JPH05109192A - Data transfer system of array disk device - Google Patents

Abstract

PURPOSE:To improve the continuity of a data transfer during a parallel reading and transferring of data from plural disk device without interruption of data transfer to a host computer even though some devices have defects in their media. CONSTITUTION:An array disk device, which consists of an array disk control device 1 and plural magnetic disk devices #0 through nu, is provided with reserved magnetic disk devices SIGMA0 through SIGMAk and the device 1 is provided with a defect control section 4 and an array defect management table 5. If there exists a defect on the medium of any one of #0 through nu devices, the defect control section 4 reads the defect location information, refers to the table 5, assigns a reserved device and the data of the defect data block are assigned in the same address. When there is a read request, eliminate the defected device and the data are read from the assigned reserved device and other devices and transferred.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data transfer system of an array disk device for reading and transferring data from a plurality (n) of magnetic disk devices in parallel.

In recent years, there has been an increasing demand for higher speed computers. However, the speedup of a magnetic disk device is inferior to that of a CPU mounted in a host computer or the like. Therefore, there is a need for an array disk technology that transfers data in parallel from a plurality of magnetic disk devices to increase the data transfer speed of the entire device.

[0003]

2. Description of the Related Art FIG. 8A is a schematic configuration diagram of a conventional array disk device, and FIG. 8B is an explanatory diagram of a conventional replacement process.

In the figure, 1 is an array disk controller, 2 is a parallel data controller, ch0 to chn are channels, 3 is a magnetic disk device, dt is a defect management table, and #
0 to #n represent the numbers of the magnetic disk device 3.

A conventional array disk device is shown in FIG.
It was configured like A. As shown in the figure, the device is composed of an array disk control device 1 and a plurality of magnetic disk devices 3 (# 0 to #n). Inside the magnetic disk device 3 (# 0 to #n), A defect management table dt corresponding to each disk is provided.

Further, the array disk controller 1 is connected to a host computer (not shown) via a host interface (Host I / f), and the parallel data controller 2 and channels ch0 to chn are provided therein. Etc. are provided, and the magnetic disk devices 3 (# 0 to #n) are connected to these channels ch0 to chn, respectively.

In the array disk device having the above structure,
At the time of writing data, the serial data sent from the host computer is divided and converted into parallel data, and the data is simultaneously written to each magnetic disk device 3 (# 0 to #n).

At the time of reading data, the parallel data read from each magnetic disk device 3 (# 0 to #n) is converted into serial data by the parallel data control unit 2 and transferred to the host computer.

By the way, in the normal magnetic disk device 3,
In order to allocate the defective data block, a plurality of spare data blocks are created at the time of formatting on the same cylinder or on the alternate cylinder.

Conventionally, in an array disk device that transfers data from a plurality of magnetic disk devices in parallel, if a defect exists on a medium (data block) of a device in the plurality of magnetic disk devices, defect management is performed. By using the method, spare data for replacing data in units of data blocks,
It was assigned to a data block.

For example, as shown in FIG. 8B, in each magnetic disk device 3 (# 0 to #n), a spare data block S is created at the time of formatting. And
For example, if the data block at address 4 is defective, the defective data block is assigned to the spare data block S created in advance to perform the alternation process.

When the data is read from the magnetic disk device 3 (# 0 to #n) including the data blocks subjected to the alternation process in this way, the data cannot be read sequentially, and the alternation allocation is performed after the seek operation. Access the data block that is being edited.

Then, the parallel data control unit 2 converts the read data into serial data and transfers it to the host computer. In this case, since it is necessary to perform a seek operation, a rotation wait, and the like in order to read the data of the data blocks that have been alternately assigned, the data transfer is delayed.

Therefore, the array disk controller temporarily suspends the data transfer to the host computer side, and resumes the transfer when the delay data (data of the data block that has been alternately allocated) can be read. It will be.

For example, in FIG. 8A, when the magnetic disk device of # 2 has a data block which has been subjected to the alternation processing as shown in FIG. 8B, the data of the magnetic disk devices other than # 2 are stored in the parallel data control unit. 2 is sent to the parallel data control unit 2 with only the data from the magnetic disk device # 2 being delayed.

Therefore, the parallel data cannot be converted into serial data, and the data transfer to the host computer is delayed.

[0017]

SUMMARY OF THE INVENTION The above-mentioned conventional device has the following problems. When data is read in parallel from a plurality of magnetic disk devices, if any of the devices has a data block subjected to alternation processing, the read data of the data block is delayed.

