JPH0232418A - Peripheral controller for magnetic disk memory - Google Patents

Abstract

PURPOSE:To shorten the response time of a DK by reading from a magnetic disk memory (DK) in which data transfer can be first executed when a processing request from a CPU is read and writing all to the DK at the time of writing. CONSTITUTION:In the access of DK, a CPU 102 prepares and stores the parameter necessary to a reading (R) action to an MEM 103 at a high speed mode, an EXECUTE operation (EXO) is executed to in IOP#0 (peripheral control part 104) and fetches the R parameter of the MEM 103. When the contents of the R parameter are judged to be the reading, first, the data are read from a DKM 108, in which the data transfer can be executed, and the data transfer is executed to the data storing area of the MEM 103. In a writing action, DK108,...,109 and DK110,...,111 are simultaneously activated, the writing action is executed and when the writing processing is completed, the completion of the writing processing is reported to the CPU 102.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a peripheral control device for a magnetic disk storage device, and more particularly to a fault-tolerant technology for realizing fault tolerance of an information processing device.

Specifically, the present invention relates to an access control method from a central processing unit to multiplexed magnetic disk storage devices in a multiplexed configuration of magnetic disk storage devices employed in fault-tolerant technology.

(Prior art) Conventionally, in order to achieve high reliability of magnetic disk storage devices, a duplex configuration of magnetic disk storage devices has been adopted in which the same data is held in two magnetic disk storage devices at the same time. .

In this case, all access control to the duplicated magnetic disk storage devices is managed and executed by a control program on the central processing unit. In other words, whether or not the magnetic disk storage device is duplicated depends on whether the magnetic disk storage device
I was not aware of the magnetic disk controller.

Therefore, in response to an access request to a duplicated magnetic disk storage device, the central processing unit activates each of the duplicated magnetic disk storage devices, and upon receiving operation completion reports from the two magnetic disk storage devices, It is necessary to complete processing for terminating access to the magnetic disk storage device.

In addition, since the magnetic disk storage devices are duplicated, the head positions of the two magnetic disk storage devices
The following functions are adopted as a magnetic disk storage device access control method that makes it possible to shorten the magnetic disk storage device access time by utilizing the positional relationship on the magnetic disk storage device where the target data is stored. ing.

(1) RE A D! The speed-up function in the 11th work is to select a magnetic disk storage device with a smaller 5EEK time, and perform RE on this magnetic disk storage device.
By instructing the AD operation, the 5EEK time can be shortened.

(2) Normally, in WRITE\JJ work, it is necessary to confirm that the data has been reliably written to a dual magnetic disk storage device. Therefore, after waiting for the completion of the WRITE operation in both of the two duplexed magnetic disk storage devices, the central processing unit performs the WRI
It is determined that the TE operation is complete.

[Problem to be solved by the invention]

In the case of (1) above, it is necessary to constantly monitor the current track address of the magnetic disk storage device, and if the managed head position shifts due to access from another system, the 5EEK time will be reduced. There may be cases where this is not possible.

Further, the access time of magnetic disk storage devices has not been shortened by focusing on search time.

In the case of (2) above, the response time is longer in the WRITE operation than in the WRITE operation for a normal non-duplexed magnetic disk storage device.

However, depending on the reliability conditions required by the information processing system and the control of the response time allowed for the magnetic disk storage device, the WRI of the dual magnetic disk storage device
Even in TE operation, high-speed response may be required.

The object of the present invention is to shorten the response time of a magnetic disk storage device to a request for access from a central processing unit to a multiplexed magnetic disk storage device.

[Means to solve the problem]

A peripheral control device for a magnetic disk storage device according to the present invention receives a processing request from a central processing unit for a multiplexed magnetic disk storage device, and transmits the processing request to an REA.
In the case of processing D, data is stored only from the magnetic disk storage device that is set as a multiplexing device to the main memory section specified by READL/, and the data is stored in the main memory section specified by READL/. The magnetic disk storage device is equipped with means for terminating the READ operation midway.

In addition, if the processing request is WRITE, WRITE is sent to all magnetic disk storage devices specified as multiplexing devices.
It is provided with a means for instructing an ITE operation and, when the WRITE operation is completed in any one of the magnetic disk storage devices, to report the completion of the WRITE process to the central processing unit.

