a conceptual diagram for explaining a difference in transfer speed between disk zones, and disk zones # Z0 to # Z0.
FIG. 4 is a conceptual diagram showing an example of correspondence between transfer speeds s1 to sn of Zn and disk addresses AdrZ1 to AdrZn.

FIG. 3 is a block diagram illustrating a configuration of a data storage control device according to the first embodiment of the present invention.

FIG. 4 is a conceptual diagram showing an arrangement of each data in the first embodiment.

FIG. 5 is a block diagram illustrating a configuration of a data storage control device according to a second embodiment of the present invention.

FIG. 6 is a block diagram illustrating a configuration of a data storage control device according to a third embodiment of the present invention.

FIG. 7 shows a sector data restoration unit 1 according to a third embodiment.
FIG. 2 is a block diagram showing an example of a configuration of 0.

FIG. 8 is a block diagram illustrating a configuration example of a buffer control unit 24 according to a third embodiment.

FIG. 9 shows four disks D1 to D3 according to the third embodiment.
It is a conceptual diagram showing arrangement of each data when D4 is used.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Disk array part (storage means) 2 Disk control part (retry control means) 3 Error processing selection part (error processing selection means) 9 Output buffer (output storage means) 10 Sector data restoration part (restoration means) 11 CPU (storage) Means) 12 input interface unit (input means)

Claims (6)

[Claims] An input unit for inputting data from a host device, a storage medium having a different data reading speed depending on a storage location, and a storage unit for storing data input from the input unit; Output storage means for temporarily holding data read from the means, readout speed required for data input from the input means, readout speed when reading data from the storage means, and Error processing selecting means for deciding whether or not to add restoration data to the data based on a retry time required for retrying; and A storage unit for storing data in the storage unit; and reading out data from the storage unit without the restoration data added thereto. When an error occurs, retry control means for performing a retry process for the storage means, and when an error occurs when reading the data to which the restoration data is added from the storage means, A data storage control device comprising: a restoration unit that restores the data held in the output storage unit according to the data. 2. The storage means comprises a plurality of storage media; the storage means stores color data input from the input means in the plurality of storage media for each color; 2. The data storage control device according to claim 1, wherein it is determined whether or not the restoration data is added to each of the plurality of storage media. 3. The storage means comprises a plurality of storage media, and the storage means, when the error processing selection means determines to add restoration data, one storage medium for the plurality of storage media as a whole. 2. The data storage control device according to claim 1, wherein data is stored in any one of the plurality of storage media. 4. The data storage control device according to claim 1, wherein said restoration unit restores the data from the restored data on a sector-by-sector basis. 5. The data storage control device according to claim 1, wherein the storage unit stores the repair data at a position adjacent to the data. 6. A step of obtaining a read speed required for input data, and a step of obtaining a storage position where the data is stored in a storage unit having a different read speed according to a storage location of the data. A step of calculating a read speed at the time of reading from the storage unit based on a storage position where the data is stored; a step of calculating a retry time required for a retry when an error occurs at the time of reading; Determining whether to add restoration data to the data based on the required read speed, the read speed, and the retry time; and if it is determined that the restoration data is to be added, Generating data and storing the data in the storage unit; and data to which the restoration data is not added. When an error occurs when reading from the storage means, a step of performing a retry process for the storage means; and when an error occurs when the data to which the restoration data is added is read from the storage means, And a step of restoring the data according to the restoration data.