For this reason, although the data from the device that has not been subjected to the alternation process can be read normally, the data transfer to the host computer is temporarily interrupted and the data transfer is delayed. Therefore, the transfer efficiency of read data is reduced.

The present invention solves such a conventional problem, and when data is read from a plurality of disk devices in parallel and transferred to a host computer, there is a defect on the medium of a certain disk device. Even in such a case, the purpose is to improve the continuity of data transfer by enabling data transfer to the host computer without interruption.

[0020]

FIG. 1 is a principle view of the present invention, in which the same reference numerals as those in FIG. 8 indicate the same elements. Also,
4 is a defect control unit, 5 is an array defect management table, SC is a spare channel, #S ₀ to #S _k are
Indicates the spare magnetic disk unit number.

In order to solve the above problems, the present invention has the following configuration. (1) In a data transfer system of an array disk device comprising a plurality of magnetic disk devices # 0 to #n and an array disk controller 1 for controlling the plurality of magnetic disk devices in parallel, If one or more spare magnetic disk devices #S ₀ to #S _K are provided and a defective data block exists on the medium of a plurality of magnetic disk devices # 0 to #n, the data blocks Data is allocated to a data block of the same address in a spare disk device, and when reading data to the allocated data block, the disk device in which the defective data block was present is removed, and the spare disk device is allocated. Then, the data is read from other disk devices in parallel and transferred.

(2) In the configuration (1), the array disk controller 1 generates an array defect management table 5 for collectively managing all defective data blocks and information for determining a magnetic disk device from which data is read. The defect control unit 4 is provided.

(3) In the configuration (2), when the defect on the medium is detected by each of the magnetic disk devices # 0 to #n and the alternation process is performed, the defect control section 4 causes the corresponding magnetic disk device to After reading the defect position information, the array defect management table 5 is referenced to generate information for allocating the defective data block to the spare magnetic disk device.

(4) In the configuration (3), the allocation processing to the spare magnetic disk device is performed based on the information generated by the defect control unit 4, and the array generated by using the information generated by the defect control unit 4 The information in the defect management table 5 is updated.

(5) In the configuration (4), when the array disk control device 1 receives a read command, the array defect management table 5 is referred to and the data block requested to be read is transferred to the spare magnetic disk device. Whether or not the allocated data block is included is checked, and the magnetic disk device from which data is read is determined based on the check result.

[0026]

The operation of the present invention based on the above construction will be described with reference to FIG. Each magnetic disk device (# 0 to #n)
If there is a defect on the medium, the data block is replaced and the defect management table dt is updated. When this update process is performed, the array disk control device 1 is notified. From this notification, the array disk control device 1 recognizes that the alternation process has been performed in the magnetic disk device.

When the array disk control device 1 recognizes that the alternation process has been performed, it refers to the array defect management table 5 and generates a spare magnetic disk device number to which a defective data block is allocated. At this time, the number of the magnetic disk device that has undergone the alternation process and the address number of the data block that has undergone the alternation process are generated as data that matches the data format of the array defect management table 5.

The generated data are temporarily stored in an internal buffer. Then, based on the information in the buffer, allocation processing to the spare magnetic disk device is executed, and the array defect management table 5 is updated.

Next, when the array disk control device 1 receives the read command, the array defect management table 5 is referred to and the data block requested to be read is
It is checked whether or not there is a data block allocated to the spare magnetic disk device.

If the data block allocated to the spare magnetic disk device is included, the defective magnetic disk device on the medium is removed, and the allocated spare magnetic disk device and other magnetic disk devices are removed. Data is read in parallel and transferred to the host computer.

By doing so, when data is read from a plurality of magnetic disk devices in parallel and transferred to the host computer, even if a defective data block exists in any of the devices, the data is transferred to the host computer. The data transfer can be performed without interruption, and the continuity of the data transfer can be improved.

[0032]

Embodiments of the present invention will be described below with reference to the drawings. 2 to 7 are diagrams showing an embodiment of the present invention, FIG. 2 is a configuration diagram of an array disk device, and FIG. 3 is an explanatory diagram when a defect on a medium is detected during normal operation. 4, FIG. 4 is an explanatory diagram of a read operation in an allocation state to the spare magnetic disk device, FIG. 5A is an explanatory diagram of allocation to the spare magnetic disk device, FIG. 5B is an array defect management table example, and FIG. 6 is an array defect management. A table update processing flowchart, and FIG. 7 is a processing flowchart during a read operation.

