JPH02222027A - Data multi-writing control system for auxiliary memory - Google Patents

Abstract

PURPOSE:To attain the automatic transfer of the correct data to a CPU even when the correct recording of data is impossible due to a fault by recording the position of an auxiliary memory where the fault is produced into a memory of the auxiliary memory. CONSTITUTION:When a data writing request is given to a logical magnetic disk device 30 from a CPU 1, a magnetic disk controller 2 decides a position of the device 30 where the data are written. Then the controller 30 selects one of magnetic disk devices 31 - 33 forming the device 30 that contains the data written correctly at the due writing position by reference to the fault information on a direct memory 25. Then the data are simultaneously sent to the magnetic disk device selected and a disk cache memory 24 of the controller 2. The controller 2 confirms that the data are correctly written into the memory 24 and the selected magnetic disk device and reports this fact to the CPU 1.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to a multiple write method in a computer system, and particularly to improving the reliability of data stored in an auxiliary storage device.
This relates to an auxiliary storage control method for

[Conventional technology]

Conventionally, as a method for increasing the reliability of data stored in an auxiliary storage device, there is a data double writing method using a magnetic disk control device as described in, for example, Japanese Patent Laid-Open No. 61-249132.

In this method, the same data is written to two magnetic disk drives by the operating system, and if one magnetic disk drive fails, processing is performed using the data stored in the other magnetic disk drive. It is possible to continue processing and prevent interruption of processing. In this case, the feature of the method described in the above publication is that the same data can be written to two magnetic disk devices by issuing a single data write command from the operating system9. A data buffer is provided in the control device that controls the magnetic disk device, which is an auxiliary storage device, and when a data write request is generated from the central processing unit, data is stored in the first magnetic disk device and the data buffer. After that, the magnetic disk control device automatically writes the write data from the central processing unit stored in the data buffer to the second magnetic disk device. This makes it possible to write the same data to two magnetic disk devices with a single data write request.

[Problem to be solved by the invention]

The above-mentioned publication does not describe anything about what to do when one magnetic disk device fails.

In the conventional method, when writing the same data to multiple magnetic disk drives and reading that data from the magnetic disk drives, if one of the magnetic disk drives fails, the operating system uses the other normal magnetic disk. It is necessary to read data from the device. To do this, the operating system must always know the status of all magnetic disk devices into which the same data has been written. As a result, there is a problem in that the processing of the operating system becomes complicated.

The purpose of the present invention is to solve such conventional problems, and when a part of a magnetic disk device that performs multiple writing fails, the correct data is automatically transferred to the central processing unit without depending on the operating system. An object of the present invention is to provide a data multiplex write control method for an auxiliary storage device that can transfer data to the auxiliary storage device.

[Means to solve the problem]

In order to achieve the above objects, the present invention provides a data multiplex writing control method for an auxiliary storage device, which is an auxiliary storage device connected to a central processing unit via a channel interface and connected to a plurality of auxiliary storage devices via a drive interface. It is a control device.

The central processing unit is made to recognize a plurality of auxiliary storage devices that have been combined in advance as one auxiliary storage device (hereinafter referred to as a logical auxiliary storage device), and a data write request is generated from the central processing unit to the auxiliary storage device. Then, one auxiliary storage device is selected from among the combinations configuring the logical auxiliary storage device, and data is written to the auxiliary storage device and the data buffer in the auxiliary storage control device,
When the writing is completed normally, the auxiliary storage control device reports the normal completion of writing to the central processing unit, and then writes data to the remaining auxiliary storage devices combined as the logical auxiliary storage device, A feature of the present invention is that when a failure occurs while data is being written to the auxiliary storage device, the location of the write failure is stored in each auxiliary storage device of the memory within the auxiliary storage control device.

[For production]

In the present invention, (i) when data transferred from the central processing unit is written to multiple auxiliary storage devices, if the data cannot be stored correctly due to a failure, the auxiliary storage device in which the failure occurred is The location is recorded in the memory in the auxiliary storage controller (see Figure 1). (n) In addition, if a failure occurs in one auxiliary storage device when data transferred from the central processing unit is written to multiple auxiliary storage devices, the location of the failure will be recorded and other auxiliary storage devices will be written. When the data can be recorded correctly, the auxiliary storage control device reports to the central processing unit that the data transfer was performed normally (see FIG. 3). (iii) When writing data to multiple auxiliary storage devices, if a failure occurs in all of the auxiliary storage devices, the central processing unit is notified that the data transfer could not be executed correctly (see Figure 3). reference). (iv) Next, when transferring the data on the auxiliary storage device to the central processing unit, refer to the failure location information and select the auxiliary storage device in which the data is correctly recorded (see Figure 3). . (V) In addition, if the data on the auxiliary storage device cannot be read correctly when transferring the data on the auxiliary storage device to the central processing unit, the location on the auxiliary storage device where the failure occurred is stored in the memory within the auxiliary storage control device (see FIG. 3). (vi
) When transferring the data on the auxiliary storage device to the central processing unit, if there is no auxiliary storage device in which the data is correctly recorded after reading the data on each auxiliary storage device, the data is transferred to the central processing unit. to report that the data transfer could not be executed correctly (see Figure 3). (Advanced) Next, when a failure occurs during writing or reading, information on the location where the failure occurs is recorded in the storage medium of the auxiliary storage device. (=4) Also, the memory in the auxiliary storage control device is made non-volatile. (ix) In addition, a hierarchically structured data table is stored in the memory in the auxiliary storage control device, and it is determined from this table whether or not data on the auxiliary storage device exists in the data buffer in the auxiliary storage control device. Determine whether the data is normal or not.

As a result 1, in the present invention, it is possible to store data in multiple auxiliary storage devices with a single write request from the central processing unit, and a failure occurs in the auxiliary storage device when writing or reading data. In this case, input/output operations can be performed using the normal auxiliary storage device without the intervention of the operating system.

Additionally, if the auxiliary storage device is a removable storage medium such as a magnetic tape device, failure information corresponding to the storage medium is recorded on the storage medium itself when the storage medium is removed, and the storage medium is attached to the auxiliary storage device. When a removable storage medium is used, multiple writing of data becomes possible even in the auxiliary storage device of a removable storage medium by expanding the failure information recorded at the time of removal into the control memory in the auxiliary storage control device.

Furthermore, if the auxiliary storage device is a magnetic disk device with disk cache memory, when the central processing unit accesses the data on the magnetic disk device, the magnetic disk device must be able to correctly record data from the cylinder table and track table. By determining whether the data exists or not, or whether the data exists on the disk cache memory, it is possible to speed up the search for correctly recorded data.

