JPH10254634A - Storage device and restoration means for storage device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To shorten restoration time or to minimize the number of recording/ reproducing devices used for restoration so as not to obstruct a processing to other media at the time of a medium fault in a library device provided with the redundant array constitution of portable media. SOLUTION: In the library device for constituting a redundant array by the plural media 20, a standby medium is loaded to a standby drive 22 at all times, and at the time of the medium fault, data restoration is quickly performed without carrying the standby medium from a storage container 21. Or a nonvolatile storage device such as a disk device 220 or the like is provided other than the plural recording/reproducing devices, and at the time of the medium fault, it is replaced by using the disk device 220 as the standby medium.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a highly available storage system for portable storage media. In particular, the present invention relates to a storage system having redundancy in each component to improve availability.

[0002]

2. Description of the Related Art As a known example closest to the invention, the following Patterson article is known.

[0003] A. C. M. Sigmod Conference Proceedings, 1988, June, pages 109-116 (D.
Patterson, et al: A Case for Redundant Arrays of Ine
xpensive Disks (RAID), ACM SIGMOD conference proceed
ing, Chicago, IL, June 13, 1988, pp. 109-116) A paper by Patterson discloses a technique relating to data arrangement on a disk array.

[0004] The disk array is a mechanism for realizing high performance and high reliability of the disk system. In a disk array, in order to improve performance, physically a plurality of disk devices appear as one disk device to a processing device. On the other hand, when a failure occurs in a disk device storing data, redundant data for performing data recovery is stored in another disk device for higher reliability. RA refers to a disk array with this redundant information.
Called ID.

[0005] In the RAID system configuration described herein, a plurality of configuration examples that enable data recovery are described. One is called "mirroring", which improves reliability by writing each block of write data from the processing unit onto at least two separate disk drives.

The other uses "parity". The mechanism is briefly described as follows. The parity disk system divides data to be written into bits, bytes, or blocks each of which is a unit of writing from a processing device, and writes the plurality of divided data to different disk drives. Then, a parity is created by calculating an exclusive OR of the plurality of divided data and written to a disk drive different from the data. When the write data is divided into n pieces, the pieces of data are d1, d2,. . . , Dn, the parity p is generated from the divided data by the following (1)
After generating the parity p, d1, d
2,. . . , Dn and p are written to different disk drives.

P = d1 (+) d2 (+). . . (+) Dn .............. (1) where (+) represents exclusive OR.

These d1, d2,. . . , Dn and p are called a parity group, and p is updated accordingly when d1, d2,..., Dn are updated so that p always keeps a value according to equation (1). For example, even when another data D is overwritten only in the area d1 in which writing has already been performed, it is necessary to calculate a new parity P for the parity p and write P together with D to the disk. In this case, the parity is calculated by the following equation (2), and the operation of writing D and P to the disk is performed.

P = D (+) d1 (+) p... (2) The content on d1 is called "old data", the content on p is called "old parity", the data D to be newly written is "new data", and the parity P to be newly written is "new parity".

In these RAIDs, data can be recovered even if one disk drive fails. In "mirroring", since there is at least one other disk drive containing the same data, the system can continue to run on the remaining disk drives and the data can be reconstructed. With "parity", if any one disk fails, its data can be recovered by calculating the exclusive OR of the data on the remaining disks. US Patent 49
No. 14656 discloses a technique of restoring data and writing the data to a spare disk device managed by the system when a failure occurs in a disk device in a device having a plurality of redundant disk arrays.

In a computer system, magnetic tapes and magnetic tapes are often used as storage devices other than disk devices.
There are optical storage devices and the like. Particularly recently, DVD (Digital)
tal Video Disk) is attracting attention. These storage devices are characterized by a storage medium and an R / W (R
In other words, the read / write device is separated, the storage medium is loaded into an arbitrary R / W device, and data on the storage medium is read and written. Generally, these media are referred to as portable media. In large-scale computer systems,
Libraries are introduced to facilitate the management of very large numbers of portable media. The library includes, in addition to the storage medium and the R / W device, a storage for storing a large number of storage media, a robot for transferring the storage medium between the storage and the R / W device, and the like. .

