JPH09218750A - Control method of magnetic disk sub-system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce probability for a data loss by evading writing impossible data to a magnetic disk and recovering data saved in an auxiliary area to a cache memory. SOLUTION: When a pin control flag inside a cache memory pin data control table 125 is invalid in stopping a plan, pin data 124 is stored in a pin data control table 133 inside a storage device 130. Then, when the data presence flag of pin data 124 to be saved is invalid, the auxiliary area where pin data 124 is stored is recognized, saving is executed and a track data invalid flag inside the pin data control table 133 is made to be valid. When starting a system, whole information of saved pin data 124 is restored to the cache memory pin data control table 125 and, after that, the pin control information flag of the pin data control table 133 is made to invalid.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of controlling a magnetic disk subsystem comprising a battery-backed storage control device with a cache memory and a magnetic disk device, and a method for controlling a cache memory before writing data in the magnetic disk device. The present invention relates to a magnetic disk subsystem with a cache memory, which is capable of temporarily storing data and then writing the data from a cache memory to a magnetic disk device.

[0002]

2. Description of the Related Art Known examples of cache memory control technology are shown below.

The data transfer rate between the cache memory and the disk device is slower than the transfer rate between the host controller and the cache memory. Therefore, Japanese Patent Application No. 55-157053
In the technology described in (3), the storage controller having the cache memory writes in response to a write instruction from the central processing unit when the data transfer between the central processing unit and the cache memory in the storage control unit is completed. A high-speed response is achieved to the central processing unit that issues a write report and issues a write instruction. Data transfer from the cache memory to the disk device is performed later by the storage control device asynchronously with the write instruction from the central processing unit. This control is called write-after control.

[0004]

Conventionally, in a magnetic disk subsystem with a cache memory, when there is a possibility of data loss in the cache memory at the time of a planned stop, etc. When there is data that cannot be written to the magnetic disk device, only the information that there is data that cannot be written to the magnetic disk device in the cache memory was left in the volatile memory with battery backup, but the data itself was saved. There wasn't.

Further, in the case of a planned stop, a volatile memory such as a cache memory loses data in the volatile memory, but Japanese Patent Laid-Open No. 60-45857 discloses a technique for transferring the data in the volatile memory to a non-volatile memory. I am presenting
There is no suggestion of a method for effectively holding data that cannot be transferred when writing to the nonvolatile memory is not possible.

Further, although the battery-backed cache memory is supposed to be able to maintain data without having to write it in the magnetic disk device, there is a limit in terms of data integrity because the backup capacity of the battery itself is limited. It was

An object of the present invention is to not only leave information that unwritable data remained in the cache memory in the cache memory due to a failure of the magnetic disk device at the time of planned stop, but also to write unwritable data. Another object of the present invention is to provide a method of controlling a magnetic disk device, which makes it possible to save the data in a spare area of the magnetic disk device or another auxiliary storage device and recover the data upon restart.

At the same time, another object of the present invention is to provide a method for controlling a magnetic disk device, in which the magnetic disk device also has position information of unwritable data and the like, which facilitates recovery.

[0009]

In order to achieve the above object, a method of saving / restoring data from a cache memory to a storage device in a magnetic disk subsystem with a cache memory according to the present invention writes data from the cache memory to the magnetic disk device. The position of the impossible data part is specified, the position information of the data is stored in the table that manages the unwritable data of the magnetic disk device, and it is judged whether it can be stored in the spare area in the magnetic disk device. The unwritable data to the magnetic disk,
If the battery backup of the cache memory is cut off when restarting, check if there is unwritable data from the table that manages unwritable data and restore the data saved in the spare area to the cache memory. Then, in place of the spare area of the magnetic disk device, unwritable data is saved and restored in the auxiliary storage device.

Conventionally, when there is unwritable data remaining in the cache memory due to a failure of the magnetic disk device at the time of planned shutdown, only the position information of the data is stored in the battery-backed volatile memory to restart. Sometimes only information was available that non-writable data existed in the cache memory before the planned shutdown. However, with that alone, there is still the possibility of data loss even if the cache memory power supply is duplicated.

Therefore, according to the present invention, when there is unwritable data due to a failure of the storage device at the time of planned stop,
Position information of unwritable data is stored in a table that manages unwritable data of the magnetic disk device. If the spare area of the magnetic disk device is sufficient, the unwritable data is saved.
Then, at the time of restart, it is confirmed whether the battery backup of the cache memory has not expired, and if so, whether or not there is unwritable data due to a failure of the magnetic disk device is searched for in the table,
The existence of unwritable data is confirmed in the table that manages unwritable data, and if unwritable data remains in the spare area, it can be restored to the cache memory.