〔Example〕

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 2 is a block diagram of a magnetic disk control device showing one embodiment of the present invention.

As shown in FIG. 2, the magnetic disk control device 2 is connected to the central processing unit 1 via a channel interface, and includes non-removable magnetic disk devices 31, 32, and 33 via drive interfaces. Connected to a group of magnetic disk devices. The channel interface and drive interface of the magnetic disk control device 2 are interconnected via a data bus 23. Connected to this data bus 23 is a disk cache memory 24 for storing data on the magnetic disk device. The magnetic disk control device 2 includes the central processing unit 1
a channel-side control unit 2o that controls data transfer between the
It is controlled by two control units of a drive-side control unit 22 that controls data transfer with a group of magnetic disk devices. These control units 20 and 22 have a control memory 21 storing a device control table for controlling the entire magnetic disk control device 2, and a correspondence relationship between data on the magnetic disk device and data on the disk cache memory 24. and a directory memory 25 that stores the fault location on the auxiliary storage device.

Although this embodiment shows an example of the magnetic disk control fold placement 2 controlled by two control units 20.22, by arranging a plurality of pairs of control units 20.22, a plurality of control units 20.22 can be used simultaneously. It is possible to process input/output requests from the central processing unit.

Further, a service processor 3 is connected to the magnetic disk control unit 2 via an SvP interface.

The service processor 3 includes a floppy disk device 4, and serves to monitor the execution status of the magnetic disk control device 2 via the SvP interface.

The magnetic disk control device 2 includes a magnetic disk device 31.3.
2 appears to the operating system 10 running on the central processing unit 1 as if it were a single magnetic disk device. In this case, one magnetic disk device in which these two magnetic disk devices [31, 32 function] will be referred to as a logical magnetic disk device 30.6 Logical magnetic disk device 30 and magnetic disk device 31.3
2 is set by an instruction from the operating system 10 or the service processor 3, and this device configuration information is stored in the control memory 21 in the magnetic disk control device 2.

An input/output request from the central processing unit l to the logical magnetic disk device 30 may specify a special device address, or may specify the address of one of the magnetic disk devices constituting the logical magnetic disk device 3o. In either case, the logical magnetic disk device 30 is accessed.

When a write request is made from the operating system 10 on the central processing unit 1 to the magnetic disk devices 31 and 32 that make up the logical magnetic disk device 30, all magnetic disk devices that make up the logical magnetic disk device 30 are requested to write. The same data is written to the disk, thereby realizing the dual write function of the magnetic disk device. Here, 2
Although an example of a logical magnetic disk device 30 consisting of two magnetic disk devices 31 and 32 is shown, it can be expanded to two or more magnetic disk devices, so multiple writing control is possible.

FIG. 3 is a schematic flowchart of the present invention.

When a data write request is issued from the central processing unit 1 to the logical magnetic disk device 30, the magnetic disk controller @2 first determines the location on the magnetic disk device that is the write target, and then writes the data to the logical magnetic disk device 30. Magnetic disk device 30
Among the magnetic disk devices 31 and 32 that make up the system, only one magnetic disk device in which data is correctly written to the write target location is selected by referring to the fault information on the directory memory 25 (step 4001). )
. The selected magnetic disk device is called a synchronous processing auxiliary storage device. Data is simultaneously transferred from the central processing unit 1 to the synchronous processing auxiliary storage device and the disk cache memory 24 in the control device 2 (step 4002). The magnetic disk control device 2 includes a disk cache memory 24
Then, it is confirmed that the data has been correctly written to the synchronous processing auxiliary storage device (step 4003). When the data is correctly stored in this combined and synchronized auxiliary storage device, it is reported to the central processing unit 1 that the data write request has been successfully completed. If the write operation to the synchronous processing auxiliary storage device is not performed normally, the location of the failure is recorded in the directory memory 25 (step 4003), or if there is another normal magnetic disk device (step 40).
05), select from among the logical magnetic disk devices 30 (
Step 4001), this is used as a synchronous processing auxiliary storage device to transfer write data from the central processing unit 1 again (Step 4002).

In this case, if a normal magnetic disk device cannot be selected, a write error is reported to the central processing unit 1 (step 4006).

When data is written to the synchronous processing auxiliary storage device (step 4003), normal completion of the input/output request is reported to the central processing unit (step 4007), and then,
Logical magnetic disk device 30 excluding synchronous processing auxiliary storage device
The write data in the disk cache 24 is sequentially written to the magnetic disk devices that make up the disk cache 24 (step 4).
009).

At this time, if data cannot be correctly written to the magnetic disk device, the location of the failure is recorded in the directory memory 25.

Next, when a data read request is issued from the central processing unit 1 to the logical magnetic disk device 30,
First, from among the magnetic disk devices constituting the logical magnetic disk drive to be read [30], select the magnetic disk device in which the data to be read has been correctly written, by referring to the directory memory 25, in the same way as when writing. Only one synchronous processing auxiliary storage device is selected (step 4001). Then, the magnetic disk control device 2 transfers the data to be read in the selected synchronous processing auxiliary storage device to the central processing unit 1 (step 4002).

If the data has been successfully transferred to the central processing unit 1 (step 4003), it is reported to the central processing unit 1 that the data read request has been successfully completed (step 4007). On the other hand, if the read processing from the synchronous processing auxiliary storage device is not performed normally (step 4
003), after recording the fault location in the directory memory 25 (step 4004), one of the magnetic disk devices in which data is normally stored is selected from among the other magnetic disk devices (step 4001). , this is used as a synchronous processing auxiliary storage device, and data to be read is transferred to the central processing unit 1 (step 4001).

FIG. 1 is a block diagram of a control memory and a directory memory showing one embodiment of the present invention, and in FIG.
A configuration related to a control table for controlling a magnetic disk controller is shown.

As a control table, a device control block D CB (Device Control) in the control memory 21 is used.
1B Lock) and a slot control block S CB (S lot Cont) in the directory memory 25.
rol B Lock). DCBlooO-
a, 1000-b.

This control block exists in each magnetic disk device controlled by the magnetic disk controller 2 and is used to execute input/output operations for the magnetic disk device.

