JP3555871B2 - Disk array device - Google Patents

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a disk array device, and more particularly to a disk array device that reads and writes data from and to a plurality of disk media in parallel, and divides and assembles the data to be read and written.
[0002]
[Prior art]
Conventionally, a disk array device has a plurality of disk media and has a magnetic head corresponding to each disk medium. The divided data is stored in parallel in the plurality of disk media, or the data that has been stored in a distributed manner is read out from the plurality of disk media. At this time, the divided data is provided with redundancy so that the data can be reconstructed even if one disk fails.
[0003]
FIG. 3 shows a conventional disk array device 100. The disk array device 100 includes a host data control unit 102 for performing data transfer with an upper-level host 200, and a plurality of disk media 101 for reading and writing predetermined data according to a command from the upper-level host 200. Then, a cache memory 103 to which predetermined data is input / output and a memory control unit 104 that controls the cache memory 103, and a RAID control unit 105 that processes data stored in the cache memory 103 in correspondence with a plurality of disks 101 A disk data control unit 106 for transferring data between the cache memory 103 and the plurality of disks 101; and a data guarantee unit 107 for guaranteeing data read / written to / from the disks 101.
[0004]
Here, an operation of writing data to the disk 101 in the conventional disk array device 100 will be described. First, the host data control unit 102 receives write data from the host 200 and stores the data in the cache memory 103. Subsequently, the RAID control unit 105 processes data in the memory according to the state of the RAID configured by the disk 101. For example, at the time of writing, data is divided.
[0005]
Subsequently, the data guarantee unit 107 has a trailer generation and check circuit 108 for generating a trailer including an error detection signal, and generates a trailer for data in the cache memory 103 in accordance with a request from the disk data control unit 106. The data is added to the data and output to the disk data control unit 106. Then, the disk data control unit 106 writes the data divided by the RAID control unit 105 to the plurality of disks 101.
[0006]
At the time of data reading, an operation reverse to the above-described processing is performed. That is, first, the divided data is read from the disk 101 by the disk data control unit 106. Subsequently, the data assurance unit 107 performs a trailer check on the data output by the disk data control unit 106, deletes the trailer, and stores the trailer in the cache memory 103. After that, the data divided by the RAID control unit 105 is assembled and transferred to the host 200 via the cache memory 103 and the host data control 102.
[0007]
[Problems to be solved by the invention]
However, the conventional disk array device has the following disadvantages. That is, since the trailer is generated and checked during the data transfer between the cache memory and the disk, there is a disadvantage that the number of signals for transfer to and from the simultaneously activated disk cannot be equal to or greater than the number of trailer generation and check circuits. Therefore, there is a problem that the data transfer speed at the time of storing and reproducing data on the disk medium cannot be improved.
[0008]
On the other hand, if the number of interfaces for transferring data between the cache memory and the disk is increased in order to improve the data transfer, a trailer generation and check circuit is required corresponding to the increased number. Therefore, the number of trailer generation and check circuits is increased, which causes a problem that the apparatus becomes complicated and the manufacturing cost increases.
[0009]
[Object of the invention]
In the present invention, the inconvenience of the conventional example is improved, and in particular, the generation and the check operation of the trailer including the error detection signal are performed independently of the data transfer to the disk, and the limitation of the number of disk transfers that can be started simultaneously is lifted. It is another object of the present invention to provide a disk array device capable of improving the storage and reproduction speed of a disk, reducing the number of trailer generation and check circuits, and reducing the manufacturing cost.
[0010]
[Means for Solving the Problems]
In view of the above, the present invention provides a host data control unit that includes a plurality of disk media, and a plurality of magnetic heads that reads and writes predetermined data on the plurality of disk media, and that performs data transfer with an upper-level host. A cache memory that is connected to the unit and temporarily holds predetermined data transmitted and received by the host, a disk data control unit that controls reading and writing of predetermined data to and from a plurality of disk media via the cache memory , and a, a RAID controller for processing the predetermined data to be read or written to a plurality of disk media at the command corresponding to the plurality of disk media. In addition, a data guarantee unit is provided for checking whether reading / writing of predetermined data to / from the disk medium is correct according to a command from the RAID control unit .