[Effect]

In this invention, the processing request from the central processing unit is
In the case of EAD processing, data is read only from the first magnetic disk storage device that can start data transfer.
, and the remaining read operations of the magnetic disk storage device are terminated midway.

In addition, if the processing request from the central processing unit is WRITE processing, WRI is sent to all magnetic disk storage devices.
A TE operation is instructed, and when the WRITE operation is completed in any one magnetic disk storage device, the completion of the WRITE process is reported to the central processing unit.

〔Example〕

Hereinafter, one embodiment of the present invention will be described with reference to the drawings.

FIG. 1 shows a configuration example of an information processing system 101 according to the present invention, in which the information processing system 101 includes a central processing unit (CPTR) 102, a main memory unit (MEM) 103,
A plurality of peripheral device control units (Ipo) 104, 105 and a plurality of magnetic disk storage devices (])K) 108, 109
, 110, 111.

Peripheral device control units 104 and 105 receive data access requests from the CPU 102 to peripheral devices and input/output devices, manage multiple peripheral device controllers, and control input/output operations. A magnetic disk storage device controller 10 that controls devices and input/output devices.
It shows that it is composed of 7.

In the information processing system 101 shown in FIG. 1, the peripheral device control unit is an example of a magnetic disk storage controller (DKC) 107 that controls a magnetic disk storage device.

Peripheral device control unit 104 is a magnetic disk storage device 108
．． 109 shows a configuration in which the peripheral device control unit 105 controls the magnetic disk storage devices 110 and 111.

FIGS. 2(a) and 2(b) show information between the central processing unit 102 and the peripheral device controllers 104 and 105, or the peripheral device controllers 104 and 105, which are necessary for explaining the operation of the embodiment shown in FIG.
105 shows an operation function issued when executing communication between 105 and 105.

Note that the information processing system 10 including the central processing unit 102, main memory unit 103, peripheral device control units 104 and 105, etc.
The constituent requirements of 1 are collectively called a module.

■The EXECUTE operator 201 executes a command from the central processing unit 102 or other peripheral device controllers 104, 105 to the peripheral device controllers 104, 105 to which the magnetic disk storage devices 108 to 111 to be processed are connected.
This is an operation for transferring commands for instructing human output control of the magnetic disk storage devices 108 to 111.

■COMPLETE operation 202 is EXEC
This is an operation for reporting the completion of the operation of the magnetic disk storage devices 108 to 111 started by the UTE operation 201 from the peripheral device control units 104 and 105 to the processing request module. COMPLETE operation 202 is performed on magnetic disk storage devices 108 to 1.
11, the peripheral device control unit 104,1
It will be issued if it is deemed necessary in the report in 05.

Also, in the COMPLETE operation 202, it is simultaneously reported whether the processing of the magnetic disk storage devices 108 to 111 has ended normally or abnormally.

■5END 5TATUS operation 203 is
This operation requests a status report of the specified peripheral device control units 104, 105 and magnetic disk storage devices 108 to 111.

Also, 5END 5TATUS operation 203
The module that receives this will report the abnormal termination as a termination report for the specified device, and if it holds detailed information regarding the abnormal termination, it will transfer the held detailed information.

■The HALT operation 204 performs the specified peripheral devices and input/output devices (magnetic disk storage devices 108 to 111).
If an operation is currently being executed, this operation is for instructing the suspension of the operation (suspend state) or forced termination of the process that is currently being executed.

(2) The RESUME operation 2o5 is an operation for instructing the restart of the input/output operation of a peripheral device whose operation has been interrupted by the HALT operation 204 and is in the process of being suspended.

Does nothing if the specified device is not suspended.

Each operation is in the format 20 when communicating the operation.
6, the information is transferred. Own module ID210
holds the identifier (ID) assigned to the module that issued this operation, and the other module ID2
11 holds an identifier (ID) given to the module that receives this operation. The own module ID 210 and the destination module TD 211 are determined by hardware, and the operation information to be subsequently transferred is stored inside the destination module.

The operation code 212 holds the code assigned to each operation, and the device identifier 213 indicates the peripheral device that is the target of the operation specified by this operation.
Holds the device identifier (ID) of the input/output device.

