JP2000267936A - Disk sub-system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce the number of writing requests in a disk in a disk sub-system having no function for non-volatilizing a cache memory. SOLUTION: Since a write-through type disk sub-system loaded with a cache memory 24 and connected to a host system 10 is constituted so that a host processor 12 informs the sub-system 20 that the processor 12 itself sends an end report or a passing point report to an operating system 11 and the sub-system 20 writes updated data in a magnetic disk device 33 at the time of receiving the report. Thereby partial post writing operation is executed without using a non-volatile memory and a battery backup device, the state of the updated data is stored in an access management table 32 included in the device 33, and even when a fault is generated during the execution of processing, the table 32 is referred to and an access range is inhibited.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk subsystem having a cache memory, and more particularly, to a disk subsystem having no nonvolatile memory or a power supply backup device, which can normally operate even when a failure occurs before writing data to a magnetic disk device. A disk subsystem that can be restored to

[0002]

2. Description of the Related Art In general, a disk subsystem for controlling the reading and writing of data from and to a plurality of magnetic disk devices is capable of high-speed reading of data by storing previously read or frequently read data in a cache memory. High-speed writing of update data by writing an update report to the host when writing to the cache memory is completed in response to a data update instruction from a higher-level operating system, and then asynchronously writing the update data to the magnetic disk This makes it possible to immediately return a response to a write request from the operating system, thereby dramatically improving write performance.

In the conventional method of reporting the completion to the host when the writing to the cache memory is completed, the writing to the disk is performed asynchronously. And the updated data exists only in the cache memory. For this reason, when data in the cache memory is erased due to a failure such as a power failure, there is a possibility that the data reported to a higher order and the data actually stored in the magnetic disk do not match. In order to prevent the above-mentioned inconveniences, the disk subsystem according to the prior art uses a non-volatile memory which does not require a power supply for holding data in a cache memory, or a power supply backup device or the like for maintaining a system power supply. Needed.

Incidentally, in a disk subsystem having no nonvolatile memory or power backup device, writing to a disk is performed in synchronization with a data update instruction (data update command) from the operating system. Only the performance was improved.

[0005]

In the disk subsystem having a nonvolatile memory and a power supply backup device according to the prior art, although reliability can be ensured, there is a problem that the cost of the system is increased.

In a disk subsystem having no nonvolatile memory or power supply backup device according to the prior art, when data in a cache memory is erased due to a failure such as a power failure, the data reported to a higher order and the magnetic disk are actually stored. There was a problem that the data did not match.

SUMMARY OF THE INVENTION An object of the present invention is to eliminate the above-mentioned disadvantages of the prior art. In a disk subsystem having no nonvolatile memory or power supply backup device, lost data or invalid data transfer in a cache memory due to a failure. To provide an inexpensive disk subsystem with improved write performance.

[0008]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention is connected between a host system and a magnetic disk drive, has a cache memory, and transfers data between the host system and the magnetic disk drive. In the controlling disk subsystem, a report indicating that the host system reports an end report of the input / output command or a passing point indicating the passing point of the input / output command, based on the input / output command from the host system to the disk subsystem. Means for notifying the notification, updating the data in the cache memory until the report notification is received, and writing the updated data in the cache memory to the magnetic disk device when the report notification is received. The feature of.

Further, according to the present invention, in the disk subsystem having the above-mentioned features, a means for storing an update data area and a means for notifying the disk subsystem of the report notification from the host system are provided, and the data update processing to the host system is completed. After the report, the area corresponding to the disk unupdated data in the cache memory is stored in the magnetic disk device as an access prohibition target area, and the cache memory is written before writing the unupdated data in the cache memory to the disk by the report notification from the host system. A second feature is that when a failure or the like occurs, access to the access prohibition target area is suppressed by referring to the access prohibition target area stored in the magnetic disk device.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a disk subsystem according to the present invention will be described below in detail with reference to the drawings. FIG. 1 is a diagram showing a block configuration of a computer system including a disk subsystem according to the present embodiment.

This computer system writes and reads data between the host system 10, a disk device group 30 including a plurality of magnetic disk devices, and the disk device group 30 in accordance with instructions from the host system 10. And a disk subsystem 20 to be controlled.

The host system 10 includes an operating system 11 including an OS, a control program, and the like;
It comprises an MS (main storage) 13 for storing application programs and data for writing and other operations on a magnetic disk, and a host processor 12 for performing arithmetic processing.

The disk subsystem 20 includes:
A channel control unit 21 that performs interface control with the host system 10 and a CCW (Channel Commander)
ndWord: a CCW operation control unit 22 for analyzing data by operating a channel command word for a channel constituting a channel program, and a disk control unit 23 for controlling data read / write of the disk device group 30.
And a volatile cache memory 24.

The disk unit group 30 is composed of a plurality of magnetic disk units 33. Inside the disk unit 33, a user area 31 for storing data from the operating system 11 and an access table for storing an access permission / prohibition control flag are stored. 32.