Further, the RAID control unit retrieves predetermined data stored in the cache memory by the host data control or the disk data control unit from the cache memory, and performs data assurance with the RAID control unit for the retrieved predetermined data. And the RAID control unit operates, and the RAID control unit stores predetermined data processed by the RAID control unit or predetermined data checked by the data guarantee unit in the cache memory .
[0011]
With such a configuration, when data is written to the disk medium in accordance with a command from the host, the data to be written is processed by the RAID control unit so as to be divided according to the number of disk media, for example. You. The processed data is written to a disk medium and then read. Then, the data assurance unit checks whether an error has occurred in the data due to writing and reading. At this time, the data assurance unit operates according to a command from the RAID control unit separately from the write and read operations. Therefore, by operating the data assurance unit, it is possible to maintain the reliability of data, and independently of the operation of the data assurance unit, it is possible to independently execute data transfer with the disk, and to store data. The reproduction speed can be improved.
[0012]
In particular, for example, a predetermined data from the upper host is stored in the cache memory, the predetermined data processing of the data by the RAID controller is retrieved from the cache memory, and error checking of data by the data guarantee section is performed After that, the data is stored again in the cache memory by the RAID control unit . Since the reading and writing of data in the cache memory to and from the disk medium is controlled by the disk data control unit, data transfer with the disk can be executed independently of the data check operation by the data guarantee unit. Can be improved.
[0013]
Also, the data assurance unit has a code generation function of generating an error detection code corresponding to the predetermined data, a code addition function of adding the error detection code to predetermined data written on the disk medium, and a read function of reading from the disk medium. It is desirable to have a code deletion function of deleting the error detection code from the predetermined data. Thus, at the time of data writing, an error detection code for detecting a data error is generated based on the data to be written by the code generation function, and the error detection code is added to predetermined data to be written by the code addition function. You. Then, at the time of data reading, an error of the data is checked, and the error detection code is deleted by the code deletion function. Therefore, since these functions operate via the cache memory, data transfer to and from the disk can be executed regardless of these operations, and the data storage / reproduction speed can be improved. Here, the error detection code is a code including, for example, a CRC code.
[0014]
Further, the RAID control unit performs a data division function of dividing the predetermined data when the predetermined data is data to be written on the disk medium, and a data division function for dividing the predetermined data when the predetermined data is data read from the disk medium. A data assembling function for assembling predetermined data; a data assurance unit configured to generate an error detection code corresponding to the divided predetermined data; and a code generation function to convert the error detection code into the divided predetermined data. A code addition function to add and a code deletion function to delete the error detection code from the data assembled by the data assembling function may be provided. Even in this case, the same operation as described above can be achieved, and the above object can be achieved.
[0015]
May further comprise an arbiter for performing arbitration is a competitive process request for using the data guarantee unit by plurality Rutotomoni, the respective RAID controller a RAID controller. Thus, the RAID controller to which use permission is given by the arbiter uses the data assurance unit, so that a plurality of RAID controllers can share the data assurance unit . Therefore, it is not necessary to increase the number of data guarantee units, and it is possible to reduce the manufacturing cost without increasing the complexity of the structure while improving the read / write processing speed.
[0016]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, an embodiment of the present invention will be described with reference to FIG.
[0017]
FIG. 1 is a block diagram showing a configuration of a disk array device 1 according to the present invention. According to the disk array device 1 of the present invention, divided data can be stored in a plurality of disk media 11 in parallel, or data already distributed and stored from a plurality of disk media 11 can be read. it can. Data to be written and read is provided with redundancy, so that data can be reconstructed even when one disk 11 fails.
[0018]
The disk array device 1 includes a plurality of disk media 11, a plurality of magnetic heads (not shown) for reading and writing predetermined data from and to the plurality of disk media 11, and a host that performs data transfer with the host 2. A data control unit 12, a cache memory 13 for temporarily storing predetermined data, a memory control unit 14 for controlling input / output of the cache memory 13, and a predetermined data stored in the plurality of disk media 11 via the cache memory 13. A RAID control unit 15 for processing correspondingly; a disk data control unit 16 for similarly reading and writing predetermined data from / to a plurality of disk media 11 via the cache memory 13; Check whether the reading and writing of predetermined data in the disk medium 11 is correct. And a chromatography data assurance unit 17.
[0019]
Hereinafter, this will be described in detail.
