JPH0916343A - Disk drive system - Google Patents

Abstract

PURPOSE: To improve the mounting density of disk drives in the disk array sub-system of a RAID system. CONSTITUTION: This disk array sub-system is provided for operating a system by recovering the data of a fault disk drive 10c3 based on the parity data of a parity disk 10d3 when any fault is generated at a certain disk drive 10c3 or the like, disk drives 10a1-10a3 or the like are connected to plural SCSII/F 12 connected with a disk controller 2 three by three and made into a unit and plural exchange units 13a-13e are provided to be exchanged for each unit. On the other hand, when trouble is generated at a certain unit 13c, the disk controller 2 writes the recovered data and the data of the other disk drives 10 in the unit 13c in a spare unit 13e and operates them.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk system which employs a RAID system for controlling a plurality of disk drives.

[0002]

2. Description of the Related Art Conventionally, in a RAID type disk system, for example, a disk array subsystem or the like, a housing design is made in consideration of workability of replacement when a certain disk drive fails, and therefore, the shape of the housing is restricted. Not only that, but the disk mounting density in the housing is low, and improvement is desired.

As shown in FIG. 4, the conventional disk array subsystem has individual disk drives 41a-41.
Assuming a failure of e, each disk drive is housed in each of the units 42a to 42e as a unit to be replaced, and the units are mounted in a row (array form) along the front or back surface of the cabinet 40. .

In this case, even if the disk drive of a certain unit fails while the system is operating, the cabinet 4
If you open the front side door of 0, all units 42a-42
e appears, and the unit of the failed disk drive can be easily removed and replaced with a new one.

By the way, normally, when designing a disk array subsystem of this type, the cabinet vertical,
The dimensions such as width and depth are predetermined to some extent from the viewpoint of obtaining stability and the like. Therefore, when the units 42a to 42e of the disk drives are arranged in the cabinet 40 under the above arrangement conditions and then the unit 44 such as the controller and the power supply device is arranged, a space 45 is formed in the internal space thereof.

[0006]

However, in this case, not only the utilization efficiency inside the cabinet is poor, but also the mounting efficiency of the disk drives is poor because the number of disk drives that can be mounted is very small.

Therefore, it is conceivable that a plurality of disk drives are arranged in the depth direction of the cabinet and connected to the interface of the controller, ignoring the exchangeability of the disk drives to some extent. There is no choice but to make the cabinet full of doors so that it can be broken, and if the failed disk drive is removed from the interface while the system is operating (such as when other disk drives are operating),
At that time, noise is generated, and noise enters other disk drives connected to the same interface,
Since the signal becomes unstable, it is usually necessary to add appropriate noise countermeasures (hardware or software countermeasures), and it is not possible to simply add a disk drive to the empty space of the cabinet. was there.

The present invention has been made in order to solve such a problem, and it is possible to improve the disk mounting efficiency in the housing without adding noise countermeasures so as not to affect others when hot-plugging and unplugging. The purpose is to provide a disk system that can.

[0009]

In order to achieve the above object, a disk drive system according to claim 1 comprises:
A plurality of disk drives and a group of these disk drives are grouped in a predetermined number in the housing, parity data is recorded in one of each group, and when a failure occurs, data is grouped into each group based on the parity data. A disk drive system provided with a disk controller for restoration is characterized in that each of the disk drives is grouped into columns between groups and is detachably unitized with respect to the disk controller.

According to a second aspect of the present invention, there is provided the disk drive system according to the first aspect, wherein the disk controller has a plurality of units after a data restoration process when a failure occurs in a unit among the unitized units. It is characterized in that a predetermined unit therein is operated as a spare unit.

According to a third aspect of the present invention, there is provided a disk drive system in which a plurality of disk drives are grouped in a housing and a predetermined number of these disk drives are grouped, and parity data is recorded in one of the groups to cause a failure. Time,
In a disk drive system provided with a disk controller that restores data in units of groups based on the parity data, disk drives grouped by the predetermined number are arranged in a row, and the disk controllers are intergrouped. And a detachable exchange unit is collectively configured for each column, and when the disk controller fails in one of the plurality of disk drives, the disk controller is configured to perform another operation based on the parity data of the parity disk for each group. Data recovery means for performing a parity check on the data of the disk drive and restoring the data of the failed disk drive, the data restored by the data restoring means, and the data of another disk drive in the unit to which the failed disk drive belongs. And among the above The data writing means writes data to the corresponding disk drive of each unit, and the control means that operates the system by replacing the predetermined unit in which the data is written by the data writing means with the unit to which the failed disk drive belongs. There is.

