JP6662530B1 - Control device, disk array system and data input / output method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a control device for improving availability and read performance of a disk array device. A controller writes writing target data composed of a plurality of unit data arranged side by side into each of a plurality of hard disks, and reads from each of the plurality of hard disks in parallel for each unit data. In the case of reading the data, the reading unit simultaneously reads the unit data adjacent to each other in the row of the data before writing from the different hard disks. [Selection diagram] Fig. 1

Description

  The present invention relates to a control device, a disk array system, and a data input / output method.

  RAID (Redundant Array of Inexpensive Disks) is a technology for improving the availability and input / output performance of HDDs (Hard Disk Drives). For example, in a striping type RAID system (RAID 0), writing data can be distributed to a plurality of disks to improve the writing speed. In a mirror type RAID system (RAID 1), availability is improved by writing the same data to a plurality of disks.

  As a related technique, Patent Document 1 discloses that when data larger than a predetermined size is read from two HDDs configured with RAID1, the former data is read from one HDD and the latter data is read from the other HDD at the same time. Discloses a technique for improving the reading speed.

JP 2017-16579 A

  In RAID1, since data is read only from one HDD at the time of reading, availability is improved but read performance is not improved. In RAID0, since different data is written to a plurality of HDDs, read performance is improved, but availability is not improved.

  Therefore, an object of the present invention is to provide a control device, a disk array system, and a data input / output method that solve the above-mentioned problems.

According to one aspect of the present invention, the control device is configured to write data to be written, in which a plurality of unit data are arranged side by side, in a writing unit that writes the data to each of the plurality of hard disks, and in parallel from each of the plurality of hard disks. A reading unit that performs reading for each unit data, wherein the writing unit writes to each of the hard disks without changing the arrangement of the data, and reads the data. each of the unit data to be continuous at the arrangement of the data, and read out simultaneously from 1 by different sectors positions of the plurality of hard disks, the reading unit detects an abnormality in the first sector position one of the hard disk, In any of the remaining hard disks, a read target sector is set at the first sector position, and the hard disk is read. For the hard disk other than a 1 by different sector position relative to the first sector position is set as a new read target sector.
Further, according to another aspect of the present invention, the control device includes: a writing unit that writes data to be written, in which a plurality of unit data are arranged in a row, into each of the plurality of hard disks; And a reading unit that performs reading in units of unit data in parallel, and wherein the writing unit performs reading in sector order starting from the same sector position in each of the plurality of hard disks. When changing the arrangement of the unit data so that the continuous unit data can be simultaneously read in the arrangement of the data before writing, writing the data to each of the plurality of hard disks, and reading the data, the reading unit includes: Each of the continuous unit data in the arrangement of the data before writing is stored in a different one of the hard disks. When reading is performed simultaneously in the sector order, and the reading unit detects an abnormality in one sector position of one of the hard disks, one of the remaining hard disks is written in the sector position where the abnormality is detected. The sector position where the same unit data is written is set as a new read target sector.

  According to one aspect of the present invention, a disk array system includes a plurality of hard disks and the above-described control device.

According to another aspect of the present invention, a data input / output method includes a step of writing data to be written, in which a plurality of unit data are arranged side by side, to each of a plurality of hard disks; Reading from the hard disk without changing the arrangement of the data, and performing the reading when reading the data. at, each of the unit data to consecutive in a sequence of the data before the write, and read out simultaneously from 1 by different sectors positions of the plurality of hard disks, in performing the read, first one of the hard disk If an error is detected at one sector position, it will be transferred to one of the remaining hard disks. Te, and sets the read target sector in the first sector position, the for the hard disk other than the hard disk, one each different sector position relative to the first sector position is set as a new read target sector.
According to another aspect of the present invention, a data input / output method includes a step of writing data to be written, in which a plurality of unit data are arranged side by side, to each of a plurality of hard disks; And reading in units of unit data in parallel from the above, wherein in the writing step, the reading is performed in the order of sectors starting from the same sector position in each of the plurality of hard disks in the reading step. In a case where the arrangement of the unit data is changed so that the continuous unit data can be simultaneously read in the arrangement of the data before writing, and the data is written to each of the plurality of hard disks and the data is read, the reading is performed. In the step of performing In the step of simultaneously reading out each of the continuous unit data from different hard disks in the order of sectors and detecting an error at one sector position of one hard disk, any one of the remaining hard disks is detected. In the above, the sector position where the same unit data is written as the one written in the sector position where the abnormality is detected is set as a new read target sector.