Since the data handled by the computer system is becoming larger and larger, it is necessary to improve the availability of the data.
Is also very high. Therefore, even in the storage device system constituted by the portable medium as described above, the Pattern
It is effective to achieve high availability by applying the concept proposed in the son's paper.

[0013] As a technique in which such a concept is applied to a portable medium, Comdex 96: DVD application is available.
Alan E. Bell (IBM Research Division): DVD
Applications, COMDEX 96, Nov. 20, 1996). In this document, an RA having redundancy is obtained by combining a plurality of ordinary libraries including DVDs, R / W devices, robots, and the like.
IL (Redundant Arrays of Inexpensive Libralies)
Has been proposed.

[0014]

It is possible to apply the configuration of the RAID system proposed in Patterson's paper to a library device using a portable medium to enable data recovery. In the case of a library device using a portable medium, the storage medium on which data is recorded can be mirrored or made redundant by parity so that data can be restored and reliability can be improved.

However, data recovery becomes possible by making the media redundant, but the recovery takes a long time. In particular, when performing redundancy by using parity, it is necessary to perform a process of reading the contents of all the media constituting the RAID, calculating the parity, and writing the data to the spare medium for the entire storage area of the medium at the time of restoration.

Further, in a library device, the number of R / W devices is usually overwhelmingly small for a large number of storage media to be managed. For this reason, usually, most media are placed in a hangar, and when it is necessary to read / write the media, the media is loaded from the hangar to the R / W device using a robot, and then the actual reading / writing is performed. The time from when the media is loaded into the device using the robot until the media becomes readable and writable is called the transport time, and this time is longer than the normal disk access time. In data recovery in library equipment,
The failed media is ejected from the R / W device, and a spare media for writing the recovered data is read / written.
Reduces media recovery time because it takes two transport times to load the device and two times to regenerate data, that is, recalculate parity and reconstruct data on spare media. Is an important issue.

SUMMARY OF THE INVENTION An object of the present invention is to provide a technique for shortening the transport time of a medium and the time for regenerating data in a storage device system based on a portable medium, particularly when restoring a storage medium failure. It is to provide the technology to do.

[0018]

Means for achieving the above object will be described below.

The storage system according to the present invention has a plurality of portable media and a spare medium, and a plurality of disk arrays (RAID) having redundant data are constituted by the plurality of portable media. It has a spare recording / reproducing device, a storage for storing portable media, a transport device for transporting the portable media, and a table for storing information on the portable media loaded in the recording / reproducing device.

In the present storage system, when a failure occurs in a portable medium, a spare recording / reproducing device is used for recovery.
A spare medium was always loaded in the spare recording / reproducing apparatus, and at the time of recovery, a failure occurred by using the spare recording / reproducing apparatus loaded with the spare medium as a write destination of the regenerated data. The transport time for discharging the medium and loading the spare medium is omitted.

Further, in the recovery processing of the storage system, the failed medium is used for the recovery processing without being ejected from the recording / reproducing apparatus. When a readable area exists in the medium in which the failure has occurred, the readable area is read, and the content is written to a spare medium loaded in a spare recording / reproducing apparatus, thereby reducing the time for data regeneration.

Alternatively, in a storage system having an auxiliary storage device having a storage capacity equal to or larger than one portable medium in addition to the above storage system, if a failure occurs in a medium, the destination of the regenerated content is written to an auxiliary destination. By using a storage device, the time required to discharge a failed medium and to transport a spare medium can be omitted.

Further, in the above storage system, when a readable area exists in a medium in which a failure has occurred, the readable area is read and the contents are written to an auxiliary storage device, thereby reducing the time for data regeneration.

[0024]

BEST MODE FOR CARRYING OUT THE INVENTION

(1) First Embodiment FIG. 1 shows a configuration example of a storage system to which the present invention is applied.
The storage system includes a control device 1 and a storage device 2 for storing data. The control device 1 is connected to the host 3 via the host interface 11, and is connected to the storage device 2 via the drive interface 12 and the transport control unit 13. Further, in the control device 1, a processor 14 for executing the control program and an RO for storing the control program are provided.
M15, RA used as a buffer for storing control information for control and for temporarily storing write / read data
M16, an NVRAM 17 for storing information about media held in the storage system, and a parity generation unit 18 of dedicated hardware for calculating parity.