On the other hand, 5CB4000-1.4000-2 is a control block for managing data stored in the disk cache memory 24. In addition, cylinder tables 2000-a, 200o-b and track tables 3000-a, 30oo-b are used to record failures and to quickly determine whether data on a magnetic disk device exists in cache memory. established for the purpose of

DCBlooO-a and DCB 10oo-b control the magnetic disk devices that constitute the logical magnetic disk device.
They are provided corresponding to the magnetic disk drives 31 and 32, respectively, and are connected by pointers 1500 and 160o to manage their configurations. That is, pointer 1500
．． Two tables 1000-a and b are joined to each other by 1600, and each is joined to 5CB4000-1.2 by pointer 1700.1800, and by pointer 1100.
cylinder table 2000-a.

It is connected to b.

Next, cylinder tables 2000-a and 2000-b are provided corresponding to the magnetic disk devices that also constitute the logical magnetic disk device 30, and are coupled to track tables 3000-a and 3000-b, respectively, by pointers. Both the cylinder table 2000 and the track table 3000 are used to determine whether data on the magnetic disk devices 31 and 32 exists on the cache memory 24 or whether a failure has occurred in the relevant track. This is the table. The cylinder tables 200Q-a and 200Q-b exist for each magnetic disk device, and their entries correspond to the cylinders of the magnetic disk device.
It shows the beginning of 000. truck table 3000
The entries correspond to tracks of the magnetic disk device, and are stored in the disk cache memory 24 along with a failure flag 3100 (indicated by $1) indicating whether or not those tracks can be accessed normally. It is composed of an area 3200 indicating the address of a slot control block (SCB) 4000, which is a control block for managing data.

A NULL code is set in the entry on the track table corresponding to a track for which no data exists in the disk cache memory 24.

The operation of multiple write control for the magnetic disk device will be described below.

When the operating system 10 on the central processing unit 1 makes an input/output request to the magnetic disk device 31, the magnetic disk control device 2 that received the input/output request stores the DCB pointer stored in the control memory 21. 1500,
1600, it is determined that the magnetic disk device 31 is a magnetic disk device constituting the logical magnetic disk device 30 since they are coupled to each other. Then, a magnetic disk device @ (referred to as a synchronous processing magnetic disk device) that executes input/output operations from the central processing unit 1 is selected from among the logical magnetic disk devices 30 . This process will be referred to as synchronous magnetic disk device selection process.

FIG. 4 is an operational flowchart of synchronous magnetic disk device selection processing in the present invention.

When an input/output request is issued from the central processing unit 1 to a magnetic disk device controlled by the magnetic disk control device 2, the DCBlooO corresponding to the magnetic disk device is
is read from the control memory 21 (step 5000),
By referring to DCBlooO, it is determined whether the magnetic disk device to which the input/output request was issued is in use (step 5001). As a result of the determination, if it is determined that the magnetic disk device is not in use, the magnetic disk device is designated as a synchronous processing magnetic disk device, and the fact that it is in use is recorded in DCBlooO (step 5005). Next, the central processing unit 1
Report device availability to (step 500
6).

Further, if it is determined that the magnetic disk device is in use (step 5001), by referring to the pointer 1500.1600 of DCBlooo, it is determined that the magnetic disk device is a magnetic disk device that constitutes a logical magnetic disk device. (Step 50)
02). As a result of the determination, if it is determined that the magnetic disk device is a magnetic disk device that configures a logical magnetic disk device, select a magnetic disk device that is not currently in use from among the magnetic disk devices that configure the logical magnetic disk device. one unit is selected (step 5003).

Next, it is determined whether all the magnetic disk devices that make up the logical magnetic disk device are in use (step 5004).
If an unused magnetic disk device exists, the unused magnetic disk device is designated as a synchronous processing magnetic disk device for executing input/output requests from the central processing unit 1 (step 5005). . In this case, after setting DCBlooO corresponding to this magnetic disk device to be in use, it reports to the central processing unit 1 that the device can be used (step 5006).

Further, if the magnetic disk device to which the input/output request was issued is in use and it is determined that the magnetic disk device does not constitute a logical magnetic disk device (step 500
1.5002), or if all the magnetic disk units constituting the logical magnetic disk unit are in use (step 5004), the CPU 1 reports device busy to the central processing unit 1 (step 5007).

Upon receiving the report that the device is available for use, the central processing unit 1 transfers the channel command word (hereinafter referred to as CCW) created by the operating system 10 at the time of an input/output request and its parameters to the magnetic disk control unit 2 .

The magnetic disk control device 2 that has received the CCW and its parameters executes the input/output request from the central processing unit 1 using a synchronous magnetic disk device. When this input/output request is a data write request to a logical magnetic disk device, after the input/output processing to the synchronous magnetic disk device is completed, it is transferred from the central processing unit 1 to all the magnetic disk devices that make up the logical magnetic disk device. Write the data.

Below, the magnetic disk controller 2 responds to a data write request from the central processing unit 1 to the logical magnetic disk device 30.
The processing will be explained using the data write CCW and its parameters.

FIG. 5 is a diagram showing a group of channel commands and their parameters prepared by the operating system when writing data on a magnetic disk device.

In FIG. 5, 5SCH is an input/output activation request command 11 executed by the operating system 10 on the central processing unit 1, and 12 is an operation request block when issuing this input/output activation request.

Further, the CCW group for performing a write operation on the magnetic disk device is 13 including (1oO to 105),
connected by pointers (110 to 1
14) are the parameters of CCW.

The input/output start command (S
When an input/output request is activated by SCH:5TART 5UBCHANNEL) 11 to a magnetic disk device controlled by the magnetic disk controller 2, the magnetic disk controller 2 selects a synchronous magnetic disk device by the channel side control unit 20. do. Here, magnetic disk device 3
1 is selected. Thereafter, the central processing unit 1 selects the CCW from the CCW group 13 indicated by the CCW address in the operation request block 12 specified by the 5SCH instruction 11.
are read out one by one and transferred to the magnetic disk control device 2. When the magnetic disk control device 2 receives these CCWs, the channel side control unit 20 processes these CCWs.

Next, the operation of the channel-side control unit 20 that receives the CCW will be described in detail.

