JPH05100801A - Highly reliable disk subsystem - Google Patents

Abstract

(57) [Abstract] [Object] The present invention relates to a disk controller, and more particularly to a method for improving fault tolerance against an access failure of a drive in a large-sized disk subsystem in which a large number of drives are connected. [Structure] Collects statistical information on correctable access errors related to the current drive (track), and if it determines that there is a sign of failure of the current drive (track) based on the collected statistical information, the disk control device determines that the current drive This is achieved by copying the data from the (track) to the spare drive (track) and then switching from the active drive (track) to the spare drive (track). [Effect] The disk control device saves the data of the working drive (track) determined to have a sign of failure to the spare drive (track), and then uses the spare drive (track) as a new working drive (track). Since it is used, it is possible to prevent the failure of the working drive proactively, reduce the probability of the drive failure, and improve the reliability of the disk subsystem.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk controller and, more particularly, to a method for improving the fault tolerance against an access failure of a drive in a large-sized disk subsystem connecting a large number of drives.

[0002]

2. Description of the Related Art When a host computer writes data to a disk drive once, the disk controller writes the same data to a primary and secondary pair drive,
Even if an access failure occurs in either the primary or secondary drive, the normal drive continues the access to improve the fault tolerance against the access failure of the drive (double writing by the disk controller) Access March / April issue (1988) page 9 to page 10 (ACCESS MAR / APR 1988 pp9-
10).

[0003]

The dual writing by the primary and secondary pair drives has high fault tolerance but also high cost. Therefore, the application is often limited to the online processing system drive. In other words, although the access frequency is not so high, even if there is a demand for improving the fault tolerance of an offline processing system drive that has a great influence on the recovery processing once the access failure occurs, there are generally many, so there are many Applying double writing is difficult in terms of cost.

An object of the present invention is to provide a method of reducing the probability of failure by performing preventive maintenance on the drive of the above-mentioned offline processing system.

[0005]

[Means for Solving the Problems] The above-mentioned object is to collect statistical information on correctable access error occurrences relating to the current drive (track), and based on this, a sign of failure of the current drive (track) can be obtained. If it is determined that there is, the disk controller copies the data from the current drive (track) to the spare drive (track), and then switches from the current drive (track) to the spare drive (track). ..

[0006]

[Operation] After the data of the working drive (track) determined to have a sign of failure is copied to the spare drive (track) by the disk controller, the spare drive (track) is used as a new working drive (track). By doing so, the failure of the working drive can be avoided proactively.

[0007]

FIG. 1 shows an embodiment in which the present invention is applied to a disk control subsystem. In FIG. 1, the disk control device 100 includes a host computer 10 on the upper side.
Drive 1 that is connected to 9 and is a storage medium on the lower side
11, 112, 113, 114 are connected. The disk control device 100 reads and writes data on these drives in response to a request from the host computer 109. The operation of the disk control device 100, which is the main content of the present invention, will be described below with reference to the configuration of the disk control device 100 in FIG.

The processors 102 and 103 incorporated in the disk controller 100 actually perform the data transfer between each drive and the host computer 109.
These processors have a channel interface 10
1, via a drive interface 107, 108, respectively connected to the host computer 109 and each drive. The control memory 104 is a common memory that can be accessed by all the processors, and is a disk controller 10.
Common control information for 0 to access the drive is stored. The contents of the common control information will be described below as needed.

FIG. 2 shows a drive control block 200 (Device Control Block; DCB for short) which is common control information.
Indicates. Each device is a disk controller 1 in the DCB.
Drive number 201 (Device Conn
ection Address; DCA for short), and 202 for each drive type drive indicating the type of model of the drive. Further, each drive is distinguished into two types, a current drive in which the disk control device 100 actually inputs / outputs data in response to a request from the host computer, and a spare drive for backup of the current drive. Whether it is used or not (current / spare) 20
Store in 3. The contents of other DCB control information will be described later.