  According to the present invention, it is possible to improve the availability and read performance of a disk array system.

It is a figure showing an example of the information processing system concerning each embodiment of the present invention. FIG. 3 is a first diagram illustrating a data input / output method according to the first embodiment of the present invention. FIG. 4 is a second diagram illustrating the data input / output method according to the first embodiment of the present invention. FIG. 6 is a third diagram illustrating the data input / output method according to the first embodiment of the present invention. FIG. 7 is a diagram illustrating a rebuild process according to the first embodiment of the present invention. FIG. 7 is a first diagram illustrating a data input / output method according to a second embodiment of the present invention. It is a figure showing an example of the data storage order by a second embodiment of the present invention. FIG. 11 is a second diagram illustrating a data input / output method according to a second embodiment of the present invention. FIG. 13 is a third diagram illustrating a data input / output method according to the second embodiment of the present invention. It is a figure explaining a rebuilding process by a second embodiment of the present invention. FIG. 11 is a diagram illustrating another example of a data storage order according to the second embodiment of the present invention. It is a figure showing the minimum composition of the control device in each embodiment of the present invention.

Hereinafter, a disk array system according to each embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a diagram illustrating an example of an information processing system according to each embodiment of the present invention.
1, the information processing system 100 includes a host device 1, a RAID controller 2, and a disk array device 3.
The host device 1 includes an OS (operating system) 11, a driver 12, and a CPU (central processing unit) 13. The OS 11 executes an IO command issued by an application or the like executed by the host device 1 and instructs writing and reading to and from the disk array device 3 composed of hard disk drives (HDDs) # 0 and # 1. The driver 12 is a device driver of the RAID controller 2. The OS 11 and the driver 12 are software. The OS 11 and the driver 12 exhibit their functions when the CPU 13 reads and executes the programs of the OS 11 and the driver 12 stored in the disk array device 3, for example.

  The RAID controller 2 includes a write controller 21 and a read controller 22. The write controller 21 and the read controller 22 are software (firmware) having functions necessary for RAID1 (mirror method). As described later, the write controller 21 and the read controller 22 have a function of accelerating the read processing in addition to the data redundancy function based on RAID1. The RAID controller 2 is connected to the disk array device 3. The write controller 21 and the read controller 22 execute an IO process on the HDD # 0 and the HDD # 1.

  The disk array device 3 includes HDD # 0 and HDD # 1. HDD # 0 and HDD # 1 constitute one logical disk that guarantees data redundancy.

  Next, the flow of the IO processing in the information processing system 100 will be described. When the OS 11 issues an IO command in the host device 1, the driver 12 receives the IO command. The driver 12 outputs the IO command to the RAID controller 2 and requests execution of the IO. When the IO instruction is a WRITE process, the write controller 21 writes the same data to HDD # 0 and HDD # 1. This ensures data redundancy, as in RAID1. When the IO instruction is a READ process, the read controller 22 simultaneously reads a part of the data to be read from each of the HDD # 0 and the HDD # 1 in parallel. In the case of RAID1, all of the data to be read is read from one HDD, so that the read controller 22 can read data at twice the speed of RAID1.

<First embodiment>
Next, the processing of the RAID controller 2 in the first embodiment will be described in detail.
FIG. 2 is a first diagram illustrating a data input / output method according to the first embodiment of the present invention.
In FIG. 2, arrows and numeral strings shown together with HDD # 0 and HDD # 1 indicate the data structure of write data or read data. For example, data “01234567” input to the write controller 21 is such that the data to be written is “data 0”, “data 1”, “data 2”, “data 3”, “data 4”, “data 5”. , “Data 6”, and “data 7”. Each of data 0 to data 7 is called unit data for convenience. In the write target data illustrated in FIG. 2, the unit data is arranged in the order of data 0 to data 7.