The 5EEK command 100 specifies the cylinder address (CC) and head address (HH) on the magnetic disk device 31.
) is specified as the seek address 110. This 5E
When the channel-side control unit 20 receives the EK command 100, the cylinder address that is the operand of this command, and the head address, if this seek address 110 is determined to be logically correct, the value is transferred to DCBlo.
oO and reports the (channel end) + (device end) DSB (device status byte) to the central processing unit 1. The central processing unit 1 that received (channel end) + (device end) is 5EEK
Following command 100 < DE F I NEEXTEN
The T command 101 is sent to the magnetic disk controller 2 together with the extent information 111 that is a parameter of the command.
as a command chain. DEFINE
The channel side control unit 20 that receives the EXTENT command 101 and parameter extent information 111 performs a logical check on the extent information 111, and if this parameter is determined to be correct, stores this value in DCBlooO. After that, the DSB of (channel end) + (device end) is reported to the central processing unit 1. (Channel end) + (Device end) D
The central processing unit 1 that received the SB is the DEFINE
Following EXTENT command 101 <SET 5ECT
The OR=+v command 102 is transferred to the magnetic disk controller 2 as a command chain together with the sector number 112 which is a parameter of this command. When the channel side control unit 20 receives the SET 5ECTOR command and the sector number, it performs a logical check of the sector number and stores this sector number in the DCB 1000. Then, by processing the 5EEK command that Uke took earlier, DCBlo
Read the seek address stored in oO and use this as a key in the cylinder table 2 on the directory memory 25.
000 and the track table 3000, and it is determined whether or not data to be input/output exists on the disk cache memory 24, that is, a hit/miss determination is performed. As a result of the hit/miss determination, if it is determined that no data exists on the disk cache memory 24 (miss), one free area in the disk cache memory 24 is secured. The disk cache memory 24 has a capacity of 5 CB, with one slot corresponding to one track capacity of the magnetic disk device.
4000.

The SCB has the following states (i) to (inn).

(i) Unused state, (5) In use state, (iii) Reusable state. An SCB in an unused state indicates that slot I- on the disk cache memory 24 managed by this SCB is free. show. The in-use state is a state in which the data in the slot on the disk cache memory 24 managed by the SCB must not be deleted. Specifically, the data in the slot is deleted due to a data read request from the central processing unit 1. is in a state where data is being read, or a data write request is issued from the central processing unit 1 to the logical magnetic disk device 30, and data writing processing is performed on all the magnetic disk devices that make up the logical magnetic disk device 30. Shows the status until the end. Furthermore, the reusable state indicates a state in which the data in the slot on the disk cache memory 24 managed by the SCB matches the contents of the magnetic disk device, and the slot can be discarded. In addition, when the above-mentioned magnetic disk device is a logical magnetic disk device 30, it is necessary that the contents match the contents of all the magnetic disk devices 31 and 32 that constitute the logical magnetic disk device 30.

FIG. 6 is a processing flowchart for securing an area (slot) on the disk cache memory of FIG. 2.

First, it is determined whether or not there is an unused SCB (step 5100), and if there is an unused SCB, one SCB is secured from among the unused SCBs, and that SCB is
is in use (step 5102). Also, unused S
If it is determined that the CB does not exist (step 51
00), it is determined whether a reusable SCB exists or not (step 5101), and when it is determined that a reusable SCB exists, one SCB is secured from among the reusable SCBs and the SCB is In use (step 5)
102). Furthermore, if it is determined that there is no reusable SCB (step 5101), the reusable SCB
The input/output request from the central processing unit 1 is placed in a waiting state until B occurs (step 5105). If an unused SCB or reusable SCB can be secured, that SC
The number of the magnetic disk device in which the data in the slot on the disk cache memory 24 managed by B is arranged, the cylinder address and track address on the magnetic disk device are stored (step 5103). Furthermore, the address of the SCB is stored in the entry of the track table 3000 corresponding to the track (step 5104). Note that if the magnetic disk devices constitute a logical magnetic disk device, the process is executed for all magnetic disk devices that make up the logical magnetic disk device (step 5104).

Through the above processing, processing for securing an empty slot is executed, and then a head positioning instruction is given to the drive-side control unit 22 for the magnetic disk device 31, and a channel end DSB is sent to the central processing unit 1. After reporting, it waits for the completion of positioning from the drive-side control unit 22. When the drive-side control unit 22 receives a head positioning request, it positions the head on the target cylinder based on the seek address in the DCB, and then selects the head corresponding to the track and performs the specified head positioning. The rotation waits until the sector number is reached. When the sector is located at the designated sector number, the drive-side control unit 22 is sent from the magnetic disk device 31.
A sector ready interrupt occurs. The drive-side control unit 22 that has harvested this interrupt notifies the channel-side control unit 20 of the occurrence of sector ready. When the channel side control unit 20 receives the sector ready notification, it reports the DSB of the device end to the central processing bag @1 and sends it to the SE.
T5 Ends ECTOR command processing (ends command 102 in FIG. 5). Furthermore, if it is determined that data exists (hit) on the disk cache memory 24 as a result of the hit/miss judgment, an address is assigned to the specified sector position on the disk cache memory 24 from the information stored in the 5CB4000. , (channel end) + (device end) DSB is reported to the central processing unit 1. The central processing unit 1 is SET 5ECTOR
When the DSB indicating the end of the command is received, as shown in FIG. Transfer to. The channel side control unit 20 controls these 5EARCHIDs.
When the EQUAL command 103 and the record ID 113 are received, if the previous hit/miss judgment has determined that it was a mistake, the key part of the first record of the sector for which the positioning instruction was given to the magnetic disk device 31 is read out. An instruction is given to the drive-side control unit 22. Next, the channel side control unit 20 compares the record ID 113 transferred from the central processing unit 1 with the key value read by the drive side control unit 22, and when the values match, (state modification) + (channel end) + (device end) DSB
is reported to the central processing unit 1. In addition, the contents of part of the keys read from the magnetic disk device 31 and the record ID 11
If the values of 3 do not match, the central processing unit 1 (
Reports the DSB of channel end) + (device end).

The channel side control unit 20 is 5EARCHIDEQUAL
When a command is received, if it is determined as a hit by the previous hit/miss determination, a part of the record addressed by the SET 5 ECTOR command processing is read from the disk cache memory 24 and transferred from the central processing unit 1. Centrally processes the DSB of (state modification) + (channel end) + (device end) when they match, and the DSB of (channel end) + (device end) when they do not match. Report to device 1.

As shown in FIG. 5, when the central processing unit 1 receives DSB of (channel end) + (device end),
Following EARCHID EQUAL command 103 <T
Execute IC command 104. TIC command 104
is a command indicating the address of the CCW to be executed next, and in the case of this embodiment, indicates the 5EARCHID EQUAL command 103, which is the previous command. Therefore, as long as the central processing unit 1 receives reports of (channel end) + (device end),
The command 103 will be repeatedly sent to the magnetic disk controller 2.

Moreover, when the central processing unit 1 receives the DSB of (state modification) + (channel end) + (device end),
WRIHT DATA105, which is the second command after EARCHID EQUAL, and its WRIHT
Data 114 to be written to the magnetic disk device by the DATA command is transferred to the magnetic disk control device 2. That is, 5 EARCHID EQUAL command 10
3 and the '0C command 104, it is possible to locate the target record for input/output on the magnetic disk device.

