JPH0962461A - Automatic data restoring method for disk array device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve fault resistance by inspecting medium surfaces while minimizing a load, and detecting a fault in its early stage and performing automatic restoration. SOLUTION: This method includes steps (S100 and S101) for detecting a wait time in which there is no access from a host system, a step (S102) for inspecting the magnetic disk medium surfaces of a plurality of magnetic disk devices in the wait time, a step (S104) for allocating a fault position to the alternate area of a magnetic disk device if the fault is detected on the magnetic disk medium surface (S103), and a step (S105) for restoring data from other magnetic disk devices and storing them in the alternate area. Further, the magnetic disk devices are equipped with buffers respectively and data stored on the magnetic disk devices are read in the buffers to inspect the magnetic disk medium surfaces. Consequently, the loss of the data is minimized and the throughput of the whole system is prevented from decreasing.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic data recovery method in a disk array device, and a disk for inspecting a medium surface by using a waiting time to detect a failure early and recovering data from another magnetic disk device. The present invention relates to an automatic data recovery method in an array device.

[0002]

2. Description of the Related Art In a conventional single magnetic disk device, a disk failure for which data recovery is unlikely to be found by a medium surface inspection is data loss. In such a case,
Data recovery is performed under the initiative of the host system, alternate allocation is performed by alternative sector processing, etc., and the data that was previously backed up to magnetic tape etc. is written back to the disk medium, but it is backed up to magnetic tape. The data is not always the latest data, but the data at the time of backup.

The disk array device is superior in fault tolerance as compared with a single magnetic disk device, and the disk array device alone has a data restoration capability.

Regarding the disk array device, a paper "A Case f" published in 1987 by Professor David A. Patterson of the University of California, Parkley, USA
or Redundant Arrays of Inexpensive Disks (RAID) ''
(Paper number UCB / CSD87
/ 391), of which the array disk device is called "RAID" and is classified into five levels.

RAID 1 is a system using a mirror disk, and a disk array controller duplicates a magnetic disk device. Read access is another RAID
Although it is faster and more reliable than the system, only 50% of the total storage capacity of the magnetic disk device can record data. This system is not usually called a disk array device.

[0006] RAID 2 is a method in which a redundant disk device for a Hamming code is added, and data is striped bit by bit or in units of several bytes and recorded on each magnetic disk medium which is grouped. In addition to the data, the group also includes a dedicated magnetic disk device that records an error detection / correction code (Hamming code) generated based on the data.

Thus, even if one magnetic disk device fails due to the redundancy, it is possible to identify the failed magnetic disk device and recover the data. In this case, for example, for 10 data magnetic disk devices, 4 Hamming code magnetic disk devices
Further, five magnetic disk devices for the Hamming code are required for 25 data magnetic disk devices.

The RAID 3 performs the identification of the failed magnetic disk device by the ECC function, and has one redundant magnetic disk device for parity generated from the data to recover the data when the magnetic disk device fails. When a failure occurs, the data is restored based on the information from the magnetic disk device for parity.

RAID4 is the above-mentioned RAID2, RA
Whereas ID3 does striping in bit units,
In the case of RAID4, striping in sector units enables simultaneous execution of a plurality of data reads.

In RAID 5, the magnetic disk medium of the parity magnetic disk device, which is fixed in RAID 4, is distributed to the magnetic disk devices in the group,
This enables a plurality of data writing processes.

In any of the RAID systems described above, the loss of data due to a failure of one magnetic disk device is restored by using another magnetic disk medium, and another track is assigned by substitution in any of the RAID systems. Alternatively, the data can be restored by writing the restored data on another magnetic disk medium.

However, when a failure (secondary failure) occurs in the magnetic disk device during the data restoration operation during data recovery due to the failure occurrence (primary failure), the data is lost as in the conventional single magnetic disk apparatus. Becomes

[0013]

The conventional magnetic disk device restores data by the method described above, and the intervention of the host system and the data backup device are necessary for the data restoration. . However, the conventional disk array device does not require the intervention of the host system and the magnetic disk device for backing up data as in the case of the conventional magnetic disk device. If a failure occurs during recovery to the disk medium or recovery to the spare magnetic disk medium, there is a drawback that data will be lost.

It is an object of the present invention to perform a medium surface inspection using a medium surface inspection means while minimizing addition in a disk array device to detect a failure early, defective replacement allocation means and automatic data recovery means. The purpose of this is to minimize the loss of data due to the secondary failure of the magnetic disk device and improve the fault tolerance by performing the data recovery using the.

[0015]

According to the present invention, there are provided a host system, a plurality of magnetic disk devices for recording / reproducing data and having redundancy, and a plurality of magnetic disks for access from the host system via an interface. A method for automatically recovering data in a disk array device comprising a disk array controller that accesses the devices in parallel, the method comprising: detecting a wait time during which there is no access from the host system; A step of inspecting the magnetic disk medium surface of the disk device, a step of allocating the failure position to an alternate area of the magnetic disk device when a failure is detected on the magnetic disk medium surface, and a data restoration from another magnetic disk device And storing data in the spare area. To.

Further, each of the plurality of magnetic disk devices is provided with a buffer, and the surface of the magnetic disk medium is inspected by reading the data stored in each magnetic disk device into the buffer. This allows
The magnetic disk medium surface is inspected without reducing the system throughput.