(Write data)
When there is a write request of 8 sector length (data 0 to data 7) to the disk array device 3, the write controller 21 writes data of 8 sectors to both HDD # 0 and HDD # 1. . When the write controller 21 acquires the data to be written (data 0 to data 7), the write controller 21 writes the data in HDD # 0 and HDD # 1 in this order. At this time, the write controller 21 sequentially writes the respective unit data in the same sector position of the HDD # 0 and the HDD # 1. The lower table of FIG. 2 shows the relationship between the sector position in HDD # 0 and HDD # 1 and the unit data written by write controller 21. As exemplified in this table, “data 0” is stored in sector 0 and “data 1” is stored in sector 1 in both HDD # 0 and HDD # 1. Thereafter, “data N” (N = 2 to 7) is similarly stored in the sector N. In the first embodiment, the operation of the write controller 21 is the same as that of RAID1.

(Normal reading)
Next, the reading process will be described. First, a normal operation (non-failure operation) will be described. When receiving the instruction to read the data having the 8-sector length from the host device 1, the read controller 22 reads the unit data in parallel from different sectors of the HDDs # 0 and # 1. For example, the read controller 22 starts from an even-numbered sector (sectors 0, 2, 4, and 6 in the table of FIG. 2) of HDD # 0 and assigns even-numbered unit data (data 0, 2, 4, 6) At the same time, odd-numbered unit data (data 1, 3, 5, 7) are simultaneously read from odd sectors (sectors 1, 3, 5, 7 in the table of FIG. 2) of HDD # 1. More specifically, the read controller 22 reads from HDD # 0 in the order of sector 0, sector 2, sector 4, and sector 6, and reads from HDD # 1 in the order of sector 1, sector 3, sector 5, and sector 7. . At this time, the read controller 22 simultaneously reads adjacent (continuous) unit data in the arrangement of the unit data in the data at the time of writing. Specifically, the read controller 22 simultaneously reads unit data (data 0 and data 1) from sector 0 of HDD # 0 and sector 1 of HDD # 1. Then, the read controller 22 simultaneously sends the read unit data (data 0 and data 1) to the request source (host device 1).

Next, the read controller 22 simultaneously reads data 2 from sector 2 of HDD # 0 and data 3 from sector 3 of HDD # 1 and sends them simultaneously to the request source. Thereafter, similarly, the read controller 22 simultaneously reads the unit data (data 4 and data 5) from the sector 4 of the HDD # 0 and the sector 5 of the HDD # 1, and sends them simultaneously to the request source. Finally, the read controller 22 reads the unit data (data 6 and data 7) from the sector 6 of the HDD # 0 and the sector 7 of the HDD # 1 at the same time, and sends them simultaneously to the request source. Thus, data for eight sectors is read.
By such processing, data can be read at twice the speed of reading data from only one of the HDDs while ensuring data redundancy.

  In addition, data in information processing has a property of spatial locality (necessary data is grouped in the vicinity of a closer address), and unit data is arranged in a meaningful order, and writing and reading are performed. It is often said. If the read processing is performed by the read controller 22, adjacent unit data (for example, data 0 and data 1, data 2 and data 3, etc.) can be obtained at the same time, so that the processing speed in the OS 11 after reading can be increased. Can be expected. For example, a method has been proposed in which the first half (data 0 to 3) and the second half (data 4 to 7) of eight sector length data are divided and the first half data and the second half data are simultaneously read out one sector at a time (Patent Document 1). 1). In the case of this method, the process of simultaneously reading the data 0 in the first half and the data 4 in the second half and transmitting the data to the request source, and then reading the data 1 and data 5 and transmitting them to the request source are repeated. In the case of such a read method, when data 0 and data 1 are required in the processing of the request source, a wait occurs until the next sector is read. In this regard, according to the present embodiment, continuous unit data can be read at high speed. As a result, occurrence of waiting can be reduced, and processing can be speeded up.