Additional information 214 holds individual information necessary for executing each operation. For example, EXECUT
In the E operation 201, the address of the main memory section 103 of the parameter that defines the input/output operation is set to COMPLE.
TE operation Holds a code indicating the status of operation 202 completion.

FIGS. 3(a), 3(b), and 3(c) show command functions transferred in EXECUTE operation 201 necessary to control the magnetic disk storage device. In this example,
The magnetic disk storage device is duplicated, and a command mechanism for accessing the duplicated device is shown.

■DEFINE EXTENT near7)' 301 is a command that assigns a logical device identifier 213 to the physical peripheral device specified by the subchannel address.

At the same time, the DEFINE EXTENT command 301
is specified in parameter format 310, and parameter format 3
10 and the number of blocks, the device identifier 213 specifies a physical device to be accessed or a specific area of the physical device, and provides a function to set it in the peripheral device control units 104 and 105.

■ASSING MIRRORDK command 302 is a command for setting two magnetic disk storage devices in a duplex relationship with each other.

The ASSINGN MIRRORDK command 302 is issued to each of the two device identifiers 213 with which a duplex relationship is set.

This command uses the parameter format 311, and passes through the duplicated partner device using the device identifier 313 and module ID 315. The operation mode 314 is REA for the duplicated magnetic disk storage devices 108 to 111.
Specifies detailed operations when requesting D/WRITE operations.

The READ mode 320 simultaneously issues a READ request to two duplicated magnetic disk storage devices, and reads data from the magnetic disk storage device for which 5EEK and 5EARCH operations have been completed first (high-speed operation mode), or
RE only from one device specified by the EAD device 321
Specify whether to perform AD operation (single mode). READ
The device 321 specifies a magnetic disk storage device that performs a READ operation when the single mode is specified.

The WRITE mode 322 performs a write operation on the duplicated magnetic disk storage devices 108 to 111, and after the WRITE operation for one storage unit is completed, the WRITE operation for the two magnetic disk storage devices is started (serial execution), or
Specifies whether to instruct and execute WRITE operations on two magnetic disk storage devices at the same time (parallel operation). WRITE
The reporting period 323 is the WRITE of the magnetic disk storage device.
Specifies when to report the completion of the operation to the access request module. The conditions that can be set are: do not report WRITE completion, report WRITE completion when the WRITE operation is completed first, and report WRITE completion when all WRITE operations of the two magnetic disk storage devices are completed.
TE completion is reported.

■CLEARMIRRORD K command 303 is issued to each of the two magnetic disk storage devices in a duplex relationship to clear the previously set duplex relationship.

This command uses the parameter format 311, and uses the device identifier 313 and module ID 315 to notify the partner device in a dual relationship. Operating mode 314 is ignored.

■READ command 304 is a command that reads data from the specified area of the magnetic disk storage devices 108 to 111 specified according to the contents of the parameters transferred by this command, and stores it in the specified main memory section 103. be.

■The WRITE command 305 writes data to the specified main memory section 10 according to the contents of the parameters transferred by this command.
Magnetic disk storage devices 108 to 111 specified from 3
This is a command to store data in the recording area of .

■The LOCATE command 306 is a command that moves the head position of the magnetic disk storage devices 108 to 111 specified according to the contents of the parameters transferred by this command to the track where the specified block number exists. be.

RE A D / W R I T E / L OCA
Parameter format 312 is used in T E Conv F. A memory address 316 and a data length 317 are specified as the data area of the main memory unit 103, and a first block number 318 in which the data to be processed exists is specified as the recording area of the magnetic disk storage devices 108-111.

LOCATE command 306 sets memory address 316
and the contents of data length 317 are ignored.

■5ENSE command 307 causes peripheral device control units 104, 105, magnetic disk storage device 108 to be stored in main memory unit 103 specified by the parameters transferred by this command.
This is a command requesting to report the status of the magnetic disk storage devices 108 to 111, or abnormal conditions of the magnetic disk storage devices 108 to 111 detected during operation.

The parameter format 312 is used in the 5ENSE command 307, and the contents of the first block number 318, which specifies the memory address 316 and data length 317 as the data area of the main memory unit 103, are ignored.