The storage device 2 includes a medium 20 and a medium 2
0, a drive 22 for reproducing / recording the medium 20, a transport device 23 for transporting the medium 20 between the storage 21 and the drive 22, a medium 2
0 is input / output to / from the storage device 2.

Each drive 22 is managed by assigning a serial number from 1 to m, and one drive 22 can be loaded with only one medium 20 at a time. Some of the m units are spare drives in case of drive failure, and the spare drives are always loaded with unused spare media. In this embodiment, the number of spare media is one.

The hangar 21 is composed of a set of racks 25 for storing the media 20.
One medium 20 is inserted, and each rack 25 is assigned a serial number from 1 to L and managed. Also, each media 20
The media always stored in the same rack 25 and stored in the rack number h are given the same serial number as the rack number, and are called media h.

Referring to FIG. 2, a data storage format of the present storage system will be described. In the media 20 in the actual storage device 2, the data d1, d2,
.. are divided and arranged on each medium 20 constituting a set of RAIDs as shown in FIG. Individual data such as d1 is a block 2 in the unit of writing from the host 3.
Call it 7. Hereinafter, the medium 20 in the actual storage device 2 is referred to as a physical medium 28. Also, p1
Is the parity (calculation result of exclusive OR) generated from d1, d2, d3, and d4, and hereafter p2 is d5, d
Parity generated from 6, d7, d8, p3 is d9,
Parities generated from d10, d11, d12 ...,
Becomes A block of data such as d1 and d2 is called a data block, and a block where parity is placed such as p1 and p2 is called a parity block.

A set of data blocks and parity blocks required to create a parity block is called a parity group 29.

However, from the host, the physical medium 28
Cannot be individually recognized, and data blocks d are continuously recorded on one logical medium as shown in FIG.
.., 1, d2,... Are arranged, and it looks as if there are a plurality of these logical media. The logical medium visible to the host is called a logical medium 26.

FIG. 3 shows a mapping table 30 for managing the correspondence between the logical medium 26 and the physical medium 28 by the system. In the present storage system, the logical media 26 managed by the system are managed by assigning serial numbers 1 to k, similarly to the physical media 28. In each entry, the number of the logical media 26 (hereinafter, referred to as a logical media number) and the numbers of all the corresponding physical media 28 (hereinafter, referred to as physical media numbers) are written. When an access to the number is made, all media of the physical media number corresponding to the logical media number can be specified by referring to the mapping table 30. This mapping table 3
0 is stored in the NVRAM 17 of the control device 1.

In the RAM 16 of the control device 1, a drive management table 40 is stored as shown in FIG. 4, which records the correspondence between the drive numbers and the numbers of the physical media loaded in the respective drives at that time. The drive management table 40 has entries for all drives in the storage system, and when a physical medium 28 is loaded in the drive, a physical medium number is recorded in the entry. Nothing is written in the drive management table 40 for a drive in which the physical medium 28 is not loaded.

When the physical medium 28 being accessed is in a failure state, "failure" is written in this entry. Since a spare medium is always stored in the spare drive, “spare” is written in the drive management table 40 to indicate that a spare medium is loaded.

Next, the flow of processing when a request is received from the host 3 will be described. The host 3 sets the target logical medium 26
Before accessing, a mount request for loading the medium 20 into the drive 22 is issued, and then a read / write request is issued. When the read / write process has been completed, the medium 20 is removed from the drive 22 and an unmount request to store the medium 20 in the storage 21 is issued.

The flow of the mounting process will be described with reference to FIG.
When issuing a mount request from the host 3, a logical media number is specified. In the control device 1, the mapping table 30
, All physical media numbers corresponding to the logical media numbers are determined (step 1001). When loading the physical medium 28 into the drive 22, the control device 1 refers to the drive management table 40 to check whether the target physical medium 28 has already been loaded into the drive 22 (steps 1002 and 1003). If not, the drive management table 40 is searched for a drive 22 in which another physical medium 28 is not loaded (step 1004), and the physical medium 28 is loaded into the drive 22 (step 1005). When the physical medium 28 is loaded, the number of the loaded physical medium 28 is recorded in the entry of the drive management table 40 (step 1006). When all the physical media 28 have been loaded into the drive 22 and the physical media numbers have been recorded in the drive management table 40, the control device 1 returns a mount completion report to the host 3 (step 100).
8).