(Read when data error is detected)
Next, an operation in the case where certain unit data (in the following example, data 2 stored in sector 2 of HDD # 0) is broken and cannot be read normally will be described as an example.
FIG. 3 is a second diagram illustrating the data input / output method according to the first embodiment of the present invention.
It is assumed that a read request for data of eight sectors has been made from the request source. As described in the normal case, the read controller 22 first reads data 0 from sector 0 of HDD # 0 and data 1 from sector 1 of HDD # 1 simultaneously, and sends data 0 and data 1 simultaneously to the request source. Send it. Next, it is assumed that the read controller 22 detects that data 2 is corrupted when trying to simultaneously read data 2 from sector 2 of HDD # 0 and data 3 from sector 3 of HDD # 1. Then, the read controller 22 temporarily discards the data 2 and the data 3. Then, this time, the read controller 22 switches the reading destination of the unit data stored in the sector 2 of the HDD # 0 in which the abnormality is detected to the sector 2 of the HDD # 1, and sets the read target sector of the HDD # 0 to the sector 2. Is set to sector 3 which is shifted to the next one. That is, the read controller 22 changes the control so as to simultaneously read unit data from the odd-numbered sector of the HDD # 0 and the even-numbered sector of the HDD # 1 starting from the sector 2 of the HDD # 1. For example, the read controller 22 first reads data 2 from sector 2 of HDD # 1 and data 3 from sector 3 of HDD # 0 at the same time, and simultaneously sends data 2 and data 3 to the request source. Next, the read controller 22 simultaneously reads data 4 from sector 4 of HDD # 1 and data 5 from sector 5 of HDD # 0 and sends them to the request source. The same applies to the subsequent unit data.

  Such read control causes a delay of one sector data read due to once discarding and re-reading data 2 and data 3, but it is possible to read simultaneously continuous unit data from HDD # 0 and HDD # 1. Therefore, high-speed data reading can be continued as in the normal case.

(At HDD failure)
Next, an operation in the case where an entire HDD (HDD # 0 in the following example) has a fixed failure will be described.
FIG. 4 is a third diagram illustrating the data input / output method according to the first embodiment of the present invention.
When the HDD # 0 has failed, the RAID controller 2 closes the HDD # 0 as in the case of the normal RAID1. Then, when there is a read request for data for eight sectors from the request source, the read controller 22 reads data only from HDD # 1. The read controller 22 reads the unit data in order from sector 0 to sector 7 one by one, and sends data 0 to data 7 to the host device 1 each time.
As for writing, the writing controller 22 writes data to the unblocked HDD # 1 as illustrated in the lower table of FIG.
The read and write processes at the time of HDD failure are the same as in RAID1.

(Rebuild processing)
FIG. 5 is a diagram illustrating a rebuild process according to the first embodiment of the present invention.
If the HDD # 0 fails, the HDD # 0 is replaced and the same data as the data written to the HDD # 1 is written to the new HDD # 0. A process of writing data to the HDD # 0 and reconfiguring it into a two-device configuration (RAID1) of the HDD # 0 and the HDD # 1 is called a rebuild process. Specifically, for example, the read controller 22 reads unit data from the HDD # 1 in the order of sectors and sends the unit data to the write controller 21. The write controller 21 writes the same unit data written in the same sector position of the HDD # 1 to each sector of the HDD # 0 where no data is written. When the rebuild process is completed, the data written to HDD # 0 and HDD # 1 match. Therefore, after the rebuilding process, as described with reference to FIG. 2, the same unit data is written in the same sector position of the HDD # 0 and the HDD # 1 at the time of writing, and the even unit and the HDD # 0 of the HDD # 0 are read at the time of reading. It is possible to control to simultaneously read out unit data from one odd-numbered sector. As a result, it is possible to secure redundancy and realize high-speed reading according to the present embodiment.