Further, the host system 10 and the disk subsystem 20 connect the host processor 12 and the channel control unit 21 with an operation control signal 15, an interrupt occurrence notification signal 16 and a data bus 17, and The signal 16 notifies the disk subsystem 20 of the end report and the passing point report to the operating system, and the data bus 17 is for performing data transfer with the MS 13. The disk subsystem 20 and the disk device group 30 are connected to the SCSI bus 3
4 are connected.

The normal operation of the computer system thus configured operates according to the following procedure. (1) The host system 10 creates a CCW (Channel Command Word) sequence 14 on the MS 13 via the host processor 12 when making a request to access a disk by the operating system 10 and
2 is instructed to start disk access. (2) Upon receiving this instruction, the host processor 12 notifies the channel control unit 21 in the disk subsystem 20 of the disk access start instruction through the operation control signal 15 after receiving the disk access start instruction from the operating system 10. .

(3) The channel control unit 21 receiving this notification
Receives the first CCW in the CCW sequence 14 via the data bus 17 after receiving the notification, and delivers it to the CCW execution unit 22. (4) The CCW execution unit 22 determines whether the CCW is a read / write command for the disk. (5) When the CCW is a read command, instruct the disk control unit 23 to read user data,
The disk control unit 23 instructs a read range to the disk device group 30, and writes the transfer data from the disk device group 30 to the cache memory 24. The disk control unit 23 reports the read completion to the CCW execution unit 22, writes the read data from the channel control unit 22 via the data bus 17 to the MS 13, and ends the CCW.

(6) If the CCW is a write command, the channel control unit 21
The data on S13 is transferred to the cache memory 24.
Thereafter, the CCW execution unit 22 instructs the disk control unit to set a status indicating “data access prohibited” in the access management table 32 in the disk device group 30 corresponding to the data update range, and after the status setting response, Control unit 2
1 to report the completion of the CCW. (7) Thereafter, when a plurality of write commands are issued for the same range, the access management table 32
Has been completed, the CCW is completed only by updating the data in the cache memory 24.

(8) When the host processor 12 issues an interrupt to the operating system 11 while the CCW sequence 14 is being executed, the host processor 12 sends the interrupt occurrence notification signal 16 so that the disk subsystem can recognize the issuance of the interrupt. And notifies the channel control 21 of the occurrence of the interrupt. The channel control unit 21 further reports the above-described interrupt occurrence notification to the CCW execution unit 22, and the CCW execution unit 22 sends the cache memory 24 to the disk control unit 23.
Is written to the disk device group 30.

(9) After the data update to the disk device group 30 is completed, the CCW execution process 22 instructs the disk control unit 23 to write “data” to the access management table 32 in the disk device group 30 corresponding to the update data. Access prohibited "
Instruct the cancellation of the state.

The operation described above is a normal operation in the present computer system. In the present embodiment, when a power failure occurs during the operation of the write command,
The following processing is performed, and will be described below with reference to the processing flows of FIGS.

The disk subsystem 2 according to the present embodiment
0 indicates that the CCW issued in the above-described process is a write command for updating data (step 30 in FIG. 3).
1), from the channel control unit 21 via the data bus 17,
The data on the MS 13 is transferred to the cache memory 24,
It is determined whether a data update command has been issued to the disk control unit by the CCW execution unit 22 before (step 302). If not, access management in the disk device group 30 corresponding to the data update range is determined. A state indicating "data access prohibition" is set in the table 32 (step 303), and then an end report or a pass point report from the host system is determined (step 304). 24 is written to the magnetic disk device (step 3).
05) Then, the "data access prohibited" state of the access management table is released (step 306).

In the present embodiment, step 3
04, the end report from the host system or the report of the passing point is determined, and when these reports are made, a path for writing the update data in the cache memory 24 to the magnetic disk device is provided. A path for notifying the disk subsystem 20 of the report of
Until the disk subsystem 20 receives the report notification, the data is updated on the cache memory 24, and at the time of the report notification, the updated data on the cache memory is written to the disk.

Therefore, if a power failure occurs before the conventional disk subsystem writes the update data to the magnetic disk device, the update data not reflected in the disk device group 30 exists in the cache memory 24 because the update data is not reflected in the disk device group 30. While the update data has been lost and the system has gone down with the status indicating “data access prohibited” set in the access management table 32 in the disk device group 30, the disk subsystem according to the present embodiment The operation is performed as shown in FIG.

That is, the disk subsystem 20 according to the present embodiment, when a power failure occurs (step 40).
1), the disk subsystem 20 is restarted (step 402), and the access management table 32 in the magnetic disk device 33 is referred to and the access management table 32 is checked (step 404). This access management table 32 corresponds to step 3 in FIG.
Since the data has already been written on the disk at 05, the data is retained through the power failure or recovery, and the access management table 3 in the area where the data was lost when the power failure occurred
2, the state indicating "data access prohibition" remains set.