Referring to FIG. 6, the reading process will be described. At the time of reading, the host 3 specifies a logical medium number and a read position on the logical medium 26, that is, a logical address. The control device 1 converts a set of a logical media number and a logical address into a set of a physical media number and a read position (called a physical address) on the physical medium 28. This conversion is fixed, and a conversion program is written in the ROM 15, and this conversion is performed when accessing the medium (step 1101). After the conversion, the control device 1 specifies the drive 22 corresponding to the physical medium 28 from the drive management table 40, reads the block of the physical address of the physical medium 28 in the drive (step 1102), and returns it to the host. .

FIG. 7 illustrates the writing process. At the time of writing, the logical media number and the logical address of the data block to be written (hereinafter referred to as new data) are designated by the host 3 in the same manner as the reading, and the physical media number and the physical address (hereinafter, this physical A set of a media number and a physical address is converted to a data address (Step 1101). Further, at the time of writing, a physical media number in which a parity block belonging to a parity group of new data is written and a physical address (hereinafter, referred to as a physical address). The combination of the physical media number and the physical address is called a parity address) is also calculated (step 1202). This conversion is also fixed, and is calculated by a conversion program written in the ROM 15. Subsequently, a block of a data address (hereinafter, referred to as old data) and a block of a parity address (hereinafter, referred to as old parity) are read out to the RAM 16 (step 1203, step 1).
204). In step 1205, the parity control unit 18 calculates a parity (this is called a new parity) from the old data, the old parity, and the write data block.
At 206, the data block to be written is written to the data address, and at step 1207, the new parity is written to the parity address.

Next, the unmount processing will be described with reference to FIG. In the unmount processing, the control device 1 specifies the physical medium 28 corresponding to the logical medium 26 with reference to the mapping table 30 (step 1001), and further searches the drive management table 40 for the drive 22 containing the physical medium 28. The process up to the step (step 1002) is the same as the mount process. Thereafter, all the target physical media 28 are returned from the drive 22 to the corresponding storage 21 (steps 1303 and 1304).

Next, when accessing the physical media numbers s to t, a procedure for recovery processing when a failure occurs in the physical medium u (s ≦ u ≦ t) and the access becomes impossible is described below.
This will be described with reference to FIG. The failure of the physical medium 28 is detected when accessing the medium. When the drive 22 cannot access the physical medium 28, the control device 1 enters a recovery process.

Normally, in the case of a failure of the physical medium 28, it is rare that the entire area of the medium becomes inaccessible.
A fault is determined when a certain threshold value set in the storage system is exceeded. For this reason, even in the physical media determined to be faulty (hereinafter referred to as faulty media), accessible blocks are often left, so in this system, blocks are read from the faulty media as much as possible and copied to the spare media. .

First, a recovery pointer p of the physical medium u determined to be faulty is prepared, and p = 0 (step 200).
1). Then, the block p is sent from the drive 22 to the control device 1.
(Step 2002). If the block 27 is accessible (step 2003), the contents of the read block 27 are copied to the block p of the spare medium loaded in the spare drive (step 2).
004). If the access is not possible, the block p of all other physical media belonging to the parity group is read out to the control device 1 (step 2005). At the same time, the exclusive OR is calculated by the parity generation unit 18 (step 2006), and the calculation result is copied to the spare medium (step 2007). If p is not the last block of the medium, the process is repeated from step 2002 to reconstruct the recovery data in the spare medium (steps 2008 and 2009).

If a read request arrives during the recovery process and the block to be read exists on the faulty medium, step 2002 is executed. If the block cannot be read, steps 2005 and 2
The contents can be read out by executing 006 to restore the content from another medium and returning it to the host. When a write request is received, writing may be performed on a spare medium.