A disk drive system according to a fourth aspect of the present invention is provided with a plurality of disk drives in a housing and a disk controller for grouping the disk drives into groups of a predetermined number and executing data processing for each group. In the disk drive system, the disk drives grouped by the predetermined number are arranged in a row,
With respect to the disk controller, a replacement unit is configured so that it can be attached / detached collectively for each row between groups,
The disk controller has one of the plurality of units as a parity unit, and parity data recording means for recording parity data for parity check when a disk drive of another unit fails in the parity unit. When a disk drive fails in a unit, the parity data is read from the corresponding disk drive of the parity unit, and data recovery means for restoring the data of the failed disk drive based on the parity data, One of the spare units is a first writing unit that writes the data of the failed disk drive restored by the data restoring unit to the corresponding disk drive of the spare unit, and another writing unit of the unit to which the failed disk drive belongs. Disk drive Second writing means for writing data to the corresponding disk drives of the spare units, and control means for operating the system by replacing the spare unit in which the respective data is written with the unit to which the failed disk drive belongs. It has.

[0013]

According to the first aspect of the invention, a plurality of disk drives grouped in a predetermined number are arranged in a row,
Each disk drive is grouped into a disk controller for each row between groups and is detachably unitized.

Therefore, when a disk drive in a unit fails, the data in the disk drive is restored, then all the data in the unit to which the disk drive belongs is copied to another unit, and the data in the failed disk drive is restored. If the unit to which it belongs is removed from the disk controller, the unit will be removed even if noise occurs, so other disk drives connected to the same interface will also be removed, and this will not cause performance degradation.

Further, according to the second aspect of the present invention, when a failure occurs in one of the unitized units, a predetermined unit among the plurality of units is operated as a spare unit after the data restoration process. Even if it is removed, the performance of the entire system will not deteriorate.

Further, according to the third aspect of the present invention, when one disk drive among a plurality of disks fails, the data restoration means of the disk controller causes another disk drive of another disk drive to be based on the parity data of the parity disk for each group. A data parity check is performed to restore the data on the failed disk drive.

Then, the restored data and the data of another disk drive in the unit to which the failed disk drive belongs are written by the data writing means to the corresponding disk drive of each of the plurality of predetermined units. When the data is written, the predetermined unit is operated by the control means as a substitute for the unit to which the failed disk drive belongs.

That is, when a failure occurs in a certain unit, the contents thereof are all reproduced in the predetermined unit, and the functions thereof are also replaced by the predetermined unit.

Therefore, even if the unit to which the failed disk drive belongs is replaced, the performance will not deteriorate.

Further, according to the present invention, when one disk drive out of a plurality of disks fails, the data restoration means of the disk controller causes the other disk drives of the other disk drives to be based on the parity data of the parity disk for each group. A data parity check is performed to restore the data on the failed disk drive.

Then, first, the restored data is the first
Is written in the corresponding disk drive of the spare unit. Then, the second writing means writes the data of another disk drive in the unit to which the failed disk drive belongs to the corresponding disk drive of the spare unit.

The spare unit in which each data is written is operated by the control means on the system as a substitute for the unit to which the failed disk drive belongs.

Therefore, even if the unit to which the failed disk drive belongs is replaced, the performance will not deteriorate.

By the above, even if the unit is removed from the system in order to replace the failed disk drive with a new disk drive, the influence of noise is eliminated, and the unit can be replaced even while the system is operating without affecting other units.

As a result, the disk mounting efficiency in the housing can be improved without adding noise countermeasures so as not to affect others when hot-plugging and unplugging.

[0026]

Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 is a diagram showing the configuration of a disk array subsystem according to an embodiment of the present invention.