[0017]

Next, the present invention will be described with reference to the drawings. FIG. 1 is a block diagram of a disk array device showing an embodiment for realizing the present invention.

The disk array device of the present invention is
It is assumed to be configured as an AID5 type disk array device.

Referring to FIG. 1, the disk array device of the present invention comprises a host system 1, a disk array control device 3 for controlling the disk array, magnetic disk devices 50-57 forming the disk array, and a host system 1. Host interface 2 which is a control and data bus between the disk array control device 3 and the disk array control device 3, and a disk interface 40 which is a control and data bus between the disk array control device 3 and the magnetic disk devices 50 to 57.
It consists of ~ 43. The magnetic disk devices 50-57 are composed of disks 70-77 and buffers 60-67 provided between the disk interfaces 40-43 and the magnetic disk devices 50-57. Note that FIG. 1 shows a state in which eight (4 × 2) magnetic disk devices 50 to 57 are connected.

The read operation in the RAID 5 type disk array device will now be described.

First, when a read request is issued from the host system 1, the disk array controller 3 receives a command through the host interface 2. The disk array control device 3 controls the operations of the magnetic disk devices 50 to 57 and the exchange of data, and reads data from a data disk having a target sector. The read data is transferred to the host system 1 via the host interface 2.

Next, the write operation in the RAID 5 type disk array device will be described.

First, when a write request is issued from the host system 1, the disk array controller 3 receives a command and data through the host interface 2. The disk array control device 3 is a magnetic disk device 50.
Controls the operation of 57 to 57 and the exchange of data. In the data writing process, old data and old parity are read from the magnetic disk devices 50 to 57 in which the corresponding data is stored and the magnetic disk devices 50 to 57 in which the corresponding parity is stored, and data to be newly written is read. Exclusive OR with old data and old parity (XO
R) Calculate the new parity from the operation. Then, the new data is stored in the magnetic disk device 50-5.
7, and the generated new parity data is also written in the magnetic disk devices 50 to 57 that store the parity in the same manner.

Next, the processing operation of the disk array device of the present invention will be described with reference to FIG.

FIG. 2 is a flow chart showing the procedure of the magnetic disk medium inspection and data recovery processing during the standby time of the disk array device shown in FIG.

First, the disk array control device 3 checks its own load state (S100) and determines that the load is so low as not to affect the requests such as read / write from the host system 1 ( S101), the medium surface is inspected by the magnetic disk medium surface inspection means (S1).
02).

This medium surface inspection is performed by the magnetic disk device 50.
The data is read from the disks 70-77 in the disk drives 50-57 to the buffers 60-67 in the magnetic disk devices 50-57, and the disk array controller 3 uses the disk interfaces 40-43 only for the results of reading the data. Receive through. And buffer 6
The data read by 0 to 67 is discarded in the buffers 60 to 67 without being sent to the disk array device 3 via the disk interfaces 40 to 43.

The disk array control device 3 makes a determination based on the read result of the transmitted data (S10).
3). When it is determined that the data needs to be restored, the defective replacement allocation means for allocating the failed position to the replacement area of the failed magnetic disk device in the magnetic disk devices 50 to 57 is used to perform the defective replacement allocation (S10).
4). Then, the automatic data recovery means reads out the data of the other normal magnetic disk devices 50 to 57, and based on the data, the exclusive logical sum (XOR) operation is performed to generate the data of the failed magnetic disk device. The data is restored by storing it in the allocated replacement area (S105).

The entire area of the magnetic disk medium can be inspected by repeating the above-mentioned processing a plurality of times for each inspection unit.

[0030]

As described above, according to the present invention, the inspection of the magnetic disk medium surface using the waiting time is performed while the bus occupancy of the disk interface is kept low, so that the throughput of the entire system is not lowered. The effect is that it can be inspected.

Further, when the data is recovered, the disk array control device 3 detects the failure early and does not need the intervention of the host system and the magnetic disk device for backing up the data to automatically recover the data. As a result, there is an effect that the loss of data due to the secondary failure of the magnetic disk device can be minimized and the fault tolerance can be improved.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a disk array device showing an embodiment for realizing the present invention.

2 is a flowchart showing a procedure of an inspection of a magnetic disk medium and a data recovery process during a standby time of the disk array device shown in FIG.

[Explanation of symbols]

 1 Host System 2 Host Interface 3 Disk Array Controller 40-43 Disk Interface 50-57 Magnetic Disk Device 60-67 Buffer 70-77 Disk

Claims (2)

[Claims] 1. A host system, a plurality of magnetic disk devices for recording / reproducing data and having redundancy, and a disk array for accessing the plurality of magnetic disk devices in parallel in response to an access from the host system via an interface. A method for automatically recovering data in a disk array device comprising a controller, wherein a step of detecting a standby time during which there is no access from the host system; A step of inspecting, a step of allocating a failure position to a replacement area of the magnetic disk device when a failure is detected on the magnetic disk medium surface, a step of restoring data from another magnetic disk device and storing the data in the replacement area A disk array device comprising the steps of: Automatic data recovery method. 2. A magnetic disk medium surface is inspected by providing a buffer in each of the plurality of magnetic disk devices, and reading the data stored in each magnetic disk device into the buffer. Automatic data recovery method for disk array device.