When the recovery of the data in the medium is completed, the spare medium is switched to the failed medium u (step 2).
010). Switching is performed by rewriting the value of the drive management table 40. “Failure” is written for the drive 22 containing the faulty media, and access to the drive 22 containing the faulty media is prohibited. For the drive 22 containing the medium 20 which has been a spare medium, the medium number u of the failed medium is assigned, and the spare drive is replaced with the normal drive 2.
2 is logically changed so that the drive 22 containing the faulty medium is treated as a spare drive. Thereafter, the failed medium is discharged from the input / output port 24, another spare medium is loaded into the drive 22 containing the failed medium (step 2011), and the content of the drive management table 40 is set to "spare" to be used as a spare drive. Switching (step 2012), the recovery processing ends.

(2) Embodiment 2 FIG. 10 shows a configuration example of a storage system to which the present invention is applied.

The storage system comprises a control device 1 and a storage device 2 for storing data. The control device 1 is connected to the host 3 via the host interface 11, and is connected to the storage device 2 via the drive interface 12 and the transport control unit 13. Further, in the control device 1, a processor 14 for executing the control program, a ROM 15 for storing the control program, a RAM 16 used as a buffer for storing control information for control and for temporarily storing write / read data, and a storage system. It includes an NVRAM 17 for storing information about the held media, and a parity generation unit 18 of dedicated hardware for calculating parity.

The storage device 2 includes a medium 20 and a medium 2
0, a drive 22 for reproducing / recording the medium 20, a transport device 23 for transporting the medium 20 between the storage 21 and the drive 22, a medium 2
0 is input / output to / from the storage device 2. Further, the storage device 2 has an auxiliary disk device 220 having a capacity equal to or larger than the capacity of the medium 20. Instead of the disk device, a nonvolatile readable / writable storage device such as a flash memory may be used.

The drive 22, the physical medium 28,
The logical media 26 and the rack 25 are assigned serial numbers and managed, and the mapping table 30 and the drive management table 40 are stored in the RAM 16 and the NVRAM 17 as in the first embodiment.

Next, the auxiliary disk device 220 will be described. In this storage system, the auxiliary disk device 220
As one or more drives loaded with media. The management of the auxiliary disk device 220 is performed by the disk management table 230 of FIG.

The auxiliary disk device 220
Is used as a storage area for temporarily storing the recovered data when the system performs recovery processing.
If the system has j drives and the capacity of the auxiliary disk device 220 is the capacity of k physical media, it is assumed that k drives exist, and the drive numbers j + 1, j + 2, .... j + k is assigned.

Each entry of the disk management table 230 has a drive number and a physical medium number as in the drive management table 40, and additionally records a set of a start address and an end address on the auxiliary disk device 220. ing. The area from the start address to the end address is equal to the capacity of one medium. When the physical media number is recorded in the entry, the auxiliary disk drive 220
Indicates that the content of the physical medium 28 has been written, and when not recorded, indicates that the area is an unused area. In the first embodiment, the drive 2 in which the physical medium 28 is loaded is determined from the physical medium number.
When searching for the drive management table 40, the drive management table 40 is referred to in the second embodiment.
The corresponding physical media number is searched for by referring to both the disk management table 230 and the disk management table 230.

Referring to FIG. 12, a description will be given of a procedure of a recovery process when the medium u (s≤u≤t) becomes inaccessible due to a failure while accessing the physical media numbers s to t. .

First, a recovery pointer for the failed medium u
Prepare p and set p = 0 (step 3001). Subsequently, referring to the disk management table 230, an unused area of the auxiliary disk device 220 is searched, and an area for the faulty medium u is allocated (step 300).
2). Subsequently, the physical medium u is read from the first block (step 3003). If the block is accessible, the contents of the read block are written to the auxiliary disk device (step 3005). The writing position to the auxiliary disk device can be easily calculated from the start address and the pointer p recorded in the disk management table.

If the area of the faulty medium u is inaccessible, the block p of all other physical media belonging to the parity group is read out to the control device 1 (step 3006). At the same time, the exclusive OR is calculated by the parity generation unit 18 (step 300).
7), and copy the calculation result to the auxiliary disk device 220 (step 3008). If p is not the last block of the medium, the process is repeated from step 3003 to reconstruct the recovery data in the auxiliary disk device 220 (steps 3009 and 3010).