In the figure, reference numeral 1 is a host device, such as a computer. Reference numeral 2 denotes a disk controller, which is a disk array controller having a RAID function. 3 is a processor, and the disk controller 2
It controls the function of. A work memory 4 is a memory used by the processor 3. A parity control circuit 5 generates parity data based on the RAID function, and records the parity data in a disk drive (hereinafter referred to as a parity disk drive) 10d1 to 10d3 in an array group.
When a disk drive (failed disk drive) in each array group fails, parity data is read from the parity disk drives 10d1 to 10d3, and the data of the failed disk drive is restored based on the data, and the spare disk 10e1 is restored. Control for writing to 10e3, etc. 6 is a buffer control circuit,
It controls the buffer memory. A data buffer memory 7 is a buffer memory for temporarily storing the data read from the disk drive. 8 is a disk interface circuit (hereinafter the interface is I / F
The I / F on the disk drive side is controlled. Reference numeral 9 is a host I / F circuit, and the host device 1
It controls the I / F on the side. 10a1 to 10e3 are disk drives, and the disk controller 2 performs S
Bit data is read / written through the CSI I / F 12. The parity data of each group is recorded in the disk drives 10d1 to 10d3. 11
Is a SCSI I / F on the host device side and is an I / F with the host device 1. 12 is SCSII / on the disk drive side
The number is F, and the number of disk drives in one array group is provided in the disk controller 2. 1
3a to 13e are a plurality of units that are exchange units,
Each is configured to accommodate a predetermined number of disk drives, for example, three disk drives 10 # 1 to 10 # 3. The unit of exchange is a vertical column unit (1 unit). 14a to 14c are array groups,
It is a logical group of disk drives that are determined by a column different from the exchange unit, that is, a horizontal column, and generate parity by the RAID function. Reference numeral 15 is an internal system bus, which is a data bus in the disk controller 2.

Here, the RAID function will be described.

Generally, as the level of RAID function,
There are level 1 (RAID 1) to level 5 (RAID 5).

RAID 1 is a mirroring method that has been widely adopted in the past (referred to as a duplicated disk, a mirrored disk, etc.).

RAID 2 is an application of the ECC function often used in memory to a disk array. Multiple data HDDs (hard disk drives)
On the other hand, a plurality of check HDDs serving as ECC check data are redundantly added, and the data is restored by the ECC function. However, this level is not generally considered as a practical solution, and there are currently few realization products.

The RAID 3 stores user data in a distributed manner in a plurality of HDDs by strapping. Also, an HDD dedicated to parity data for these HDD groups
Is a function for adding data and restoring data. The strapping size is 1 byte or block. User data is stored in parallel in a plurality of HDDs, at the same time a parity is generated, and the parity is stored in the parity HDD. The merit of this function is that since the HDDs are accessed in parallel, a high data transfer rate can be realized.

The RAID4 has a HDD dedicated to strapping + parity like the RAID3, but the strapping size is set to be relatively large. One user access is performed in parallel with RAID 3 in which the HDDs in the array are accessed in parallel.
In No. 4, since the strapping size is large, only some HDDs in the array are accessed. The merit of this function is that a single user data access can result in partial H
Since this is performed only for the DD, the user can access the remaining HDDs, and multiple requests can be executed simultaneously. This level is generally not considered as a practical solution and currently there are few realizable products.

RAID 5 has the same strapping size as RAID 3 but has a function of storing parity data in each HDD in the array in a distributed manner without fixing HDDs for storing parity data. The merit of this function is that RAID4 has parity H
Since the DD is fixed, the access to the parity HDD becomes a performance bottleneck, whereas the RAI
At D5, the parity data is distributed and stored in each HDD, so that the particular HDD does not become a bottleneck.

Next, the operation of this disk array subsystem will be described with reference to FIG.

As shown in FIG. 2, in this disk array subsystem, when one disk drive, such as the disk drive 10c3, fails, the disk controller 2 causes the same array group 14c as the failed disk drive 10c3. Disk drives 10a other than the failed disk drive 10c3 belonging to (group of disk drives for which parity is generated)
The contents of 3, 10b3 and 10d3 (bit data such as "1" and "0") are read. Then, the parity control circuit 5 takes the exclusive OR of the bit data of the disk drives 10a3 and 10b3 and compares it with the parity data of the disk drive 10d3 to restore the contents (data) of the failed drive, Is written in the data buffer memory 7 by the buffer control circuit 6.