As described above, according to the present embodiment, while maintaining redundancy by writing the same data to a plurality of HDDs, reading another unit data from a plurality of HDDs at the same time increases the read performance. Can be planned. In particular, since continuous unit data can be read at the same time, it is possible to contribute to speeding up of processing after data reading.
Although the example in which the disk array device is configured by two HDDs has been described, the number of HDDs may be three or more. In this case, the read controller 22 controls to simultaneously read the unit data from the sector positions of the plurality of HDDs shifted by one.

<Second embodiment>
In the first embodiment, control for simultaneously reading unit data from the even sector of HDD # 0 and the odd sector of HDD # 1 has been described. In the second embodiment, control for simultaneously reading unit data from the same sector position of HDD # 0 and HDD # 1 will be described.
FIG. 6 is a first diagram illustrating a data input / output method according to the second embodiment of the present invention.
FIG. 7 is a diagram illustrating an example of a data storage order according to the second embodiment of the present invention.
The description of the second embodiment will also be made on the premise of the configuration of the information processing system 100 illustrated in FIG. However, since the functions of the write controller and the read controller are different from those of the first embodiment, they are described as a write controller 21a and a read controller 22a.

(Write data)
When an 8-sector-length (data 0 to data 7) write request is issued to the disk array device 3, the write controller 21a writes 8 sectors of data to both the HDD # 0 and the HDD # 1. In the second embodiment, the write controller 21a switches the order of the unit data to be written to the HDDs # 0 and # 1 in advance. Regarding the arrangement, when the read controller 22a performs reading in the sector order starting from a certain same sector position in the HDD # 0 and the HDD # 1, the arrangement of data (data 0 to data 7) before writing is performed. Is rearranged so that adjacent unit data can be read simultaneously. For example, with respect to HDD # 0, the write controller 21a extracts, from the pre-write data, unit data to which even numbers are assigned, arranges them in the first half while maintaining their arrangement order, and thereafter adds units to which odd numbers are assigned. The data is arranged continuously without changing the arrangement order, and the data is written to the HDD # 0 as it is. Specifically, the write controller 21a rearranges the order of data 0 to data 7 in the order of data 0, data 2, data 4, data 6, data 1, data 3, data 5, and data 7, and The data is stored in the sector order as it is.

  Similarly, for the HDD # 1, the write controller 21a extracts the odd-numbered unit data and arranges them in the first half while maintaining their arrangement order, followed by the arrangement order of the even-numbered unit data. Line up without changing. Specifically, the write controller 21a rearranges the order of data 0 to data 7 in the order of data 1, data 3, data 5, data 7, data 0, data 2, data 4, and data 6, and The data is stored in the same sector position in HDD # 1 as in HDD # 0. The state after writing is shown in FIGS. 6 and 7A. 7A, data 0 is stored in sector 0 of HDD # 0, data 1 is stored in sector 0 of HDD # 1, and data 2 is stored in sector 1 of HDD # 0. Data 3 is stored in sector 1 of HDD # 1. The same applies to the subsequent sectors. That is, by reading unit data from sectors at the same position in HDD # 0 and HDD # 1 at the same time, continuous unit data in the original data (data 0 to data 7) can be obtained. As described above, in the second embodiment, when the read controller 22a reads the next sector position one by one in order from, for example, sector 0 of HDD # 0 and HDD # 1, it is possible to obtain simultaneously continuous unit data. Rearrange and write. Since the write sector position differs depending on the write sector length, the read controller 22a also writes and holds the sector length information in the HDDs # 0 and # 1 simultaneously (for example, in association with the write start sector position). deep. FIG. 7B shows a write sector position when the length of the data to be written is 4 sectors.

(Normal reading)
Next, a reading process according to the second embodiment will be described. First, a normal operation (non-failure operation) will be described. Upon receiving an instruction to read data having a length of 8 sectors from the host device 1, the read controller 22a reads unit data simultaneously and simultaneously from the same sector position of the HDD # 0 and the HDD # 1. In the example of FIG. 7A, first, the read controller 22a simultaneously reads unit data from sector 0 of HDD # 0 and sector 0 of HDD # 1. The read controller 22a simultaneously sends the read unit data (data 0 and data 1) to the request source (host device 1).