FIG. 4 shows duplicated magnetic disk storage devices 108 to 108.
Peripheral device control units 104 and 10 necessary for explaining the processing flow of access from CPU 1o2 to 111
5 shows the internal configuration of the common section 106 of No. 5.

The common unit 106 is a microprocessor (MPU) that performs control functions necessary for internal operations such as managing the state of the peripheral device control units 104 and 105, analyzing and executing operations, fetching and executing commands, and creating and reporting completion statuses.
unit 401 , a local memory unit (LM) 402 that holds a memory area necessary for the operation of the common unit 106 , a buffer area for data transfer, configuration control information regarding a duplicated partner device, etc., and a central processing unit 102 . ．． A system bus interface section 403 that controls interface functions with the main memory section 103 or other peripheral device control sections 104 and 105, and a magnetic disk storage device controller 1.
07 to the main memory section 103, and data transfer operation between the microprocessor section 401 and the main memory section 103.
04, and the central treatment device 102 or peripheral device control unit 1
04, 105 and the peripheral device control unit 104, 105 in the interprocessor communication, and recognizes the operation for its own modulation, receives the transferred information, holds it internally, and sends a microprocessor to confirm that the operation has been received. Processor section 4
An operation communication control unit 405 that has a function of reporting using an interrupt to 01, and a common unit 1 that displays an abnormal state.
The configuration includes a debug control unit 406 that provides debug functions for the firmware of 06, and each component is connected by an internal bus of the common unit 106.

When an operation is issued from the central processing unit 102 to the peripheral device control unit 104, the system bus interface unit 403 sequentially transfers the transferred operation information to the operation communication control unit 405.

The operation communication control unit 405 checks the partner module ID 211 of the operation information and determines whether it matches the module ID assigned to itself.

As a result, if it is determined that an operation has been issued to itself, the operation code 212, device identifier 213, and additional information 214 subsequently transferred from the central processing unit 102 are internally captured and held. The operation communication control unit 405 can hold a plurality of pieces of operation information.

When the operation communication control unit 405 receives an operation from the conversion module, the operation communication control unit 405 transmits the operation to the microprocessor unit 4.
01 by interrupt. Upon receiving the interrupt, the microprocessor unit 401 directly accesses the internal register of the operation communication control unit 405 to determine and execute the specified operation.

When parameters are required to execute an operation, the parameters prepared in the main memory unit 103 are transferred to the internal local memory unit 4 via the data transfer control unit 404.
02 and instructs input/output operations according to the contents of this parameter.

Peripheral device control in this embodiment requires (1) assignment of device identifiers to peripheral devices and input/output devices and setting of duplex devices, and (2) execution of normal input/output operations.

Each process will be explained below.

[1] Assignment of device identifiers to peripheral devices and duplication device settings FIGS. 5(a) and 5(b) show EXECUTE operation 201 when allocating the magnetic disk storage device 108 and the magnetic disk storage device 110 as a duplex device pair.
shows. Each magnetic disk storage device 108, 110 is given a device identifier ``DKOOI''.

First, the entire magnetic disk storage device 108 and magnetic disk storage device 110 are designated as the device identifier 213 by °DKO.
Issue an operation to allocate OI'. The central processing unit 102 sends operation information 501 to the peripheral device control unit 104 as the content of the operation.
and parameter information 510 to the peripheral device control unit 105 as operation information 502 and parameter information 511.
Issue.

In this operation, the magnetic disk storage device 108 is used as a subchannel number (address unique to the physical device).
It is assumed that 1234' is set and consists of 320000 blocks. Similarly, the magnetic disk storage device 110 has a subchannel number (address unique to the physical device) set to 5678°.
It is assumed that it consists of 320,000 blocks.

Each peripheral device control unit 104, 105 has a command code (7
) By recognizing the DEFINE EXTENT command, a table is created and held internally that holds the device identifier, subchannel address, and area of the physical device that can be accessed by the device identifier.

As a result, by providing the device identifier 'DKOOI' to the peripheral device control units 104 and 105, it becomes possible to access the target magnetic disk storage device.

Next, an operation for specifying a duplex device is issued. As for the contents of the operation, operation information 503 and parameter information 512 are issued to the peripheral device control unit 104, and operation information 504 and parameter information 513 are issued to the peripheral device control unit 105.