Subsequently, the disk controller 2 transfers the data written in the data buffer memory 7 to the spare unit 13 corresponding to the failed disk drive 10c3.
Disk drive in e (spare disk drive) 1
Write to 0e3.

Next, the disk controller 2 transfers the contents of the other disk drives 10c1 and 10c2 of the same unit 13c as the failed disk drive 10c3 to the disk drives 10e1 and 1e in the corresponding spare unit 13e.
Copy to 0e2. It should be noted that the above-mentioned restoration processing and copy processing are executed by the disk controller 2 when there is no access request from the host device 1.

When the restoration process and the copying process are completed, the unit 13c including the failed disk drive 10c3
The spare unit 13e completely replaces the function of, and the unit 13c of the failed disk is no longer used.

Therefore, the failed disk drive 10c3
In order to replace the disk with a new one, replace the unit 13c containing the failed disk drive 10c3 with SCSI I / F 12
Even when the disk drive 10c3 is removed from the system, the spare unit 13e completely replaces the function, so that the entire system operates as in the state before the disk drive 10c3 fails, and the system can be operated without affecting the performance. .

If the unit containing a plurality of disk drives is replaced in the conventional control mode, the data of the failed disk drive is replaced by the spare disk drive, but other disks are replaced. Since the drive is used as it is, after the unit is removed, data restoration processing is performed on the disk drive 10c1 and the disk drive 10c2 that were not copied to the spare. For this restoration processing, the disk system itself is restored. Performance degradation occurs.

On the other hand, in the present embodiment, a plurality of disk drives connected to the same I / F are integrated into one unit, and even if the system is in an operating state, the plurality of disk drives are unit-by-unit without affecting performance and reliability. This makes the space in the depth direction of the cabinet available as a space for mounting the disk drives, and many disk drives can be mounted in the cabinet.

The failure unit 13c is replaced with SCSI I / F 1
Even if noise is generated when the disk drive is removed from disk 2, all disk drives 10c1 to 10c3 connected to the same I / F
Since they will be removed together, the noise at that time will not affect the others and the performance will not deteriorate.

As described above, according to the disk array subsystem of this embodiment, two or more disk drives 10a1 are provided.
10a3 are connected to the SCSI I / F 12 in a plurality to form one unit, the units are arranged in an array, a spare unit 13e is set in the plurality of units 13a to 13e, and a disk drive in the plurality of units, for example, When the disk drive 10c3 of the unit 13c fails,
Unit 1 that stores the failed disk drive 10c3 after restoring the data of the failed disk drive 10c3 and writing it to the corresponding spare disk drive 10e3
By copying the data of the other disk drives 10c1 and 10c2 in 3c to the respective disk drives 10e1 and 10e2 of the spare unit 13e and operating the spare unit 13e as an alternative, the failure disk can be operated without affecting the performance even during system operation. It is possible to replace the drive 10c3 with the unit 13c.

Therefore, as shown in FIG. 3, a plurality of units 13a accommodating a plurality of disk drives 10a1 to 10a3 in the depth direction of the cabinet 20 can be arranged in an array, and the conventional unit shown in FIG. The space 45 in the depth direction, which is a wasted space in the cabinet 40, can be effectively used as a mounting space for the disk drive.

As a result, the effect that the mounting density of the disk drive system is improved and the restrictions on the device shape are alleviated can be obtained.

In the above embodiment, the parity structure of RAID function level 3 or level 4 has been described.
It goes without saying that the present invention can be applied to other levels in the same manner as above.

[0049]

As described above, according to the first aspect of the present invention, since the disk drives are grouped into the disk controller for each row between groups and are detachably attached, the disk drives are unitized. Even if noise is generated when the drive is removed, it will not affect other disk drives and performance will not deteriorate.
According to the second aspect of the present invention, when a failure occurs in a unit that has been unitized, the disk controller operates a predetermined unit among a plurality of units as a spare unit after the data restoration process, so the failure occurs. Even if the unit is removed, the performance of the entire system will not deteriorate.

Further, according to the third aspect of the invention, the restored data and the data of the other disk drive in the unit to which the failed disk drive belongs are associated by the data writing means with each of the plurality of predetermined units. Since the predetermined unit is written in the disk drive to be replaced and the unit to which the failed disk drive belongs is operated as a substitute, the performance deterioration does not occur even if the unit to which the failed disk drive belongs is replaced.