The channel side control unit 20 of the magnetic disk control device 2 receives the WRIHT DATA command 10 from the central processing unit 1.
5 and write data 114, the previous hit/
If a hit is determined as a result of a miss determination, a positioning instruction is given to the drive-side control unit 22 based on the seek address, sector number, and record ID stored in the DCB of the magnetic disk device 31. This is to ensure data reliability by writing the data directly onto the magnetic disk device when a write request is received from the central processing unit 1.

After completing the positioning at the time of hit, the channel side control unit 20
Or, immediately after receiving the WRIHT DATA command in the event of a miss, it prepares for data transfer to the disk cache memory 24 and instructs the drive-side control unit 22 to prepare for data transfer. When the drive-side control unit 22 completes the preparation for data transfer to the magnetic disk device 31, the write data 114 from the central processing unit 1 is transferred to the disk cache memory 24 and the magnetic disk device 31, respectively. When this data transfer is completed, the channel side control unit 20 checks the data transfer result of the drive side control unit 22 to the magnetic disk device 31 and the data transfer result to the disk cache memory 24, and both end normally. In this case, the record position (seek address, sector number, record ID) where the write operation was performed and the end of the write process to the magnetic disk device 31 are specified in 5CB40.
0o and transfers the (channel end)+(device end) DSB to the central processing unit 1.

As a result, the data write input/output request to the magnetic disk device 31, which is a synchronous magnetic disk device, ends, and the input/output request issued from the central processing unit 1 ends. The write operation to the magnetic disk device described above is called online destaging processing. After completing the online destaging process.

The drive-side control unit 22 selects a magnetic disk device that has not executed the write request requested by the central processing unit 1 from among the magnetic disk devices configuring the logical magnetic disk device by referring to 5CB4000. . Here, the magnetic disk device 32 is selected, so
A write operation to the magnetic disk device 32 is executed asynchronously with the central processing unit 1. This write operation is called offline destage processing. The offline destaging process will be described below.

The drive-side control unit 22 starts offline destaging processing at a certain cycle after the online destaging processing from the central processing unit 1 to the magnetic disk devices constituting the logical magnetic disk device 30 is completed. In the offline destaging process, first, among the magnetic disk devices configuring the logical magnetic disk device 30, a data output request from the central processing unit 1 is sent to a magnetic disk that has not yet been processed by the online destaging process or the offline destaging process. Select a device (in this embodiment, the magnetic disk device 32
). The drive-side control unit 22 issues a positioning request to the selected magnetic disk device 32 based on the seek address, sector number, and record ID stored in the 5CB 4000 during online destage processing.

When the positioning process is completed, the data on the disk cache memory 24 is transferred to the magnetic disk device 32, and the completion of data transfer to the magnetic disk device 32 is recorded in 5CB 4000. This ends the offline destaging process. When the offline destaging process to the magnetic disk device 32 is completed, data writing from the central processing unit 1 to all the magnetic disk devices making up the logical magnetic disk device 30 is completed.

At this time, 5CB4000 is assumed to be a reusable SCB.

If a logical magnetic disk device consists of two or more magnetic disk devices, offline destage is performed for all magnetic disk devices that make up the logical magnetic disk device, except for the synchronous processing magnetic disk device. By executing the process, N-overwrite operation of the magnetic disk device is realized.

Next, the operation at the time of failure will be explained in detail.

When the magnetic disk control device 2 detects some kind of failure in a track on the magnetic disk device to which data is written during data writing operation for online destaging processing, and the destaging processing cannot continue. As shown in FIG. 1, the fault (fist 1) is recorded in the fault flag 3100 of the entry in the track table 3000 corresponding to the track where the fault has occurred. Also, 5CB
4000, the completion of the write operation to the magnetic disk device in which the failure occurred is recorded.

Thereafter, only one magnetic disk device for which the write operation has not been completed is selected from among the magnetic disk devices constituting the logical magnetic disk device, and the online destage process is performed again on this magnetic disk device as a synchronous magnetic disk device.

FIG. 7 is an operational flowchart in the case of a track failure during online destage processing according to the present invention, and FIG. 8 is a diagram showing a method for searching the track table and cylinder table according to the present invention.

If a fault is detected in a track of a magnetic disk device during online destage processing, first, it is determined whether the magnetic disk device is a magnetic disk device constituting a logical magnetic disk device (step 5200). When it is determined that the magnetic disk device constitutes a logical magnetic disk device, the occurrence of a failure is recorded in the entry on the track table 3000 corresponding to the track to which data is written (step 5201).

As shown in FIG. 8, the entries in the track table 3000 include a fault flag 3100 and an SCB address 3200.
When a track failure occurs, (*1) is recorded in this failure flag 3100.

When the recording of the track failure is completed, the 5CB400 reserved for processing write requests from the central processing unit 1 is
0, the completion of the write operation to the magnetic disk device in which the track failure occurred is recorded (step 5202). Thereafter, it is determined whether or not there is a magnetic disk device that is not writing data from the central processing unit 1 among the magnetic disk devices that constitute the logical magnetic disk device (step 5203).

As a result of this determination, if it is determined that there is a magnetic disk device in which no data write operation has been performed, the magnetic disk device in which no data write operation has been performed is selected and secured (
Step 5204). At the same time, the information related to the input/output request from the central processing unit 1 stored in the DCBlooO corresponding to the magnetic disk device in which the track failure occurred is transferred to the newly secured DCBlooO corresponding to the magnetic disk device.
Copy to CB 1000. Next, use the newly secured magnetic disk device as a synchronous processing magnetic disk device, and use the DC
Positioning processing is executed at the data write target position (position indicated by the seek address, sector number, and record ID) copied from the central processing unit 1 onto the B100O (step 5205). When the positioning process is completed, a request is made to the central processing unit 1 to retransfer the write data (step 5206). Thereby, it is possible to write data from the central processing unit 1 to a magnetic disk device that is not causing a failure.

On the other hand, if it is determined that the magnetic disk device in which the track failure has occurred does not constitute a logical magnetic disk device (step 5200), or if there is an unwritten write state in the magnetic disk device that constitutes the logical magnetic disk device. If it is determined that there is no magnetic disk device (step 5203), 5CB4'OOO secured for processing the data write request from the central processing unit 1 is made into an unused SCB (step 5207), and the central processing unit Report DSB including unit check for 1 (
Step 5208). When the central processing unit 1 receives the DSB including the unit check, it detects that the input/output request to the magnetic disk device has ended abnormally, and the subsequent C
Stop running CW and restart operating system 10.
Reports abnormal termination of I/O requests to.