FIG. 3 shows replacement track management information 300 which is common control information. For each drive, an unused spare track that can be used as a backup for the working track that stores data and a spare track that is already being used as a backup for the working track are separately managed.
Specifically, the replacement track control block that stores the track number of the spare track is managed in a queue by dividing it into an unused track and a used track.

FIG. 4 shows track access / temporary error statistical information 800 which is common control information. Each time the processor in the disk control device 100 detects a correctable data transfer error in the data read process on the track of each drive, the corresponding track column 802 of the corresponding drive 801 of the track access / temporary error statistical information 800. The total number of temporary access errors 803 stored in is incremented. These pieces of information are used to detect the onset of a failure of the truck or the drive, and the details will be described later.

Next, the operation of each processor in the disk controller 100 according to the present invention will be described. FIG. 5 shows a flow of drive health check processing. This process is repeatedly executed by each processor in the disk control device 100 at a certain cycle. First, 0 is set as the drive number in the check drive number storage table 800 (FIG. 8) which is the common control information. First, the number of the drive to be checked is read from the check drive number storage table 800, and the local memory (15
No. 2, 153) (501).

Next, it is checked whether the swap request 901 in the swap request table 900 (FIG. 9) of the drive having the drive number is'off '(50).
2). Here, in the swap request table 900,
When the swap request 901 is “off”, it means that there is no request to switch to the spare drive or the spare track for the drive. Swap request 901 is'off
If it is', it is checked whether or not the drive number is equal to the maximum drive number. If they are equal, -1 is set in the check drive number storage table 800.

Then, the track search pointer, which is a local variable, is cleared to zero (505), and for the drive having the check drive number previously stored in the local memory of its own processor, the track number equal to the track search pointer is set. It is determined whether or not the number of temporary access errors of the own track exceeds the upper limit (threshold). If the number of temporary access errors exceeds the upper limit (threshold), the total number of temporary error frequently occurring tracks 804 is incremented by +1 (514), and the temporary error of the drive is increased. It is determined whether or not the total number of frequently-occurring tracks 804 exceeds the upper limit (threshold), and if it exceeds, the processed mark 801 of the drive is set, and then the swap request of the drive is registered (511). The check drive number is incremented by 1 and the process ends. That. If it does not exceed the upper limit, after setting the processed mark 805 of the track, the swap request of the track is registered (509),
The track search pointer is incremented by 1 and the health check of the next track is performed.

FIG. 6 is a flow chart showing the processing contents of the track swap request registration processing 509 in FIG. First,
The lock of the drive having the track to be swapped is acquired (601), and one replacement track to be the copy destination is selected from the unused replacement track queue 310 (FIG. 3) (604). If there is no unused replacement track, it is displayed that the track swap is impossible (606), the acquired lock is released, and the processing is ended. If there is an unused replacement track, the replacement track number 204 is set in the DCB of the swap target drive, and the own drive number is set as the copy destination drive number 205. Further copy request 206
Set to track copy, and copy start track 212
Set the swap target track number to. Then, the swap request 208 is turned on (708), and the acquired lock is released.

FIG. 7 is a flow chart showing the processing contents of the drive swap request registration processing 511 in FIG. First,
The lock of the drive to be swapped is acquired (701), the DCB of the drive connected to the disk control device 100 is checked, and the application distinction 203 is the spare drive,
Moreover, one drive having the same drive type as the swap target drive is selected (704). If there is no spare drive, swap not possible is displayed (706), the above lock is released, and the process ends. If there is a spare drive, the spare drive number is set in the copy destination drive number 205 of the DCB of the swap target drive, and the copy request 206 is made.
Is set to drive copy, and copy start track 21
The copy start track number and the copy end track number corresponding to the drive type are set in 2 and the copy end track 213, respectively (707). Then, the swap request 208 is turned on (708), and the acquired lock is released.