[0020]
The disk array device 1 is a storage device such as a hard disk which is built in or externally attached to a host 2 such as a personal computer. The disk array device 1 is connected to a host 2 via a host interface 18. The host interface 18 is, for example, a SCSI interface. The host interface 18 is connected to the host data control unit 12 of the disk array device 1.
[0021]
The host data control unit 12 receives data to be written to the disk medium 11 from the upper host 2 or transmits data read from the disk medium 11 to the upper host 2 according to a command from the upper host 2. It performs data transfer with the host 2. Further, the host data control unit 12 is further connected to the cache memory 13. Specifically, it is connected to a memory control unit 14 that controls input and output of the cache memory 13.
[0022]
The cache memory 13 is a RAM having a predetermined capacity, and temporarily stores data received from the upper host 2 or data to be transmitted to the upper host 2. Therefore, when a command to write data to the disk medium 11 is issued from the host 2, the host data control unit 12 receives the write data, and the host data control unit 12 outputs the write data to the cache memory 13. ing. Then, the output write data is stored in the cache memory 13 by the memory control unit 14. On the other hand, when a command to read from the disk medium 11 is issued from the host 2, data read from the disk medium 11 is stored in the cache memory 13 by the memory control unit 14 as described later. Then, the stored read data is input to the host data control unit 12 by the memory control unit 14 and transmitted to the host 2 by the host data control unit 12.
[0023]
The RAID control unit 15 processes write and read data temporarily stored in the cache memory 13 in correspondence with the plurality of disk media 11. For example, when data is stored in a plurality of disk devices using RAID-3, the data to be written is allocated according to the number of disks 11 in order to allocate and write the data together with the parity data (data division function). At the time of reading, the divided data as read data is assembled and the data is restored (data assembling function). After this division and assembling operation, the processed data is stored in the cache memory 13 again. However, the processing operation of the RAID control unit 15 is not limited to the above operation.
[0024]
The data assurance unit 17 checks whether or not the write and read data in the cache memory 13 are executed without error according to a command from the RAID control unit 15. Specifically, the data assurance unit 17 includes a code generation function of generating an error detection code corresponding to the divided predetermined data, a code addition function of adding the error detection code to the divided predetermined data, A code deletion function of deleting an error detection code from the data assembled by the data assembling function.
[0025]
The code generation function includes, for example, a CRC generation circuit 17b that generates a CRC code by cyclic error detection (CRC), and an additional data generation circuit 17c that generates management information (for example, a serial number and an address) of the divided data. Consists of Here, the CRC generation circuit 17b checks the CRC of the read data, and the additional data generation circuit 17c also checks the additional data of the read data. The sign addition function includes a trailer addition circuit 17d for adding a trailer composed of the CRC code and the additional data to the divided data. Here, the trailer adding circuit 17d creates an initial value of the additional data to be read from the information of the disk to be read, and also predicts the additional data to be read. The code deletion function includes a trailer deletion circuit 17e for deleting a trailer of read data.
[0026]
Further, the data assurance unit 17 includes a trailer detection circuit 17a and a trailer storage unit 17f. The trailer detection circuit 17a detects a trailer including the CRC code and the additional data of the read data. Then, the detected trailer is deleted by the trailer deletion circuit 17e, and the trailer is stored in the trailer storage unit 17f. That is, the trailer storage unit 17f is a memory having a predetermined capacity. Here, the trailer stored in the trailer storage unit 17f is deleted after the trailer check is completed. Therefore, the capacity of the trailer storage unit 17f only needs to be at least equal to the trailer size (for example, 8 bytes).
[0027]
The disk data control unit 16 writes and reads data to and from the disk 11. That is, the divided data is stored and reproduced on the disk 11 via the cache memory 13. At this time, the disk data control unit 16 and each disk 11 are connected by the disk interface 19.
[0028]
Next, the operation of the present embodiment will be described separately for the operation of writing data to the disk 11 and the operation of reading data from the disk 11.