Also, unlike a conventional magnetic disk subsystem with a cache memory, a magnetic disk array with a cache memory is shown as a conventional single or multiple magnetic disks by using a plurality of small magnetic disk devices. In the device, a spare disk (spare disk)
To improve maintainability. However, this spare disk is rarely used except during maintenance. Therefore, in the present invention, data that cannot be written to the magnetic disk device from the cache memory due to a failure of the magnetic disk device in the array configuration can be saved to the spare disk at the time of the planned stop.

[0013]

BEST MODE FOR CARRYING OUT THE INVENTION The battery-backed magnetic disk subsystem with a cache memory of the present invention temporarily stores data in the cache memory before writing the data to the magnetic disk device by the storage controller. Although data is stored in this cache, when the data is written to the magnetic disk device regularly or irregularly, it may not be writable due to a failure of the magnetic disk device. This data is called pin data. This pin data cannot be written from the cache memory to the magnetic disk device, but it remains in the cache memory, so
Data is guaranteed when the cache memory is energized, and there is no problem. However, when the state of the planned stop occurs and the battery backup of the cache memory is cut off, the reliability of the data becomes doubtful because the pin data cannot be written in the relevant part of the magnetic disk device.

Therefore, data including pin data is saved in a spare area of the magnetic disk device or a spare disk,
A method of recovering the pin data at the time of restart by providing the related information of the pin data will be described below with reference to the drawings.

FIG. 1 is a schematic view of a portion where control related to the present invention flows. In the following, as an embodiment of the present invention, the evacuation area will be described by using the spare area of the magnetic disk device, but it is not necessarily limited to the spare area, and a magnetic disk device with a cache including a spare disk is also used. is there.

The magnetic disk subsystem with cache memory comprises a central processing unit 110 and a storage control unit 120.
And a storage device 130. Storage controller unit 12
0 is requested by the central processing unit of the host device to transfer data, the control unit 121 temporarily stores the data in the cache memory 122, and reports the completion of data transfer to the host device. After that, the data is transferred from the cache memory 122 to the storage device 130. However, in this state, due to a failure of the storage device 130, it may not be possible to write data to the target access position. At this time, the cache memory 12
The untransferable data left in 2 is called pin data 124.

FIG. 2 shows a cache memory pin data management table 125 for managing the pin data of the cache memory 122. The cache memory pin data management table 125 is in the shared memory 123, the cache memory pin data position information 203 of the pin data 124 of the cache memory 122, and the cache memory pin management information flag 2 indicating that there is a pin in the cache memory.
01 and a cache memory pin data counter 202 indicating the number of pin data in the cache memory. The cache memory pin data position information 203 includes
It has a device number 204, a cylinder number 205, a track number 206, and a data presence flag 207.

FIG. 3 shows a pin data management table 133 for storing the pin data information at the time of planned stop when the pin data 124 exists in the cache memory 122.
Pin data 1 in the cache memory 122 in the table
If 24 exists, pin management information flag 301 with pin data indicating that there is at least one pin data 124
There are the pin data number 302 indicating the number of the saved pin data 23, and the pin data position information which is the position information of the pin data 124. Further, the information of the pin data 124 includes a device number 304, a cylinder number 305, a track number 306, and a track data valid flag 307 indicating whether the pin data stored in the spare area is valid or invalid.

4 and 5 show the flow of the present invention.

FIG. 4 shows the procedure for saving the pin data. When a data transfer request is issued from the normal cache memory 122 to the storage device 130, it is determined whether the data can be transferred to the storage device 130. At this time, if the data cannot be transferred from the cache memory due to the failure of the storage device 130, the data becomes the pin data 124, and the pin management information flag 201, the cache memory pin data counter 202 in the cache memory pin data management table 125, Necessary information is recorded in the cache memory pin data position information 203.

At the time of the planned stop, if the pin management information flag 201 in the cache memory pin data management table 125 is valid (indicates 1 as a value) to confirm that the pin data 124 exists, then the pin data 12
4 is a pin data management table 133 in the storage device 130.
To memorize. If the data presence flag 307 of the pin data 124 to be saved next is valid, the spare area 132 in which the pin data 124 can be stored is confirmed and saved, and the track data valid flag 30 in the pin data management table 133 is saved.
Enable 7. On the contrary, when the data presence flag 207 is invalid, or the spare area 132 in which the pin data 124 can be stored
If there is not, the track data valid flag 307 is invalidated. Then, when the position information of all the pin data 124 is written in the pin data management table 133,
The number of pin data in the cache memory pin data counter 202 is written in the pin data number 302 in the pin data management table 133, and the pin data management table 13 is written.
3 pin management information flag 301 is enabled.