FIG. 10 is a flow chart showing the processing contents of the track swap processing. This processing is also executed by an arbitrary processor in a certain cycle. First, a swap process candidate drive number is read from the swap process drive number storage table 900 and stored in the local memory of the processor (1001). If the drive number is equal to the maximum drive number, the table 900 is cleared to zero (1003), and if not equal, the table 900 is incremented by 1 (1015). Next, the DCB of the drive
Check if the swap request 208 inside is'on ',
If not'on ', the process ends. If it is'on ', it is determined whether the copy request of the drive is a track copy or a drive copy. If it is a drive copy, the drive swap processing is performed and the processing ends. If it is a track copy, the swap request 208 is processed (1006), the lock of the drive is acquired (1007), the track data from the track having the copy start track number is loaded into the cache memory 105 (1008), and the replacement is performed. After storing the above load data for the track having the track number (1010), a swap processing request is issued.
Set it to f '(1010) and release the above lock, and finish the process.

FIG. 11 is a flow chart showing the processing contents of the drive swap processing. This processing is also executed by an arbitrary processor in a certain cycle. First, the swap request 208 is set in process, and the copied track number 214 is initialized to the value of (copy start number-1) (1101). The following is repeated until the copied track number 214 becomes equal to the copy end track number. First, the locks of the copy source drive (the drive concerned) and the copy destination drive are both acquired (1103), and track data is loaded from the uncopied track (copied track + 1 track) into the cache memory 105 (1105). ),
After storing the load data in the track of the copy destination drive (1010), the acquired lock is released and the number of the copied track is incremented by 1 (1107). Then, in order to be able to receive the data input / output from the host computer to the drive under the swap processing,
After the swap processing is interrupted for a certain period of time (1108), the next track copy processing is started.

At the time when the copied track number 214 becomes equal to the copy end track number, the locks of the copy source and copy destination drives are both acquired, and destaging processing of the differential data to the copy destination drive (1111), and , Perform access path swap processing (1112),
Finally, the acquired lock is released (1114), and the process ends.

In the following, destage processing of the difference data (11
The content of 11) will be described, but before that, the management of the differential data during the drive swap processing will be described. When a write request is sent from a host computer to a track on the copy source drive, if the track number is greater than or equal to the copied track number, the data write target track is an uncopied track. It suffices to simply write the data on the track, and at the end of the copy, the data does not differ between the copy source drive and the copy destination drive. On the other hand, when the track number is smaller than the copied track number, the track to which data is to be written is the copied track, and therefore when data is written to the track,
At the end of copying, the data differs between the copy source and copy destination drives.

To deal with the latter case, as shown in FIG. 14, when data is written to a copied track on the copy source drive, the track (such a track is referred to as a differential track) is written. The write data for the call) is also held in the cache body 402 in the cache 105. For example, this is the case when writing to track 2 (TR2). In this case, first
A slot control block (SC that holds the address of an empty data storage location in the cache body 402)
B) Select one SCB (406) in the queue 43, store the write data in the storage location (slot 2) pointed to by the SCB, and store the write data in the track table 403 provided in the directory 401 in the cache 105. The pointer to the SCB is stored in the entry corresponding to TR2.
By doing this, it is possible to know which difference track is and what write data is by looking at the cache. The data of the differential track is at the stage when the drive copying is completed, that is, the copied track number 214
When becomes equal to the copy end track number, it is necessary to write again to the copy destination drive.

Now, FIG. 12 is a flow showing the contents of the destage processing (1111) of the difference data. First, it is checked whether or not the differential track is linked to the track table 403 of the directory 401 (120
1) If there is no link, the process ends. If they are linked, all the write data on the cache pointed to by the SCB of the differential track is written to the corresponding track of the copy destination drive (1204).