In this way, even if a track failure occurs in some of the magnetic disk units constituting the logical magnetic disk unit, the operating system 10 on the central processing bag @1 can operate without being aware of the trunk failure. Processing can continue.

FIG. 9 is a processing flowchart when a track failure occurs during offline destaging processing.

When a track failure occurs during data transfer processing to the magnetic disk device during offline destage processing, the track table 30 corresponding to the track where the failure has occurred is
The occurrence of a failure is recorded in the entry 00 (step 5300), as in the online destage process. After the recording, the completion of data writing to the magnetic disk device is recorded in the 5CB 4000 (step 5301).

Next, when it is determined that the data write operation of all the magnetic disk units [6 constituting the logical magnetic disk unit] has been completed (step 5302), the 5CB400
0 to reusable SCB state (step 5303)
. That is, free the slot.

FIG. 10 is a processing flowchart for determining whether data to be accessed by the central processing unit exists on the disk cache memory of FIG. 2.

Below, the failure flag 310 on the track table 3000
A method of referring to the track failure information recorded in 0 will be explained. This truck fault information is.

When an input/output 8' @ request is issued from the central processing unit 1, this is referred to in the hit/miss determination process that determines whether or not the track targeted for the input/output process exists on the disk cache memory 24. be done.

In FIG. 10, the seek address specified by the 5EEK command from central processing unit 1 is stored in seek address 1200 on DCB 100O shown in FIG. In the hit/miss determination process, the track table entry corresponding to the seek address is read (step 5500).

This read operation is performed as follows. The start address 1100 of the cylinder table 2000 on the DCB 100O in FIG. 8 and the cylinder address 1210 in the seek address 1200 are added (5401) to obtain the cylinder table entry address. As a result of this addition, the cylinder table entry indicated by the obtained cylinder table entry address is obtained by adding the start address 2100 of the track table 3000 corresponding to the trunk included in that cylinder and the track address 1220 within the seek address 120°. By (5402)
, ask for the track table entry address. As a result, the track table entry indicated by the obtained track table entry address is read.

When a track table entry is read in this way, as shown in FIG. is occurring (step 5501)
). As a result of this determination, if it is determined that Access 1 to Rack are inaccessible due to a failure, it is determined whether or not there is an unselected magnetic disk device among the logical magnetic disk devices targeted for input/output processing. (Step 5502
). As a result of this determination, if it is determined that there is an unselected magnetic disk device, one magnetic disk device is selected from the unselected magnetic disk devices and secured (step 5503). Using the newly secured magnetic disk device, go back to the beginning, read the track table entry, and repeat the same process again (step 5).
500). This makes it possible to select a magnetic disk device that has access tracks that can be accessed normally. If a magnetic disk device with access tracks that can be accessed normally is selected, the SCB shown in FIG.
Determine the value of address 3200 (step 5504)
. As a result of the determination, if a NULL value is set in the SCB address 3200, it is determined that the access track does not exist on the disk cache memory 24. In other words, it is determined to be a mistake (step 5506).

At the same time, its status is recorded on the DCB.

On the other hand, when it is determined that the scB address is stored in the 5CB7 address 32oo (step 5504
), it is determined to be a hit (step 5505), and its status is recorded on the DCB.

Furthermore, if it is determined that there is no magnetic disk device that has access tracks that can normally access all the magnetic disk devices that make up the logical magnetic disk device (step 5502), the input/output from the central processing unit 1 is Since the request cannot be executed normally, the DSB including the unit check is reported to the central processing unit 1 (step 5507).
).

This allows the magnetic disk controller 2 to automatically select the magnetic disk device in which the data requested to be transferred by the central processing unit 1 is correctly stored.

Next, the operation of the magnetic disk $J device 2 when a data read request is issued from the central processing unit 1 will be described in detail.

FIG. 11 is a diagram showing a group of channel commands and their parameters prepared by the operating system when reading data on a magnetic disk device.

In FIG. 11, 5SCH is an input/output activation request instruction 11 executed by the operating system 10 on the central processing unit 1, and 12 is a control block when issuing the input/output activation request.

14 is a CCW group for performing a data reading operation for the magnetic disk device, and 210 to 214 connected to these CCWs by pointers are all CCW parameters. Of the CCW group 14, the operations up to the TIC command 204 are the same as those at the time of the data write request described in detail above. That is, TIC command 20
By the time the next READ DATA command 205 of No. 4 is executed, the magnetic disk device in which the data of the access track is correctly stored is selected as the synchronous processing magnetic disk device, and the record to be accessed is positioned and prepared. In this case, if there is a hit, the record is located at the record position to be accessed on the disk cache memory 24, and if it is a miss, the record is located at the record position to be accessed on the magnetic disk device. In this state, the central processing unit 1 issues a READ DATA command 2.
05, the magnetic disk control device 2 transfers the record located on the disk cache memory 24 in the case of a hit, and the record located on the magnetic disk device in the case of a miss to the central processing unit 1. Transfer to.

Here, in the case of a miss, the record on the magnetic disk device is added to the central processing unit 1 and the disk cache memory 24
However, the process to be performed when a failure occurs during this transfer will be explained below.

FIG. 12 is a processing flowchart when a track failure occurs when reading data on a magnetic disk device.

If a failure occurs while transferring records on a magnetic disk device to the central processing unit 1, first it is determined whether the failed magnetic disk device constitutes a logical magnetic disk device (step 5600). As a result of the determination, if it is determined that the magnetic disk device constitutes a logical magnetic disk device, the failure flag 3100 in the entry of the track table 3000 corresponding to the track to which the accessed record belongs is set. , record the track failure (step 5601). Thereafter, it is determined whether or not there is a magnetic disk device among the logical magnetic disk devices to which the magnetic disk device belongs, which has not yet been selected in response to an input/output request from the central processing unit 1 that is currently processing. (Step 5602). As a result of the determination, if it is determined that there is an unselected magnetic disk device, one magnetic disk device is selected from the unselected magnetic disk devices and reserved (step 5603). At this time, information regarding the currently executing input/output request from the central processing unit 1 is copied from the contents of the DCB corresponding to the failed magnetic disk device to the DCB corresponding to the newly secured magnetic disk device. Then, positioning processing is executed for the newly secured magnetic disk device (step 5604). Next, the read data is retransferred to the central processing unit 1 (step 5605).