Next, the read controller 22a simultaneously reads data 2 from sector 1 of HDD # 0 and data 3 from sector 1 of HDD # 1 and sends them simultaneously to the request source. Thereafter, similarly, the read controller 22a simultaneously reads the unit data (data 4 and data 5) from the sector 2 of the HDD # 0 and the sector 2 of the HDD # 1, and sends them simultaneously to the request source. Finally, the read controller 22a reads the unit data (data 6 and data 7) from the sector 3 of the HDD # 0 and the sector 3 of the HDD # 1 at the same time, and sends them simultaneously to the request source. Thus, data for eight sectors is read.
With such processing, as in the first embodiment, data can be read at twice the speed as compared with the case of reading data from only one of the HDDs while ensuring data redundancy.

  Here, a comparison is made with the read processing of the first embodiment. In the first embodiment, unit data is simultaneously read from the even-numbered sector of HDD # 0 and the odd-numbered sector of HDD # 1. Focusing on one HDD, data is read by skipping one sector. As a property of the HDD, when data is read from discontinuous sectors, a process of re-searching the sectors is performed. From that point of view, in the read processing of the first embodiment, a slight speed reduction due to the re-search occurs. On the other hand, in the second embodiment, the read controller 22a sequentially reads data from consecutive sectors in the HDD # 0 and the HDD # 1. For this reason, the sector is not re-searched, and higher-speed reading can be performed.

(Read when data error is detected)
Next, an operation in the case where certain unit data (in the following example, data 2 stored in the sector 1 of the HDD # 0) is broken and cannot be read normally will be described as an example.
FIG. 8 is a second diagram illustrating the data input / output method according to the second embodiment of the present invention.
It is assumed that a read request for data of 8 sectors has been made from the request source. As described in the normal case, the read controller 22a first reads data 0 from sector 0 of HDD # 0 and data 1 from sector 0 of HDD # 1 at the same time, and sends data 0 and data 1 simultaneously to the request source. Send it. Next, suppose that the read controller 22a detects that data 2 is corrupted when trying to simultaneously read data 2 from sector 1 of HDD # 0 and data 3 from sector 1 of HDD # 1. Then, the read controller 22a once discards the data 2 and the data 3. Then, this time, the read controller 22a switches the read destination of the unit data stored in the sector 1 of the HDD # 0 in which the abnormality is detected to the HDD # 1, and replaces the sector 5 in the HDD # 1 with the data 2 stored therein. This is set to the position of the sector to be read from HDD # 0 and HDD # 1. Then, the read controller 22a simultaneously reads the data 3 from the sector 5 of the HDD # 0 and the data 2 from the sector 5 of the HDD # 1, and simultaneously sends the data 3 and the data 2 to the request source. Next, the read controller 22a reads data 5 from the sector 6 of the HDD # 0 and data 4 from the sector 6 of the HDD # 1 at the same time, and simultaneously sends the unit data to the request source. Finally, the read controller 22a simultaneously reads the data 7 from the sector 7 of the HDD # 0 and the data 6 from the sector 7 of the HDD # 1, and simultaneously sends the unit data to the request source.

  As shown in this example, before detecting abnormal data, the read controller 22a operates to simultaneously read even-numbered unit data from HDD # 0 and odd-numbered unit data from HDD # 1. After detecting the abnormal data, the read controller 22a changes the control so that the odd-numbered unit data is simultaneously read from the HDD # 0 and the even-numbered unit data is simultaneously read from the HDD # 1.

  As a result, although the data 2 and the data 3 are once discarded and read again, there is a delay corresponding to the data reading of one sector, but subsequently, unit data that is simultaneously continuous can be read from the HDDs # 0 and # 1. High-speed data reading can be continued as in the normal case.