[0029]
In the case of a write operation, first, when a data write command is issued from the upper host 2, the host data control unit 12 receives the write data from the upper host 2 and stores it in the cache memory 13. Subsequently, the RAID control unit 15 retrieves the data stored in the cache memory 13 from the cache memory 13 and processes the data (EXOR) according to the state of the disk 11 (RAID information) where the write data can be stored. , Etc.), and then store it in the cache memory 13. Further, the RAID control unit 15 takes out the processed data from the cache memory 13 again, and calculates the CRC of the write data using the CRC generation circuit 17b. Further, an initial value of the additional data is created from the information of the disk 11 to be written, and the additional data is calculated using the additional data generation circuit 17c. Then, the generated CRC code and the additional data are combined into a trailer, the trailer is added to the write data using the trailer adding circuit 17d, the data is converted, and the trailer is stored in the cache memory 13. The trailer addition position is determined by the instruction of the trailer detection circuit 17a. Subsequently, the disk data control unit 16 retrieves the data to which the trailer has been added from the cache memory 13 and transfers the data to the disk 11 for storage.
[0030]
On the other hand, in the read operation, first, data is read from the disk 11 by a magnetic head (not shown), and the read data is received by the disk data control unit 16 and stored in the cache memory 13. Subsequently, the RAID control unit 15 retrieves the read data stored in the cache memory 13 from the cache memory 13, and the trailer detection circuit 17a detects the trailer portion of the read data. The detected trailer is deleted by the trailer deletion circuit 17e and stored in the trailer storage unit 17f. In addition, in order to check whether there is an error in the read data, a CRC code is generated from the previously read data by using the CRC generation circuit 17b, and the additional data generation circuit reads the disk code. The initial value of the additional data is predicted and created from the information of No. 11, and a trailer for checking is created from these. Then, by comparing the trailer for this check with the trailer deleted from the read data and stored in the trailer storage unit 17f, an error is detected by reading. Thereafter, the read data from which the trailer has been deleted is stored in the cache memory 13 again. Subsequently, the RAID control unit 15 performs data processing (eg, EXOR processing) in accordance with the state (RAID information) of the disk 11 on which the read data has been stored. Then, the host data control unit 12 transfers the processed data from the cache memory 13 to the host 2.
[0031]
By doing so, the trailer generation and check operations, which are error detection processes in reading and writing to the disk 11, are performed on the memory. In other words, the RAID control unit 15 retrieves data from the cache memory, 15 or the data assurance unit 17 performs the error detection processing, so that the transfer to the disk 11 and the trailer generation and check operations can be performed independently. Accordingly, a plurality of transfers to and from the disk 11 can be performed simultaneously regardless of the trailer generation and check operations, and the speed of storage and reproduction on the disk 11 can be improved.
[0032]
Furthermore, since the trailer addition and check operations are added as one operation of the RAID control unit 15, a trailer generation and check circuit is not added in accordance with the number of transfer signals to and from the disk 11 that starts simultaneously. The number of circuits can be reduced, and the manufacturing cost of the disk array device 1 can be reduced.
[0033]
Next, another embodiment of the present invention will be described with reference to FIG. FIG. 2 is a block diagram showing a configuration of another embodiment of the present invention.
[0034]
This other embodiment has substantially the same components as the disk array device 1 shown in FIG. The disk array device 2 illustrated in FIG. 2 further includes a plurality of RAID controllers 15A and 15B, and an arbiter 20 that performs a trailer generation by the RAID controllers 15A and 15B and a competition process of a check circuit use request. ing.
[0035]
The operation of the other embodiment will be described. In the write operation on the disk 11, first, the host data control unit 12 receives a plurality of read / write data from the host 2 and stores them in the cache memory 13. Data processing and trailer generation are performed using the RAID control units 15A and 15B for each of the write data. Also, arbitration, which is a conflict process for a request to use the trailer generation circuit 17, is performed, and the RAID control units 15A and 15B to which use permission has been given can use the trailer generation circuit 17. The disk data control unit 16 transmits the data on which the trailer has been generated and added to the disk 11.
[0036]
In the read operation, first, the disk data control unit 16 extracts data from the disk 11 in which the read data is stored and stores the data in the cache memory 13. Subsequently, the RAID controllers 15A and 15B check and delete the trailer, process the trailer, and reconstruct the read data. At the time of reading, arbitration is similarly performed to use the check circuit of the trailer. Then, the host data control unit 12 transmits the reconstructed read data to the host 2.