In the figure, the same symbols as in FIG. 1 indicate the same components.
Reference numeral 6 represents an MPU (microprocessor). The configuration of the array disk device in this embodiment is shown in FIG. This device includes an array disk control device 1, a plurality of magnetic disk devices 3 including # ₀ to #n, and #S ₀ to #S _k.
And a spare magnetic disk device 3 consisting of.

The array disk controller 1 includes a parallel data controller 2, a defect controller 4, an array defect management table 5, an MPU (microprocessor) 6, channels ch0 to ch2, and spare channels SC0 to SCK. In addition to the above, a defect management table dt is provided for each magnetic disk device 3.

The defect management table dt is a table for managing the defect position on the medium of each magnetic disk device, and this table data is updated in each magnetic disk device 3. The defect control unit 4 is a control unit for generating information for determining a magnetic disk device from which data is read.

The array defect management table 5 is a table for collectively managing all defective data blocks, and the MPU (microprocessor) 6 is a processor for performing various controls in the array disk controller 1.

The parallel data control unit 2 divides the serial write data sent from the host computer into parallel data, or converts the parallel data read from each magnetic disk device 3 into serial data. It controls.

Channels ch0 to chn are used when there is no defect on the medium, and are spare channels SC.
0 to SCK are spare channels used when a replacement process is performed due to a defect on the medium (data block).

Among the magnetic disk devices 3, # 0 to #n are normally used devices, and #S ₀ to #S _K are spare magnetic disk devices which are allocated when the alternation process is performed. ..

The operation of the array disk device in the embodiment will be described below. During normal operation, as shown in FIG. 3, the channels ch0 to ch0 are controlled by the MPU 6.
Only chn is connected and spare channels SC0 to S
Separate CK. In this state, data is written on the medium of the magnetic disk devices # 0 to #n or data is read from the medium.

The data read in parallel from each magnetic disk device 3 (# 0 to #n) is transferred to the parallel data control unit 2, and the parallel data is converted into continuous sequential data (serial data). ..

Thereafter, the sequentially converted data is transferred to the host computer through the host interface (Host I / f). In such an operation, for example, among the data blocks accessed in parallel by the array disk control device 1, the address number “04” (see FIG. 5A) of the data block on the medium in the magnetic disk device 3 of # 2 is Determined to be defective,
It is assumed that the replacement process has been performed.

In such a case, as in the conventional example, the data is once allocated to the alternate spare data block in the magnetic disk device # 2. After the completion of the alternation process, the information is stored in the defect management table dt, and the contents of the defect management table dt are updated (the processes up to this point are the same as those in the conventional example).

When the defect management table dt is updated, the # 2 magnetic disk device notifies the MPU 6 in the array disk control device 1 to that effect. After that, based on the instruction from the MPU 6, the defect control unit 4 reads out the above-mentioned information from the defect management table dt (processes in a free time).

Next, from the read information and the information obtained by referring to the array defect management table 5,
The number "# 2" of the magnetic disk device that has undergone the replacement process
The address number "04" of the data block on which the alternation process has been performed, and the spare magnetic disk device number (for example, #S ₀ ) to be assigned instead for each data block are matched with the data format of the array defect management table. Generate by converting to data.

The generated data is stored in a buffer (not shown) in the defect control section 4. When this process ends, the MPU 6 connects the spare channel (eg, SC0), reads the data of the alternate-processed data block from the magnetic disk device (eg, # 2) subjected to the alternate process, and reserves the data. Allocation processing is executed by writing to the same data block address “04” of the magnetic disk device (for example, #S ₀ ).

Then, according to an instruction from the MPU 6, the defect control section 4 updates the data in the array defect management table 5 using the data in the buffer. The processing from reading the information of the defect management table dt to the update of the array defect management table is performed by using the free time.

As the array defect management table 5, for example, a table as shown in FIG. 5B is used. In this table example, it is composed of an address and a data section of the management table. Each address from the start address “0000” to the end address “1000” of the data block is registered in the address of the management table, and the allocation process is performed in the data part in correspondence with each of these addresses. Data on whether or not it is done is registered.

For example, "0000" in the data section is data indicating that allocation processing has not been performed, and "FF"
“F0” and “FFF1” are data indicating that the replacement is assigned. In this case, "FFF0", "FFF
"1" indicates the address of the array defect management table storing the replacement information.

In the address of the array defect management table, as an address for storing allocation information,
The above-mentioned "FFF0", "FFF1" ... Are secured, and the data portion corresponding to each of these addresses is
Data "01A" showing the disk configuration for data generation
3 ”,“ A123 ”, ... In this case, "A" indicates a spare disk.