(At HDD failure)
Next, an operation in the case where an entire HDD (HDD # 0 in the following example) has a fixed failure will be described.
FIG. 9 is a third diagram illustrating a data input / output method according to the second embodiment of the present invention.
When the HDD # 0 has failed, the RAID controller 2 closes the HDD # 0 as in the case of the normal RAID1. Then, when there is a data write request for eight sectors from the request source, the write controller 22a writes only to the unblocked HDD # 1. At this time, the write controller 22a performs data writing and sector length information by performing rearrangement according to the sector length as described above. For example, the write controller 22a rearranges data 0 to data 7 in the order of data 1, data 3, data 5, data 7, data 0, data 2, data 4, and data 6, and writes the data to HDD # 1. As described below, by rearranging and writing in the same manner as in the normal case, the disk configuration illustrated in FIGS. 6 and 7A can be reproduced after the rebuild process.

  Then, when there is a read request for data for eight sectors from the request source, the read controller 22a reads data only from HDD # 1. The read controller 22a reads the unit data in the order of the pre-write data, and sends the read unit data to the host device 1. For example, the read controller 22a reads unit data in the order of sector 4, sector 0, sector 5, sector 1, sector 6, sector 2, sector 7, and sector 3, and sends the data to the host device 1 each time.

(Rebuild processing)
FIG. 10 is a diagram illustrating a rebuild process according to the second embodiment of the present invention.
When the HDD # 0 fails, the HDD # 0 is replaced and the rebuild process is performed. More specifically, the read controller 22a operates the HDD # 1 so that the arrangement of the read unit data is the original arrangement order (data 0 to data 7), as described in the above "when an HDD failure occurs". The unit data is read from an appropriate sector position and sent to the write controller 21a together with information on the sector length (8 in this example). The write controller 21a performs the rearrangement for the HDD # 0 according to the sector length and writes the data to the HDD # 0, as described in the above-mentioned "data writing". When the rebuilding process is completed, the data written to HDD # 0 and HDD # 1 are as illustrated in FIGS. 6 and 7A. After the rebuild processing, as described with reference to FIG. 6, the write controller 21a performs rearrangement according to the sector length and sequentially writes the data in consecutive sectors. Then, the read controller 22a simultaneously reads the unit data at the same sector position of the HDD # 0 and the HDD # 1 in the order of the sectors. As a result, it is possible to secure redundancy and realize high-speed reading according to the present embodiment.

According to the present embodiment, similarly to the first embodiment, the availability and read performance of the disk array device 3 can be improved. Further, the reading speed can be further increased as compared with the first embodiment.
Although the example in which the disk array device 3 is configured with two HDDs (HDD # 0, HDD # 1) has been described, the same write control and read control may be applied even when the number of HDDs is three or more. Can be. For example, when the number of HDDs is three (HDD # 0, HDD # 1, HDD # 2) and the arrangement of unit data to be written is data 0 to data 7, the rearrangement is performed as shown in FIG. May be written to. Specifically, the write controller 21a rearranges the data of HDD # 0 in the order of data 0, data 3, data 6, data 2, data 5, (empty), data 1, data 4, and data 7, and in this order. The data is stored in the sector order as it is. Further, the write controller 21a sorts the HDD # 1 in the order of data 1, data 4, data 7, data 0, data 3, data 6, data 2, data 5, and (empty), and keeps the sector in this order. Store them in order. The write controller 21a sorts the HDD # 2 in the order of data 2, data 5, (empty), data 1, data 4, data 7, data 0, data 3, and data 6, and keeps the sector in this order. Store them in order.

FIG. 12 is a diagram illustrating a minimum configuration of the control device according to each embodiment of the present invention.
As shown in FIG. 12, the control device 200 includes at least a writing unit 210 and a reading unit 220.
The writing unit 210 writes data to be written composed of a plurality of unit data arranged in each of a plurality of hard disks.
The reading unit 220 reads data from each of the plurality of hard disks in parallel for each unit data. When reading certain data, the reading unit 220 simultaneously reads, from different hard disks, unit data adjacent to each other in the arrangement of the data before writing.
In one embodiment, the reading unit 220 may simultaneously read unit data from different sector positions of each of a plurality of hard disks.
In one embodiment, when the reading unit 220 performs reading in the sector order starting from the same sector in each of the plurality of hard disks, the writing unit 210 simultaneously reads adjacent unit data in the data arrangement before writing. The arrangement of the unit data may be changed so that the data can be written to each of the plurality of hard disks.
The control device 200 corresponds to the RAID controller 2 in each of the above embodiments. The writing unit 210 corresponds to the writing controllers 21 and 21a. The read unit 220 corresponds to the read controllers 22, 22a.