[0037]
By doing so , the trailer generation and check circuit 17, which is the data assurance unit 17, is shared by the plurality of RAID control units 15A and 15B. Therefore, even if a plurality of RAID control units 15A and 15B are mounted, the trailer generation and check are performed. Can be suppressed, the speed of reading / writing from / to the disk 11 can be increased, and the manufacturing cost of the disk array device 1 can be reduced. Further, in the present embodiment, a plurality of host interfaces 18 for connecting the host 2 and the host data control unit 12 or a plurality of disk interfaces 19 for connecting the disk data control unit 16 and the disk 11 may be connected. Also in this case, it is not necessary to increase the number of circuits 17 for generating a trailer or the like even though the RAID control units 15A and 15B are added, and the increase in the number of circuits inside the apparatus 1 can be suppressed.
[0038]
【The invention's effect】
As described above, in the disk array system of the present invention, a RAID controller for processing corresponding predetermined data to be read or written to a plurality of disk media at a command from the upper level host to a plurality of disk media, at the command of the RAID control unit, e Bei a data guarantee unit for checking whether the read or write to the disk medium in a predetermined data is correct, the RAID control unit, a host data control or disk data controller The predetermined data stored in the cache memory is retrieved from the cache memory, and the RAID control unit and the data guarantee unit operate on the retrieved predetermined data, and the predetermined data processed by the RAID control unit is processed. The RAID control unit stores the data or predetermined data checked by the data guarantee unit in the cache memory. , Since a configuration, when writing data to the disk medium by a command from the upper host, the write data is retrieved by the RAID control unit from the cache memory, while being processed by the RAID control unit, which is the processed data relative data assurance section is added guarantee data operated, are stored in the cache memory again at RAID, then Ru is written to the disk medium. When the data is read, the data is fetched from the cache memory by the RAID control unit as described above, and the data assurance unit checks whether an error has occurred by writing and reading the data according to a command from the RAID control unit. Ru is run by. At this time, the data assurance unit operates independently of the write and read operations, so that the data assurance unit operates to maintain the reliability of the data and to operate independently of the operation of the data assurance unit. The data transfer to and from the disk can be performed independently and the data storage / reproduction speed can be improved.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a configuration of an embodiment of the present invention.
FIG. 2 is a block diagram showing a configuration of another embodiment of the present invention.
FIG. 3 is a block diagram showing a configuration of a conventional disk array device.
[Explanation of symbols]
1 disk array device 2 host host 11 disk medium 12 host data control unit 13 cache memory 14 memory control unit 15, 15A, 15B RAID control unit 16 disk data control unit 17 data guarantee unit 18 host interface 19 disk interface 20 arbiter 17a trailer detection Circuit 17b CRC generation circuit 17c Additional data generation circuit 17d Trailer addition circuit 17e Trailer deletion circuit 17f Trailer storage unit

Claims (3)

A plurality of disk media; a plurality of magnetic heads for reading and writing predetermined data from and to the plurality of disk media; a host data control unit for performing data transfer with an upper host; A cache memory for temporarily storing predetermined data transmitted and received by the host, a disk data control unit for controlling the reading and writing of the predetermined data to and from the plurality of disk media via the cache memory; A RAID control unit for processing predetermined data read / written to / from the plurality of disk media in accordance with the plurality of disk media,
A data assurance unit that checks whether reading or writing of the predetermined data to the disk medium is correct according to a command from the RAID control unit,
The RAID control unit, said predetermined data stored in the cache memory in the host data controller or the disk data controller, and Eject from the cache memory,
The RAID control unit and the data assurance unit operate on the extracted predetermined data,
The RAID controller stores the predetermined data processed by the RAID controller or the predetermined data checked by the data assurance unit in the cache memory ,
The disk array device, wherein the disk data control unit simultaneously performs data transfer with the plurality of disks regardless of operations of the RAID control unit and the data guarantee unit . A data division function for dividing the predetermined data when the predetermined data is data to be written to the disk medium, wherein the predetermined data is data read from the disk medium; Data assembling function to assemble the predetermined data in the case,
A code generation function for generating an error detection code corresponding to the divided predetermined data, a code addition function of adding the error detection code to the divided predetermined data, 2. The disk array device according to claim 1, further comprising a code deletion function for deleting an error detection code from the data assembled by the above function. 3. The disk array device according to claim 1, further comprising a plurality of said RAID control units, and an arbiter for performing arbitration, which is a competition process for a request for use of said data guarantee unit by each of said RAID control units. .

Images