The array defect management table 5 does not manage a plurality of magnetic disk devices for each magnetic disk device, but handles all the devices as if they were one device. Therefore, the address numbers of the respective devices are not used, but all devices are represented by a series of address numbers.

Next, when a read command is issued to the data block that has undergone the allocation processing, the processing is performed as follows. Assignment In this case, as shown in FIG. 4, the data of the address number "04" in the data blocks in the magnetic disk apparatus # 2, the address number "04" in the data block in the spare magnetic disk drive # S ₀ It has been processed.

When the array disk controller 1 receives a read command in this state, the MPU 6 refers to the array defect management table 5 and refers to the read-requested data block in the spare magnetic disk device (# in this example). It is checked whether or not there is a data block allocated to S ₀ ).

If there is a data block that has been allocated, when reading that data block, MP
U6 changes the channel configuration (changes to non-standard configuration).
That is, the channel ch2 is temporarily disconnected and the spare channel SC0 is connected.

Then, the data of the same data block address number "04" in the magnetic disk devices # 0 to #n except # 2 and the spare magnetic disk device #S ₀ are read in parallel, and the parallel data control unit is read. Transfer to 2.

The parallel data control unit 2 converts the parallel data into sequential data (serial data) and transfers it to the host computer. By doing so, even when a read command is issued to a defective data block, all the data is read at the same time and transferred at the same time by reading the data allocated to the spare magnetic disk device. It becomes possible.

The processing of the array disk device will be described below with reference to the flowcharts of FIGS. 6 and 7. The process numbers in FIGS. 6 and 7 are shown in parentheses. (Update process of array defect management table) ... See FIG. 6 In each of the magnetic disk devices # 0 to #n, a defect exists on the medium when the medium is accessed (S1), and the alternation process is performed. When the above is performed (S2), the defect management table dt is updated (S3).

In such a case, free time (time when there is no access from the host computer and no internal processing)
Use to execute the following process. First, from the magnetic disk device (the device # 2 in this example) that has undergone the alternation process,
Since the MPU 6 is notified that the defect management table dt has been updated by performing the alternation process, the MPU 6 that has received this notification issues an instruction to the defect control unit 4 to read the information of the defect management table dt.

Thereafter, the defect control section 4 reads information from the defect management table dt of the magnetic disk device (S4). Next, the defect control unit 4 refers to the array defect management table 5 to find the number of the spare magnetic disk device to be allocated.

Then, the number "# 2" of the magnetic disk device subjected to the alternation processing, the address "04" of the data block subjected to the alternation processing, and the number of the spare magnetic disk device for allocating the data block instead. "#S ₀ " is generated (S5).

In this case, for the device number “# 2” and the block number “04”, the data read from the defect management table dt is not used as it is, but is converted into the data format used in the array defect management table 5. , Generate as described above.

The data thus generated is temporarily stored in a buffer (not shown) provided in the defect control unit 4. Next, the MPU 6 refers to the data in the buffer and performs allocation processing to the spare magnetic disk device (S6).

In this process, the MPU 6 connects the spare channel SC0, then reads the data of the data block address "04" which has been subjected to the alternation process of the magnetic disk device of # 2, and reads this data from the spare of #S ₀ . This is done by writing to the same data block address "04" in the magnetic disk device.

When this allocation process is completed, the MPU 6
In response to the instruction from, the defect control unit 4 takes out the data stored in the buffer and updates the data in the array defect management table 5 (S7). (Processing during read operation) ... See FIG. 7. When the array disk control device 1 receives a read command from the host computer, the MPU 6 first refers to the array defect management table 5 (S
11) It is checked whether or not the read-requested data block has a data block allocated to the spare magnetic disk device (S12).

As a result, if there is no allocation, the standard configuration read operation (normal read operation) is performed (S13).
However, if the data block is allocated to the spare device, the data is read as follows.

Data reading from the medium is performed for each data block. Therefore, first, it is determined whether the next data block (the data block to be read next) is allocated to the spare device (the MPU 6 determines) (S1).
4) If not assigned, data read (S18) is performed with the standard configuration.

However, if it is assigned, the channel configuration is changed to a non-standard configuration (S15). In this process, as shown in FIG. 4, the channel ch2 is temporarily cut off and the spare channel SC0 is connected (both are processed by the MPU 6).

In this state, data is read in parallel from each magnetic disk device (S16). In this data read, each of the magnetic disk devices # 0 to #n except # 2,
Data is simultaneously read from the same data block address in the spare magnetic disk drive of S ₀ and transferred to the parallel data control unit 2.