  In addition, it is possible to appropriately replace the components in the above-described embodiment with known components without departing from the spirit of the present invention. The technical scope of the present invention is not limited to the above embodiment, and various changes can be made without departing from the spirit of the present invention.

100 Information processing system 1 Host device 2 RAID controller 3 Disk array device 11 OS
12 ... Driver 13 ... CPU
21, 21a Write controller 22, 22a Read controller 3 Disk array device # 0, # 1 HDD
200 control device 210 writing unit 220 reading unit

Claims (6)

A writing unit that writes data to be written composed of a plurality of unit data in each of a plurality of hard disks;
A reading unit that reads in units of unit data in parallel from each of the plurality of hard disks,
The writing unit writes to each of the hard disks without changing the arrangement of the data,
When reading the data, the reading section, each of said unit data successive in arrangement of the data before the write, and read out simultaneously from 1 by different sectors positions of the plurality of hard disks,
When detecting an abnormality at a first sector position of one of the hard disks, the reading unit sets a read target sector at the first sector position in one of the remaining hard disks, and sets the hard disk other than the hard disk. Is set as a new sector to be read, a sector position that differs by one from the first sector position.
Control device. A writing unit that writes data to be written composed of a plurality of unit data in each of a plurality of hard disks;
A reading unit that reads in units of unit data in parallel from each of the plurality of hard disks,
The writing unit, when the reading unit performs reading in the order of sectors starting from the same sector position in each of the plurality of hard disks, sequentially reads the unit data consecutive in the data arrangement before writing. Changing the arrangement of the unit data so that it can be written to each of the plurality of hard disks,
When reading the data, the reading unit reads each of the unit data consecutive in the data arrangement before writing, simultaneously from different hard disks in sector order,
When the reading unit detects an abnormality at one sector position of one of the hard disks, the same unit data as that written at the sector position at which the abnormality is detected is applied to one of the remaining hard disks. Setting the written sector position as a new read target sector,
Control device. If one of the hard disks fails, close the hard disk,
The writing unit writes the unit data to the other hard disks in the same arrangement order as before the blockage,
The control device according to claim 2 . Multiple hard disks,
A control device according to any one of claims 1 to 3 ,
A disk array system comprising: Writing data to be written composed of a plurality of unit data to each of the plurality of hard disks;
Reading in units of unit data in parallel from each of the plurality of hard disks,
In the writing step, writing to each of the hard disks without changing the arrangement of the data,
When reading the data, in performing the read, each of the unit data to consecutive in a sequence of the data before the write, and read out simultaneously from 1 by different sectors positions of the plurality of hard disks,
In the reading step, when an abnormality is detected at the first sector position of one hard disk, a read target sector is set at the first sector position in any of the remaining hard disks, In the hard disk, a sector position that differs by one from the first sector position is set as a new read target sector.
Data input / output method. Writing data to be written composed of a plurality of unit data to each of the plurality of hard disks;
  Reading in units of unit data in parallel from each of the plurality of hard disks,
In the writing step, when reading is performed in the sector order starting from the same sector position in each of the plurality of hard disks in the reading step, the unit data consecutive in the data arrangement before writing is simultaneously read Changing the arrangement of the unit data so that it can be written to each of the plurality of hard disks,
In the case of reading the data, in the step of performing the reading, each of the unit data continuous in the data arrangement before writing is simultaneously read from different hard disks in sector order,
In the step of performing the reading, when an abnormality is detected at one sector position of one of the hard disks, in any of the remaining hard disks, the same as the one written at the sector position at which the abnormality is detected is used. Setting the sector position where the unit data is written as a new read target sector,
Data input / output method.

Images