On the other hand, when it is determined that the failed magnetic disk device does not constitute a logical magnetic disk device (step 56
00), or when the unselected magnetic disk device is not among the logical magnetic disk devices (step 5602),
5CB400 that manages slots on the disk cache memory 24 that stores records on the magnetic disk device
0 as an unused state (step 5605), and reports the DSB including the unit check to the central processing unit 1 (step 5607), thereby notifying that the input/output request could not be executed normally.

In the above embodiment, when a failure occurs in all the magnetic disk units that make up the logical magnetic disk unit, the failure is reported to the central processing unit 1. However, when the number of failed magnetic disk units exceeds the specified value, In this case, a failure or warning may be reported to the central processing unit 1.

Note that if the directory memory 25 in which the track table 3000 for recording fault information is arranged is a volatile memory, the fault information is also recorded on the magnetic disk device. As a result, even if a power failure occurs, failure information is retained.

If the directory memory 25 is a non-volatile memory, this fault information is transferred to the service processor 3 (see FIG. 1) when the power to the magnetic disk control device 2 is turned off. records this failure information in the floppy disk device 4. Then, when the power is turned on again, the service processor 3
reads the fault information from the floppy disk device 4.

Track table 3000 in directory memory 25
The failure flag 3100 of is initialized.

In this way, it is possible to perform multiple write operations on the magnetic disk drive, and even if a failure occurs in part of the multiplexed data, only the magnetic disk drive can be controlled without the intervention of the operating system. After automatically selecting a magnetic disk drive in which normal data is stored, data can be correctly read from the magnetic disk drive.

FIG. 13 is a block diagram of a data multiplex write control system showing a second embodiment of the present invention, in which a magnetic tape device is used as the auxiliary storage device.

In the computer system of FIG. 13, the magnetic disk control device 2 is replaced with a magnetic tape control device 7 compared to that of FIG.
The magnetic disk devices 31 to 33 are magnetic tape devices 51 to 5.
3, the only difference is that the directory memory 25 is changed to a data management memory 75, and a floppy disk device is not connected to the service processor 5.

Other than that, the configuration is the same.

The central processing unit 1 is connected to a magnetic tape controller 7 via a channel interface.

Furthermore, the magnetic tape control device 7 is connected to magnetic tape devices 51, 52, 53, etc. via a drive interface.
・Connected to a group of magnetic tape devices. A data buffer memory 74 for storing data on the magnetic tape device is connected to the data bus 73 connected to the channel interface and the drive interface. The magnetic tape control device 7 includes a channel side control section 7o for controlling data transfer with the central processing unit 1, and a drive side control section 72 for controlling data transfer with the magnetic tape device.
It is controlled by two control units. These control units 70
．． 72 is connected to a control memory 71 for controlling the entire magnetic tape control device 7 and a data management memory 75 for managing data on the data buffer memory 74.

As in the case of a magnetic disk device, in this embodiment as well, by setting a logical magnetic tape device, it is possible to write and read data from the central processing unit 1 to one logical magnetic tape device. In practice, the same data is written to or read from multiple magnetic tape devices. Here, the magnetic tape control device 7 acts on the operating system 10 running on the central processing unit 1 as if the magnetic tape devices 51, 52 were one magnetic tape device.

The correspondence between logical magnetic tape devices and magnetic tape devices is as follows:
Settings are made by instructions from the operating system 10 or the service processor 3, and information for the settings is recorded in the control memory 71 and the recording medium of the magnetic tape device. The configuration information of the logical magnetic tape device recorded on the recording medium of the magnetic tape device is automatically stored in the control memory 71 when the recording medium is attached to the magnetic tape device.

A magnetic tape device 51 that constitutes a logical magnetic tape device 50 from the operating system 1o on the central processing unit 1
Or if a write request is made to 5L52, all magnetic tape devices [(5L52
) is written with the same data. This results in
The dual write function of a magnetic tape device can be realized. Although this embodiment shows an example of a logical magnetic tape device consisting of two magnetic tape devices, it is of course possible to expand the logical magnetic tape device to three or more.

FIG. 14 is a related diagram of the control table in the magnetic tape control device in FIG. 13.

Device control block
1B 1ock (hereinafter referred to as DCB) 6000-a,
6000-b is stored in the control memory 71, exists in each magnetic tape device controlled by the magnetic tape control device 7, and constitutes a control block for executing input/output operations for the magnetic tape device. The DCBs that control the magnetic tape devices constituting the logical magnetic tape device are connected to each other by pointers. Furthermore, each DCB entry is connected to a hash table 7000 on the data management memory 75 by a pointer.

The hash tables 7000-a and 7000-b are tables for determining whether or not data on the magnetic tape device exists on the data buffer memory 74. A control block for managing stored data, that is, a block control block (B 1ock Control
B 1ock: Hereinafter referred to as BCB) 8000 address 7200 and a failure flag (hereinafter referred to as BCB) that stores information indicating that data on the magnetic tape device cannot be read correctly.
-)7100.

The operating system 10 on the central processing unit is
When a data write request is made to the magnetic tape device 51, the magnetic tape control device 7 accepts this input/output request, stores the subsequently transferred write data in the data buffer memory 74, and then sends it to the central processing Reports completion of input/output to device 1. After that, the magnetic tape control device 7 independently writes the data on the data buffer memory 74 to the magnetic tape device 51 and the magnetic tape device 52 asynchronously with the central processing unit 1. In this case, confirm that recording media are installed in all the magnetic tape devices that make up the logical magnetic tape device, and if they are not installed.

The operating system 10 on the central processing unit 1 is requested to attach a recording medium, and after all the magnetic tape devices constituting the logical magnetic tape device are attached, a data write operation is executed. Magnetic tape control device 7
In order to manage the data on the data buffer memory 74, write request units (
This is called a block) and is managed by referring to the BCB 8000. In the BCB8000, an area is secured when a data write request is issued from the central processing unit 1 to the magnetic tape device, and the sequence number assigned to the block, that is, the first block of the magnetic tape device, is set to 1.
It is pointed to from the entry on the hash table 7000 obtained by the hash function using the sequence number as a key. When data is written to all the magnetic tape devices that make up the logical magnetic tape device, the data on the data buffer 74 is deleted, and at the same time, BCB 8000 is invalidated, and at the same time, the address of BCB 8000 is deleted from the entry on the hash table 7000. . This realizes multiple write processing for the magnetic tape device.