On the contrary, if the pin management information flag 201 is invalid (indicates 0 as a value), the pin management information flag 301 in the pin data management table 133 is invalidated so that the pin data 124 does not exist. Record. Then, when the information of all the pin data 124 is stored in the storage device 130, the power is turned off.

FIG. 5 shows the procedure of pin data recovery processing. When the system is restarted, it is first checked whether the battery of the cache-backed-up cache memory 122 has run out. If the battery has not run out, the pin management information flag 301 in the pin data management table 24 is invalidated, and the pin data is recovered. The process ends.

If the battery is dead and the pin management information flag 301 of the pin data management table 133 is invalid, the cache memory pin data management table 12
The pin management information flag 201 in 5 is invalidated, and the pin data recovery process ends.

If the battery is dead and the pin management information flag 301 of the pin data management table 133 is valid, the following processing is executed.

First, the pin data position information 303 of the pin data 124 in the pin data management table 133 is written in the cache memory pin data position information 203. If the track data valid flag 307 is valid, the pin data 133 saved in the spare area 132 of the magnetic disk device 130 is restored in the cache memory 122. Then, the data presence flag 207 of the cache memory pin data management table 125 is validated. On the contrary, if the track data valid flag 307 is invalid, the data presence flag 20 of the cache memory pin data management table 125 is set.
Disable 7.

Then, all the saved pin data 12
The information of 4 is the cache memory pin data management table 1
After being restored to 25, the pin management information flag 301 of the pin data management table 133 is invalidated.

By the above steps, even if the pin data 124 is obtained, the data can be preserved in the same way as other data.

[0029]

According to the present invention, in a magnetic disk subsystem with a cache memory, when there is pin data that cannot be written from the cache memory to the magnetic disk device due to a failure of the magnetic disk device, and when there is a planned stop, the data in the cache memory is The track including the pin data can be saved as much as possible, and conversely, at the time of restart, the pin data can be restored in the cache memory like other data by using the information in the pin data management table. As a result, in the above information system, if the data from the host controller can be updated in the cache memory, the possibility of data loss becomes very small.

[Brief description of drawings]

FIG. 1 is a schematic diagram of a magnetic disk subsystem including parts relevant to the present invention.

FIG. 2 is a diagram showing a table for managing data in the cache memory that could not be written from the cache memory to the magnetic disk device due to a failure of the magnetic disk device.

FIG. 3 is a diagram showing a table in which management information of data that cannot be written from the cache memory to the magnetic disk device due to a failure of the magnetic disk device at the time of planned stop is stored.

FIG. 4 is a diagram showing a flowchart of processing for saving and restoring data that could not be written in the magnetic disk device from the cache memory due to a failure of the magnetic disk device.

FIG. 5 is a diagram showing a flowchart of processing for saving and restoring data that could not be written in the magnetic disk device from the cache memory due to a failure of the magnetic disk device.

[Explanation of symbols]

110 .... Central processing unit 120 .... Storage control unit 121 .... Control unit 122 .... Cache memory 123 .... Shared memory 124 .... Pin data 125 .... Cache Memory pin data management table
130 .... Storage device 131 .... Storage area in storage device 132 .... Reserve area in storage device 133 .... Pin data management table 134 .... Saved pin data

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Isamu Kurokawa 2880, Kozu, Odawara-shi, Kanagawa Hitachi Storage Systems Division (72) Inventor, Kuji Takeuchi 2880, Kozu, Odawara, Kanagawa Hitachi Storage Co., Ltd. System Division

Claims (2)

[Claims] 1. A cache memory, a magnetic disk device, a storage control device for controlling the cache memory and the magnetic disk device, and data transfer between the cache memory, the magnetic disk device and the storage control device. A method for controlling a magnetic disk subsystem with a cache memory, comprising: a data transfer means for controlling a magnetic disk subsystem from a central processing unit, which is a host controller of the magnetic disk subsystem with a cache memory, using the control of the data transfer means. When the write data is temporarily stored in the cache memory and then the data is stored in a target access position in the magnetic disk device, the cache memory cannot be written in the target access position due to a failure of the magnetic disk device. For the data remaining in
The unwritable data is saved in the spare area of the magnetic disk device to guarantee the data at the time of the planned stop, and when the magnetic disk device with the cache memory is restarted, the write failure stored in the spare area is lost. A method of controlling a magnetic disk subsystem for recovering possible data in the cache memory. 2. The method according to claim 1, wherein data is saved to a spare disk in place of the spare area of the magnetic disk device.