According to the invention described in claim 4, the restored data is first written in the corresponding disk drive of the spare unit by the first writing means, and then by the second writing means. The data of another disk drive in the unit to which the failed disk drive belongs is written to the corresponding disk drive of the spare unit, and the spare unit is operated on the system as a substitute for the unit to which the failed disk drive belongs. Even if the unit to which the disk drive belongs is replaced, the performance does not deteriorate.

As described above, even if the defective disk drive is replaced with a new one, the influence of noise disappears even if the unit is removed in units, and the unit can be replaced without affecting other units even during system operation.

As a result, it is possible to improve the disk mounting efficiency in the housing without adding noise countermeasures so as not to affect others when hot-plugging and unplugging.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of a disk array subsystem according to an exemplary embodiment of the present invention.

FIG. 2 is a diagram showing the operation of this disk array subsystem.

FIG. 3 is a diagram showing an example in which a plurality of disk drives are mounted in a housing in this disk array subsystem.

FIG. 4 is a diagram showing a mounting example of a conventional disk array subsystem.

[Explanation of symbols]

1 ... Host device, 2 ... Disk controller, 3 ... Processor, 4 ... Work memory, 5 ... Parity control circuit, 6
... buffer control circuit, 7 ... data buffer memory, 8 ...
Disk I / F circuit, 9 ... Host I / F circuit, 10 ... Disk drive, 11, 12 ... SCSI I / F, 13a-13
e ... Replacement unit unit, 14a to 14c ... Array group.

Claims (4)

[Claims] 1. A plurality of disk drives in a housing and a group of these disk drives are grouped in a predetermined number. Parity data is recorded in one of the groups. When a failure occurs, each disk drive is grouped based on the parity data. A disk drive system provided with a disk controller for restoring data in units of groups, wherein each of the disk drives is grouped into columns between groups and detachably unitized with respect to the disk controller. Disk drive system. 2. The disk drive system according to claim 1, wherein the disk controller operates a predetermined unit among the plurality of units as a spare unit after a data restoration process when a failure occurs in a unit among the units. A disk drive system characterized by: 3. A plurality of disk drives in a housing and a group of these disk drives are grouped by a predetermined number, and parity data is recorded in one of the groups. When a failure occurs, each disk drive is based on the parity data. In a disk drive system provided with a disk controller for restoring data in units of groups, disk drives grouped by the predetermined number are arranged in a row, and the disk controllers are collectively attached / detached for each row between groups. The disk controller freely configures a replacement unit, and when one of the plurality of disk drives fails, the disk controller is configured to parity data of other disk drives based on the parity data of the parity disk for each group. Check and restore failed disk drive data A data recovery unit, a data recovery unit that writes the data recovered by the data recovery unit and the data of another disk drive in the unit to which the failed disk drive belongs to a disk drive corresponding to a predetermined unit of the plurality of units. A disk drive system comprising: an input unit; and a control unit that operates the system by replacing a predetermined unit in which data is written by the data writing unit with a unit to which the failed disk drive belongs. 4. A disk drive system comprising a plurality of disk drives in a housing, and a disk controller for grouping the disk drives into groups of a predetermined number and executing data processing in units of the groups, wherein Grouped disk drives in a row, and for the disk controller, a replacement unit is configured so as to be attachable and detachable collectively for each row between groups, and the disk controller is one of the plurality of units. Is used as a parity unit, and parity data recording means for recording parity data for parity check when a disk drive of another unit fails in that parity unit, and when a disk drive fails in a certain unit,
A data restoring means for reading the parity data from the corresponding disk drive of the parity unit and restoring the data of the failed disk drive based on the parity data; and a data restoring means using one of the plurality of units as a spare unit. The first writing means for writing the data of the failed disk drive restored by the corresponding disk drive of the spare unit, and the data of the other disk drive in the unit to which the failed disk drive belongs to the spare unit. Disk for writing data into a disk drive for operating the system, and a control means for operating the system by replacing the spare unit in which the respective data is written with the spare unit to which the failed disk drive belongs. Drive system .