By issuing an operation to both the peripheral device control unit 104 and the peripheral device control unit 105, each peripheral device control unit can confirm that the magnetic disk storage device 108 and the magnetic disk storage device 110 have been assigned as bare of the duplex device. 104 and 105.

In this embodiment, the central processing unit 102 issues each operation four times independently, but by specifying the command chain in the flag part of the parameter, it can be issued once (total) to each peripheral device control unit 104, 105. It is also possible to set the device identifier and duplex device by issuing the operation twice.

The central processing unit 102 simply specifies the device identifier 'DKOO1' to the peripheral device control unit 104 and executes READ/W.
When a RITE processing request is issued, the peripheral device control unit 10
4 independently determines whether duplexing is set and the address of the magnetic disk storage device 108 to which the duplexing device is connected, and issues the same READ/WRITE processing request to the magnetic disk storage device 110 of the peripheral device control unit 105. As a result, access control of the dual configuration of the magnetic disk storage device is realized by hardware.

Conversely, the device identifier 'DK' is sent to the peripheral device control unit 105.
When a READ/WRITE processing request is issued specifying OOI°, the peripheral device control unit 105 issues a processing request to the peripheral device control unit 104, thereby realizing duplication of the magnetic disk storage device.

[2] Flow of READ/WRITE processing of magnetic disk storage device (1) Execution of RE A D operation Figures 6 and 7 show R when high-speed operation mode is specified.
The processing flow of EAD operation is shown.

In the normal operation mode, if a failure is detected in accessing the magnetic disk storage device, the peripheral device control units 104 and 105 automatically retry the READ operation using another duplicated magnetic disk storage device. This is the same as accessing a magnetic disk storage device for which duplexing is not set, except that it is performed in a fixed manner.

FIG. 6 shows that in the high-speed operation mode, the mask device (the peripheral device under the module that has received the activation of the magnetic disk storage device from the central processing unit 102) is accessed by the slave device (the activation of the magnetic disk storage device from the peripheral device control unit). This shows the flow of processing when the peripheral device under the module that received the request ends earlier than the peripheral device under the module that received the request.

The central processing unit 102 creates and stores READ parameters 601 in the main memory unit 103 with parameters necessary for the READ operation, specifies the device identifier 'DKOO1', and executes the IO
1 for P#0 (peripheral device control unit 1゜4). :EXEc
Issue UTE operation 602.

IOP#0 executes EXECUTE operation 602 and performs parameter fetch 603 of READ parameter 601 in main memory section 103. READ parameter 6
It is determined that the content of 01 is a READ command, and the magnetic disk storage device 108 is activated 604.

Subsequently, it is determined 605 whether or not the device identifier 'DKOOI' is a duplex device by checking the internal control table. If it is not set as a duplication device, a waiting process for a positioning completion report 610 is performed in order to execute a normal READ operation.

If it is set as a duplication device, IOP #1 is activated. First, parameters 606 necessary for READ operation are created, and an EXECUTE operation is issued 607 using the same procedure as the central processing unit 102.

When IOP #1 receives the EXECUTE operation 607 from IOP #O, it maintains that the magnetic disk storage device 110 under its own device has been activated as a slave device, and fetches the parameter 606. Determine that the content of parameter 606 is a READ command,
The magnetic disk storage device 110 is activated 609 .

In this operational example, when accessing the magnetic disk storage device, a positioning completion report 610 is first generated from the magnetic disk storage device 108. When IOP #0 receives the positioning completion report 610 from the magnetic disk storage device 108, it issues an instruction 611 to start data transfer to the magnetic disk storage device 10B and HALT requesting IOP #1 to immediately suspend processing of the magnetic disk storage device 110. Operation 612 is issued.

As a result, the magnetic disk storage device 108 reads data from the designated recording area of the magnetic disk storage device, and performs data transfer 613 to the data storage area of the main memory section 103 designated by the parameter 601.

When IOP#1 receives the HALT operation,
Instructs magnetic disk storage device 110 to end operation 6
14, and COMP notifies IOP #0 that access to the magnetic disk storage device has been interrupted midway through the operation completion report 615 from the magnetic disk storage device 110.
Send LETE operation 616.