For this reason, in the present system, after restarting, the operating system 11 refers to the access management table 32 in step 403, and the management table 32
Is determined in the “data access prohibited” state stored in the storage area, and if an attempt is made to access the prohibited area (step 405), the CCW execution processing 22
Sends a report to the operating system 11 indicating that access to the above-mentioned range is prohibited, so that the user area corresponding to the range in which the access management table 32 whose data has been lost is in the “data access prohibited” state. Of the old data (data before the update) can be prevented from being transferred to the operating system 11. If an attempt is made to access an area for which access is not prohibited (step 404), the CCW
An execution process 22 permits the operating system 11 to permit normal access.

As described above, according to the present embodiment, in the disk subsystem having no nonvolatile memory or power supply backup device, the host processor does not report the end to the operating system or report the passing point.
By notifying the disk subsystem and writing the update data to the disk when the disk subsystem receives the notification, a plurality of data update requests are collectively issued to the disk device and a write request is issued to the disk device. Overhead can be reduced and performance can be improved.

When the end report or the passing point is reported to the host system, the data update range is set to the access prohibition area when the disk subsystem is notified, and the update data is lost due to a power failure or the like. In addition, by prohibiting access to the range to be updated, the old data in the user area corresponding to the range in which the access management table from which data has been lost is in the “data access prohibited” state (before updating). Data) can be prevented from being transferred to the operating system 11.

It should be noted that the disk subsystem according to the present invention can be represented as the following embodiments. First Embodiment In a disk subsystem equipped with a cache memory connected to a host system and having no function of making the cache memory non-volatile, an input / output command from the host system to the disk subsystem causes the host system Has a means for notifying the disk subsystem of a report of completion of an I / O instruction or a report of a passing point indicating a passing point of an I / O instruction. A disk subsystem for updating data and writing updated data in a cache memory to a disk when a report is issued.

<Second Embodiment> The disk subsystem according to the first embodiment has means for storing an area for update data and means for notifying the disk subsystem of a report notification from the host system. After the update processing completion report, the area corresponding to the disk unupdated data in the cache memory is held as the access prohibition target area, and the cache memory failure occurs before the host system writes the unupdated data in the cache memory to the disk by the report notification. A disk subsystem that suppresses access to the access-prohibited area by means for storing the area of the update data when such a situation occurs.

[0031]

As described above, according to the present invention, based on the input / output command from the host system to the disk subsystem, the host system reports the completion of the input / output command or the passing point indicating the passing point of the input / output command. Providing means for notifying a report notification indicating that a report is provided, and updating data on the cache memory until the report notification is provided,
With the configuration in which the update data in the cache memory is written to the magnetic disk device when the report notification is received, the write performance to the disk can be improved without having a failure backup device in the nonvolatile memory or the battery device. Can be planned. Further, when the update data from the operating system is lost due to a failure, it is possible to perform normal recovery by setting the update request range to the access prohibited state.

[Brief description of the drawings]

FIG. 1 is an exemplary diagram showing a block configuration of a computer system including a disk subsystem, a host system, and a disk device group according to an embodiment of the present invention.

FIG. 2 is a view showing a configuration of access management information in a disk device according to the embodiment.

FIG. 3 is an exemplary diagram showing transition of access management information setting according to the embodiment;

FIG. 4 is an exemplary diagram showing transition of access prohibition state management according to the embodiment;

[Explanation of symbols]

10: host system, 11: operating system, 12: host processor, 13: MS (main storage), 14: CCW (channel command word)
Column, 15: operation control signal, 16: interrupt occurrence notification signal, 17: data bus, 20: disk subsystem,
21: channel control unit, 22: CCW (channel command word) execution unit, 23: disk control unit, 24: cache memory, 30: disk device, 31: user area, 32: access management table, 33: disk unit, 34 : SCSI bus.

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) G11B 20/10 G11B 20/10 D

Claims (2)

[Claims] 1. A disk subsystem connected between a host system and a magnetic disk device, mounted with a cache memory, and controls data transfer between the host system and the magnetic disk device. Based on the input / output instruction of
A means for notifying a report notification indicating that the host system reports an end report of the input / output instruction or a passing point indicating the passing point of the input / output instruction is provided, and data is stored in the cache memory until the report notification is received. A disk subsystem for updating and writing update data in a cache memory to a magnetic disk device when the report notification is received. 2. A means for storing an update data area and a means for notifying the disk subsystem of the report notification from the host system, and after reporting the end of data update processing to the host system, the disk unupdated data in the cache memory is provided. In the case where a cache memory failure or the like occurs before writing the unupdated data in the cache memory to the disk by a report notification from the host system, the area corresponding to
2. The disk subsystem according to claim 1, wherein access to the access prohibition target area is suppressed by referring to the access prohibition target area stored in the magnetic disk device.