This parallel data is converted by the parallel data control unit 2 into sequential data (serial data) and transferred to the host computer. When the data reading is completed in this way, the MPU 6 returns the channel configuration to the standard configuration (a state in which ch2 is connected and SC0 is disconnected) (S17).

Next, it is judged whether or not all the read-requested data blocks have been read (S19), and when all the data blocks have been read, the process ends. However, if it has not been completed, the counter is updated to the next read target data block (S20), and the above processing (S14 to S17) is repeated.

(Other Embodiments) The embodiments have been described above, but the present invention can also be implemented as follows. (1) In the above-described embodiment, it is necessary to configure the array defect management table by the magnetic disk device or the non-volatile memory (SRAM) so that the position information of the data block including the defect is not lost even when the power is turned off. However, it is possible to configure the array defect management table with an inexpensive volatile memory by using the method of creating the array defect management table every time by reading the information of each defect management table at the time of power-on. .

(2) In the array defect management table shown in FIG. 5B, the number of spare magnetic disk devices is one (the device shown in FIG. A), or even if a plurality of spare devices are provided. Then you can create a table. For example, a portion corresponding to A in the figure (A = # S0 in the embodiment) is B
(B = # S1) C (C = # S2) or the like may be used for allocation.

An address for storing allocation information may be secured next to the final address of the data block. (3) If there are a plurality of magnetic disk devices that have undergone the replacement process for the defective part, if multiple spare magnetic disk devices are used for allocation, the same processing as in the above-described embodiment is performed. Is possible.

[0074]

As described above, the present invention has the following effects. (1) When transferring data in parallel from n magnetic disk devices, even if there is a defective data block in one of the n magnetic disk devices, temporarily remove the corresponding magnetic disk device, By generating continuous sequential data from the data that can be read normally (without delay) and the data of the spare magnetic disk device, the data can be transferred to the host computer without any data delay, and the array Enables high-speed data transfer of magnetic disks.

(2) The read data can be transferred to the host computer without interruption, and the continuity of data transfer can be improved. As a result, read data transfer efficiency can be improved.

[Brief description of drawings]

FIG. 1 is a principle diagram of the present invention.

FIG. 2 is a configuration diagram of an array disk device according to an embodiment of the present invention.

FIG. 3 is an explanatory diagram when a defect on a medium is detected during normal operation.

FIG. 4 is an explanatory diagram during a read operation in an allocation state to a spare magnetic disk device.

5A is an explanatory diagram of allocation to a spare magnetic disk device, and FIG. 5B is an example of an array defect management table.

FIG. 6 is an array defect management table update processing flowchart.

FIG. 7 is a processing flowchart during a read operation.

FIG. 8A is a schematic configuration diagram of a conventional array disk device,
B is an explanatory view of a conventional replacement process.

[Explanation of symbols]

 1 Array Disk Controller 2 Parallel Data Controller 3 Magnetic Disk Device 4 Defect Controller 5 Array Defect Management Table

Claims (5)

[Claims] 1. A plurality of disk devices (# 0 to #n)
And an array disk controller (1) for controlling a plurality of disk devices in parallel, the data transfer method of an array disk device comprising: one or more spare disk devices (#S ₀ ~ #S _K ) is provided, and if a defective data block exists on the medium of a plurality of disk devices (# 0 to #n), the data of the data block is replaced with the data of the same address in the spare disk device. Allocated to blocks, when reading data to the allocated data block, remove the disk device in which the defective data block existed and read the data in parallel from the allocated spare disk device and another disk device. A data transfer method for an array disk device, which is characterized in that the data is transferred after being transferred. 2. An array defect management table (5) for collectively managing all defective data blocks in the array disk controller (1), and a defect controller for generating information for determining a disk device from which data is read. The data transfer system for an array disk device according to claim 1, further comprising (4). 3. When the respective disk devices (# 0 to #n) detect defects on the medium and perform replacement processing, the defect control section (4) outputs defect position information from the corresponding disk devices. The information for allocating a defective data block to a spare disk device is generated by referring to the array defect management table (5) after the reading, and the information is generated. Data transfer method of array disk device. 4. An array defect management table using allocation information to a spare disk device based on the information generated by the defect control unit (4) and using the information generated by the defect control unit (4). 4. A data transfer system for an array disk device according to claim 3, wherein the information in (5) is updated. 5. The array disk controller (1),
When the read command is received, it is checked by referring to the array defect management table (5) whether or not the read-requested data block includes the data block allocated to the spare disk device. 5. The data transfer method for an array disk device according to claim 4, wherein the disk device to read the data is determined based on the result.