Next, access path mapping table 1
400 (FIG. 15) will be described. The access path mapping table 1400 (FIG. 15) shows the drive identification number CCA (Channe) seen from the host computer 109.
l Connection Address), and the identification number DCA (DeviceCon
nection Address) correspondence table. The drive number (CCA) designated by the input / output request of the host computer 109 is D by the disk controller 100.
Converted to CA, the drive corresponding to DCA is actually accessed. Specifically, assuming that the value of CCA is k in FIG. 15, C of the access path mapping table 1400 is set.
The value i in the Bk column of the CA mapping information 1401 is read and set as the value of DCA, and the drive of drive number i is actually accessed. With this mapping function, even if the current drive is switched to the spare drive in the above drive swap processing, the correspondence between CCA and DCA is changed so that the drive number does not need to be changed on the host computer side.

Finally, the access path swap processing using the above access path mapping table 1400 will be described according to the flow of FIG. First, the copy source drive DCA (denoted as i) and the copy destination drive D
The CA (j) is stored (1301). Then DC
Value in the Bi column of the A mapping information 1402 (k)
And the value (l in the Bj column of the DCA mapping information 1402
() Is read (1302). Then, the value 1 is written in the Bi field of the DCA mapping information 1402, and C
The value i is written in the Bl column of the CA mapping information 1401 (1303). Finally, DCA mapping information 1402
Write the value k in the Bj column of CCA mapping information 1
The value j is written in the Bk column of 401 (1303).

By these processes, the disk controller copies the data of the working drive (track) determined to have a sign of failure to the spare drive (track), and then the spare drive (track) is replaced with a new working drive. It can be automatically used as a (track),
Improves disk subsystem reliability.

[0026]

As described above, the disk controller copies the data of the current drive (track), which is judged to have a sign of failure, to the spare drive (track), and then the spare drive (track) is replaced with a new current drive (track). Therefore, it is possible to prevent the failure of the current drive proactively, reduce the probability of occurrence of the drive failure, and improve the reliability of the disk subsystem.

[Brief description of drawings]

FIG. 1 is a control configuration diagram showing an embodiment of the present invention.

FIG. 2 is a configuration diagram of a drive control block.

FIG. 3 is management information of a replacement truck.

FIG. 4 is a diagram showing track access / temporary error statistical information.

FIG. 5 is a flow chart of drive health check processing.

FIG. 6 is a flow chart of a track swap request registration process.

FIG. 7 is a flow chart of drive swap request registration processing.

FIG. 8 is a check drive number storage table.

FIG. 9 is a swap processing drive number storage table.

FIG. 10 is a flow of track swap processing.

FIG. 11 is a flow chart of a dry swap process.

FIG. 12 is a flowchart of a differential track destage process.

FIG. 13 is a flow chart of access path swap processing.

FIG. 14 is a configuration diagram of a cache memory.

FIG. 15 is a configuration diagram of an access path mapping table.

[Explanation of symbols]

100 ... Disk control device, 102, 103 ... Processor, 104 ... Control memory, 109 ... Host computer, 111, 112, 113, 114 ... Drive.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Yoshiro Shirayanagi 2880 Kozu, Odawara City, Kanagawa Stock Company Hitachi Ltd. Odawara Factory

Claims (2)

[Claims] 1. A disk subsystem comprising a disk controller and a drive which is a magnetic storage medium, wherein the drive connected to the disk controller is divided into one or more working drives and one or more spare drives for disk control. Registered in the device in advance, the number of correctable data transfer system failures that occurred during input / output from the host computer to the active drive is counted for each drive by the disk controller, and the number of occurrences is a predetermined value. A high reliability system for a disk subsystem, characterized in that the contents of the active drive are exceeded by the disk controller to an appropriate spare drive, and the spare drive is used as a substitute for the active drive. 2. The disk control device according to claim 1, wherein the tracks of the drive are divided into an active track and an alternate track and registered in advance in the disk control device, and when the input / output from the host computer to the active track is executed. The number of correctable data transfer system failures that have occurred is counted for each track by the disk control device, and for the current track in which the number of occurrences exceeds a predetermined value, the disk control device selects an appropriate number in the same drive. A high reliability system for a disk subsystem, characterized in that, after copying the contents to a spare track, the replacement track is used as a working track instead of the working track.