After this, IOP#O reports a COMPLETE operation 618 to notify the central processing unit 102, which has requested access to the magnetic disk storage device, of normal completion based on the READ operation completion report 617 from the magnetic disk storage device 108, and Terminates disk storage access.

FIG. 7 shows the processing flow of a READ operation when the access by the slave device ends earlier than the access by the mask device.

In the process flow shown in FIG. 7, the process up to when IOP#1 starts up the magnetic disk storage device 110 609 is the same as that shown in FIG.

In this operational example, when accessing the magnetic disk storage device, first, a positioning completion report 701 is generated from the magnetic disk storage device 110.

When IOP #1 receives a positioning completion report 701 from the magnetic disk storage device 110, since it is a slave device, it issues a COMPLETE operation 702 to report positioning completion to IOP #O. After this, IOP#1 becomes lOP# within a certain period of time.
By receiving the RESUME operation 703 from o, data transfer 7.4 of the magnetic disk storage device 110 is continued.

IOP#O receives a COMPLETE operation 702 from IOP#1 that reported completion of positioning.
Instructs magnetic disk storage device 108 to end operation 7
06, and receives an operation completion report 707 from the magnetic disk storage device 108.

After this, IOP#1 receives a READ operation completion report 708 from the magnetic disk storage device 110, and receives a COMPLE message indicating that the access to the magnetic disk storage device 110 has been successfully completed.
Issue TE operation 709 to IOP#O. I
OP#O is a C0MPL that reports normal completion to the central processing unit 102 that has requested access to the magnetic disk storage device by COMPLETE operation 709.
Reports ETE operation 710 and terminates access to the magnetic disk storage device.

(2) Executing the WRITE operation The WRITE operation is performed using the ASSIGN MIRRORD command that sets duplexing.
By setting the K command 302 (D7-y), the combination of operation modes shown in FIG. 8 can be set. Each operation mode is arbitrarily set by parameters.

In the following execution example of the WRITE operation, the dual magnetic disk storage devices 1OA and 110 in the operation mode 803 shown in FIG.
The central processing unit 102 that made the WRITE request to the magnetic disk storage device when the ITE operation was completed first
An example of operation in high-speed access mode (FIG. 9) in which a COMPLETE operation is reported to
8.110, performs a WRITE operation on the mask device, and upon completion of the WRITE operation of the mask device, instructs the slave device to perform the WRITE operation, and upon completion of the slave operation, executes the WRITE operation on the central processing unit 102. COM
An operation example (FIG. 10) of a highly reliable access mode is shown in which the normality of data in one of the duplexed devices is always guaranteed by reporting the PLETE operation.

(A) Processing flow of high-speed access mode The central processing unit 102 creates and stores the parameters necessary for the WRITE operation in the main memory unit 103 as parameters 901, specifies the device identifier 'DKOOOI', and selects IOP#0 (peripheral device). EXECUTE operation 90 to the control unit 104)
Issue 2.

IOP#0 executes an EXECUTE operation 902 and fetches 903 a parameter 901 from the main memory section 103. It is determined that the content of the parameter 901 is a WRI TE command, and the magnetic disk storage device 10
Activation 904 is performed for 8. followed by the device identifier'
It is determined 905 whether DKOOolo is a duplication device or not and its operating mode by checking an internal control table. If it is not set as a duplex device, in order to execute a normal WRITE operation, it waits for a WRITE operation completion report 911 and sends a COMPLETE operation 912 to the central processing unit 102.
notification processing.

If it is set as a duplex device, IOP #1 is activated, but first a parameter 906 necessary for the WRITE operation is created, and an EXECUTE operation 907 is issued in the same procedure as the central processing unit 102.

When IOP#1 receives the EXECUTE operation 907 from IOP#0, it maintains that the magnetic disk storage device 110 under its own device has been activated as a slave device, and fetches the parameter 906. It is determined that the content of the parameter 906 is a WRITE command, and the magnetic disk storage device 110 is activated 909.

In this operation example, in accessing the magnetic disk storage device, first WRITE is sent from the magnetic disk storage device 108h).
A case where an operation completion report 911 is generated is shown. IOP#0
When it receives a WRITE operation completion report 911 from the magnetic disk storage device 108, it reports a COMPLET operation 912 to notify the central processing unit 102, which has requested access to the magnetic disk storage device, of normal completion, and the software Notifies the end of storage device access.