When the recovery of the data in the faulty medium is completed, the auxiliary disk device 220 is switched to the faulty medium u and started to be used (step 3011). Switching is performed by rewriting the value of the drive management table 40 as in the first embodiment.

When the data recovery is completed,
u is discharged from the input / discharge port 24, and the area of the faulty medium u in the drive management table 40 is left blank (step 3012).

Thereafter, the recovery is completed by copying all the contents from the auxiliary disk device 220 to the spare medium (step 3013), but this copy does not necessarily have to be performed immediately after the step. If a read / write request is received from the host 3 before copying is performed, the auxiliary disk device 220 may be used as a physical medium u, and the copy will not access the medium u. This may be performed, for example, when an unmount request for the logical medium 26 to which the medium subjected to the recovery processing belongs comes and all the physical media 28 belonging to the logical medium 26 are returned to the storage 21.

[0057]

According to the present invention, since a spare medium is pre-loaded in a spare drive or an auxiliary disk device is provided in place of a spare medium, a spare medium is immediately provided when a failure occurs in a medium. Data can be restored to the media or the auxiliary disk device, and the media transport time can be reduced. Further, data is read out of a range readable from the failed medium and copied to a spare medium or an auxiliary disk device, so that data recovery processing can be hastened.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration example of a storage system to which the present invention is applied.

FIG. 2 is a diagram showing an example of the arrangement of data and parity on a medium.

FIG. 3 is a diagram showing a configuration of a mapping table.

FIG. 4 is a diagram showing a configuration of a drive management table.

FIG. 5 is a diagram showing a flow of a mounting process.

FIG. 6 is a diagram showing a flow of a reading process.

FIG. 7 is a diagram showing a flow of a writing process.

FIG. 8 is a diagram showing a flow of an unmount process.

FIG. 9 is a diagram showing a flow of a recovery process.

FIG. 10 is a diagram showing a configuration example (second embodiment) of a storage system to which the present invention is applied.

FIG. 11 is a diagram showing a configuration of a disk management table.

FIG. 12 is a diagram showing a flow of a recovery process.

[Explanation of symbols]

1: control device, 2: storage device, 3: host, 11: host interface, 12: drive interface, 13: transport control unit, 14: processor 14, 15: ROM, 16: RA
M, 17: NVRAM, 18: parity generation unit, 20: media 20, 21: hangar, 22: drive, 23: transport device, 2
4: Input / output port, 25: Rack, 26: Logical media, 2
7: block, 28: physical media, 29: parity group, 30: mapping table, 40: drive management table, 220: auxiliary disk device, 230: disk management table

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁶ Identification code FI G11B 20/18 570 G11B 20/18 570Z

Claims (4)

[Claims] A plurality of portable recording media for storing data and error correction data; a spare portable recording medium for the recording media;
A plurality of recording / reproducing devices for recording / reproducing the portable recording medium, a spare device for the recording / reproducing device, a storage for storing the portable recording medium, a storage for storing the portable recording medium, and a recording / reproducing device In a storage device having a transfer device for transferring between,
A storage device and a fault recovery means, wherein a spare device is always loaded with a spare recording medium, and when a portable recording medium fails, the spare device loaded with the spare portable recording medium is used for recovery. 2. The fault recovery means according to claim 1, wherein when a faulty portable recording medium has a readable range, the content is read from the readable range and copied to a spare portable recording medium. Failure recovery means for reconstructing the contents of a portable recording medium in which a failure has occurred. 3. A storage device system having an auxiliary storage device having a capacity equal to or greater than one portable recording medium in addition to the storage device according to claim 1, wherein the storage device is connected to the storage device when a failure occurs in the portable medium. A storage device and a failure recovery means, wherein the contents of a portable recording medium in which a failure has occurred are reconfigured in an auxiliary storage device. 4. The fault recovery means according to claim 3, wherein when the faulty portable recording medium has a readable range, the content is read from the readable range and copied to an auxiliary storage device. Failure recovery means for reconstructing the contents of a portable recording medium in which a failure has occurred.