After this, WRITE from the magnetic disk storage device 110
IOP#0 independently waits for a report of operation completion. I
OP#1 is WRITE from the magnetic disk storage device 110.
When the operation completion report 914 is received, WR is sent to IOP#O.
Reports a COMPLETE operation 915 to signal completion of an ITE operation. As a result, IOP#O
performs termination processing 916 of the WRITE operation request from the central processing unit 102 and ends the WRITE processing.

The above-mentioned termination condition for the magnetic disk storage device access is from the magnetic disk storage device 108 to the magnetic disk storage device 11.
If 0 occurs first, the COMPLETE operation 91 receives the report of the COMPLETE operation 915 and notifies the central processing unit 102 of normal completion.
2, and the software is notified of the end of access to the magnetic disk storage device.

(B) Processing flow of highly reliable access mode In the processing flow of Figure 10, IOP#0 is the magnetic disk storage device 1.
The processing up to starting 904 of 08 is the same as that shown in FIG.

Next, it is determined 1001 whether or not the device identifier 'DKOOOI' is a dual device and its operation mode by checking the internal control table.

In this operation example, it is determined 1002 that a duplication device is set and that serial execution is set (mode 801 in FIG. 8 is set). As a result, after waiting for the completion of access to the magnetic disk storage device 108,
A WRITE operation of the magnetic disk storage device 110 is performed.

When IOP#O receives the access completion report 1003 of the magnetic disk storage device 108, it sends W to IOP#1.
In order to start RITE, parameters necessary for the WRITE operation are created 1004, and an EXECUTE operation 1005 is issued in the same procedure as the central processing unit 102.

When 10P#1 receives (7) EXECUTE operation 1005 from IOP#0, it holds that the magnetic disk storage device 110 under its own device has been activated as a slave device, and fetches the parameter 1004 (1006).
do. It is determined that the content of the parameter 1004 is a WRITE command, and the magnetic disk storage device 110 is activated 1007.

After this, when a WRITE operation completion report 1009 is generated from the magnetic disk storage device 110, 10P#1 sends a COMPLETE operation 1010 indicating the WRITE operation completion report from the magnetic disk storage device 110 to IOP#1.
Report to O. C0MPLETE operation 101
0 report, IOP#O reports a COMPLETE operation 1011 to notify the central processing unit 102 that has requested the magnetic disk storage device access of normal completion, and the software is instructed to terminate the magnetic disk storage device access. Notice.

In this highly reliable access mode, access time to the magnetic disk storage device is twice as long on average as when it is not duplexed, but the advantage is that the recorded contents of one side of the duplexed magnetic disk storage device are always guaranteed. There is.

In this embodiment, the CPU 102 boots only one of the dual magnetic disk storage devices;
It is also possible for the PU 102 to access two dual magnetic disk storage devices. In this case, the control procedure described above is applicable, except that the master/slave relationship must be set in advance for the dual magnetic disk storage device.

In addition, although we have shown an example of operation in which magnetic disk storage devices under different peripheral device control units are specified as a pair of duplexing devices, two magnetic disk storage devices under the same peripheral device control unit are specified as bare in a duplexing device. However, the control procedure described above is applicable.

Furthermore, the number of magnetic disk storage devices is not limited to two, and it goes without saying that the present invention can also be applied to multiplexing of more than two magnetic disk storage devices.

[Effects of the Invention] As explained above, in the present invention, when a processing request is issued from a central processing unit to a multiplexed magnetic disk storage device and the processing request is a READ process, the system is configured as a multiplexing device. Reads data only from the first magnetic disk storage device that can start data transfer and stores it in the designated main memory section, and terminates the READ operation of the remaining magnetic disk storage devices midway. Since the means is provided, when executing the READ process to the magnetic disk storage device, data is read from the magnetic disk storage device in which the 5EEK operation and the search operation have been completed first.
It is possible to control transfer to a specified area of the main memory section. As a result, compared to the conventional high-speed access control method that transfers data from a magnetic disk storage device that finishes 5EEK operation earlier, for example, in the case of a duplex magnetic disk storage device, the magnetic disk storage device response is 1/4 rotation on average. It is possible to save time.

Since the positioning status of the magnetic disk storage device is reported through communication between peripheral device control units, there is no need to manage the head position to determine which multiplexed magnetic disk storage device has the shortest 5EEK time.

In this invention, a WRITE operation is instructed to all magnetic disk storage devices designated as a multiplexing device, and a WRITE operation is performed in any one magnetic disk storage device.
When the operation is complete, write WRIT to the central processing unit.
Since we have provided a means to report the completion of E processing, WRITE
The response time required of the system in access request processing,
It is possible to select WRl and TE processing according to the reliability configuration. As a result, as a control method for realizing high-speed access to a magnetic disk storage device by utilizing a multiplexed magnetic disk storage device, the same control procedure as the READ operation is applied, and 5EEK operation, search operation, and data By reporting the completion of access to the magnetic disk storage device to the central processing unit that generated the processing request at the end of the magnetic disk storage device in which the total write operation time is short, the response to WRI'TE can be shortened. In the case of a highly reliable configuration, by sequentially executing multiple magnetic disk storage devices, even if a failure occurs during writing in one magnetic disk storage device, processing is interrupted due to the failure and the other magnetic disk Since the contents of the storage device are guaranteed, failures that result in complete loss of previous data can be avoided.

[Brief explanation of the drawing]

FIG. 1 is a configuration diagram of an information processing system in this invention.
Figures 2 (a) and (b) are diagrams showing operation functions used for communication between modules necessary to explain the operation example, and Figures 3 (a), (b), and (C) are magnetic disk Figure 4 shows the contents of the EXECUTE operation command necessary to control the storage device. Figure 4 shows the internal configuration of the peripheral device control unit. Figures 5 (a) and (b) show the dual magnetic disk storage device. FIG. 6 is a diagram showing the contents of the EXECUTE operation and the parameters when allocating devices as a pair; FIG. 6 is a diagram showing the flow of processing when access to the mask device is completed earlier than the slave device in the high-speed operation mode of the READ operation; FIG. 7 is a diagram showing the flow of processing when the slave device access ends earlier than the mask device access in the high-speed operation mode of the READ operation, and FIG. 8 is a diagram showing the WRITE operation mode that can be specified by the ASSING MIRRORDK command. , FIG. 9 is a diagram showing the processing flow of a WRITE operation with the high-speed access mode specified, and FIG. 10 is a diagram showing the processing flow of the WRITE operation with the high-reliability access mode specified. In the figure, 101 is an information management system 102 is a central processing unit, 103 is a main memory section, 104 and 105 are peripheral device control sections, 106 is a common section, 107 is a magnetic disk storage device controller, and 108 to 111 are magnetic disks. It is a storage device. Figure 1 (a) (b) Format 206 for operation communication

Claims (2)

[Claims] (1) Consists of a central processing unit, a main memory section, one or more peripheral control devices, and a plurality of magnetic disk storage devices connected to the peripheral control device. A peripheral control device that processes as a multiplexing device that holds data, receives processing requests from the central processing unit to the magnetic disk storage device, executes the processing requests, and reports processing results to the central processing unit. A processing request is issued from the central processing unit to the encoded magnetic disk storage device,
If the processing request is a READ process, data is read only from the first magnetic disk storage device that is set as a multiplexing device and is capable of starting data transfer, and stored in the designated main memory section. A peripheral control device for a magnetic disk storage device, comprising means for terminating the READ operation of the remaining magnetic disk storage devices midway. (2) Consisting of a central processing unit, a main memory section, one or more peripheral control devices, and a plurality of magnetic disk storage devices connected to the peripheral control device, the plurality of magnetic disk storage devices are connected to the same A peripheral control device that processes as a multiplexing device that holds data, receives processing requests from the central processing unit to the magnetic disk storage device, executes the processing requests, and reports processing results to the central processing unit. A processing request is issued from the central processing unit to the encoded magnetic disk storage device,
If the processing request is WRITE processing, WRITE processing is performed for all magnetic disk storage devices specified as multiplexing devices.
A magnetic disk storage device comprising a means for instructing an ITE operation and reporting completion of the WRITE process to a central processing unit when the WRITE operation is completed in any one magnetic disk storage device. peripheral control device.