Next, when a data read request is issued from the central processing unit 1 to the magnetic tape devices that make up the logical magnetic tape device, first any one of the magnetic tape devices that make up the logical magnetic tape device is read. selected. For this selected magnetic tape device, the entry 1 of the hash table 7000 is referred to using the sequence number assigned to the block as a key. If the address of the BCB 8000 is set in this entry, the write request previously issued by the central processing unit 1 has not yet been reflected in all the magnetic tape devices that make up the logical magnetic tape device.
In other words, since it has been determined that writing has not been done to all of the specified plurality of magnetic tape devices, the operation of those devices is terminated first. That is, first, unreflected data is written to all the magnetic tape devices constituting the logical magnetic tape device, and then the data read request issued this time from the central processing unit 1 is executed by the previously selected magnetic tape device. Additionally, if the entry in the hash table 7000 indicates that the block cannot be read due to a failure, select an unselected magnetic tape device from among the magnetic tape devices that make up the logical magnetic tape device. , executes a data read operation using the magnetic tape device.

If a failure occurs in the data read operation, the failure is recorded in the failure flag 7100 of the entry on the hash table 7000 corresponding to the block to be read. Thereafter, an unselected magnetic tape device is selected from among the magnetic tape devices constituting the logical magnetic tape device, and a data read request from the central processing unit 1 is executed. In addition.

In the above case, when selecting an unselected magnetic tape device, there may be cases where no recording medium is attached to any of the magnetic tape devices that make up the logical magnetic tape device. At that time, the magnetic tape control device 7 instructs the operating system 10 of the central processing unit 1 to attach a recording medium to the magnetic tape device constituting the logical magnetic tape device. Then, when the recording medium is attached to the magnetic tape device, the failure information is read from the recording medium and developed on the hash table 7000, and the central processing unit 1
Executes a request to read data from. Note that writing of the failure information to the recording medium is executed when the recording medium is removed.

As a result, it is possible to read multiplexed data from a plurality of magnetic tape devices.

As described above, in each embodiment of the present invention, when the auxiliary storage control device automatically performs multiple writing of data in the auxiliary storage device, the auxiliary storage control device Since the data is stored in the control memory of the device, by referring to this information, it is possible to select the auxiliary storage device in which the data is correctly stored. In addition, since information on locations where data is not stored correctly in the auxiliary storage device is also recorded in the auxiliary storage device itself, it is also possible to store information on auxiliary storage devices that have removable recording media, such as magnetic tape devices. , data multiplexing can be automatically executed in the control device.

〔Effect of the invention〕

As explained above, according to the present invention, correct data can be automatically transferred to the central processing unit by the magnetic disk control device and the magnetic tape control device without the intervention of the operating system, and the correct data can be automatically transferred to the central processing unit without the intervention of the operating system. It is possible to automatically multiplex write data not only to an auxiliary storage device having a recording medium but also to an auxiliary storage device having a removable recording medium.

[Brief explanation of the drawing]

FIG. 1 is a diagram related to a control table in a magnetic disk control device according to the present invention, and FIG. 2 is a block diagram of a magnetic disk control device showing an embodiment of the present invention. FIG. 3 is a processing flowchart when the magnetic disk device receives an input/output request from the central processing unit. Flowchart of the processing, FIG. 5 is a diagram showing the channel command group and its parameters prepared by the operating system when writing data on the magnetic disk device in FIG. 2, and FIG. 6 is a diagram showing the channel command group and its parameters on the disk cache memory in FIG. FIG. 7 is a processing flowchart for securing an area, and FIG. The figure shows how to search the track table and cylinder table in Figure 1, and Figure 9 shows how to search for the track table and cylinder table in Figure 1.
FIG. 10 is a processing flowchart when a failure occurs in a track on a magnetic disk device. FIG. 11 is a diagram showing a group of channel commands and their parameters prepared by the operating system when reading data from the magnetic disk device in FIG. 13 is a block diagram of a magnetic tape control device showing another embodiment of the present invention; FIG. 14 is a flowchart of the process when a track failure occurs while reading the above data; FIG. 14 is a block diagram of the magnetic tape control device of FIG. FIG. 2 is a configuration diagram of a table for controlling. 1: Central processing unit, 2: Magnetic disk control head, 3.5
: Service processor, 7: Magnetic tape control device, 10
: Operating system, 20° 70: Channel side control unit (processor), 22° 72 Drive side control unit (processor), 21° 7: Control memory, 24: Disk cache memory, 25: Directory memory, 31
~33: Magnetic disk device, 30: Logical magnetic disk device, 1000-a, 1000-b: Device control block, 1100, 1500, 1600, 1700° 180
0: pointer, 2000-a, 2000-b = cylinder table, 3000-a, 3000-b = track table, 4000-a, 4000-b = slot control table, 51 to 53: magnetic tape device, 50: logic magnetic Tape device, 74: Data buffer memory, 75: Data management memory, 6000-a, 6000-b: Device control block, 7000-a, 7000-b: Hash table, 8000-a, 8000-b Niblock control block, 3100.7100: Failure flag. Figure Figure Figure Figure

Claims (1)

[Scope of Claims] 1. An auxiliary storage control device connected to a central processing unit via a channel interface and connected to a plurality of auxiliary storage devices via a drive interface; When a plurality of auxiliary storage devices combined in advance are recognized as one auxiliary storage device (hereinafter referred to as a logical auxiliary storage device) and a data write request is generated from the central processing unit to the logical auxiliary storage device, the logical auxiliary storage device One auxiliary storage device is selected from the combinations that make up the device, data is written to the auxiliary storage device and the data buffer in the auxiliary storage control device, and when the writing is completed normally, the central In an auxiliary storage control device that reports normal write completion to a processing device and then writes data to the remaining auxiliary storage devices combined as a logical auxiliary storage device,
An auxiliary storage device characterized in that when a failure occurs while data is being written to the auxiliary storage device, the location of the write failure is stored in each auxiliary storage device of the memory in the auxiliary storage control device. data multiplex writing control method. 2. In the data multiplex writing control method for the auxiliary storage device according to claim 1, when a write request is made from the central processing unit, a failure occurs while writing to one selected auxiliary storage device; If normal writing is not possible in the auxiliary storage device, the location of the failure is recorded in each auxiliary storage device of the memory in the auxiliary storage controller, and then the auxiliary storage device that constitutes the logical auxiliary storage device other than the auxiliary storage device where the write failure occurred. Select one auxiliary storage device from the group, write the same data to the auxiliary storage device and the data buffer in the auxiliary storage control device, and when the operation is completed normally, the central A data multiplex write control method for an auxiliary storage device characterized by reporting the completion of normal writing to a processing device. 3. In the data multiplex writing control method for an auxiliary storage device according to claim 1, M auxiliary storage devices (of the N auxiliary storage devices constituting the logical auxiliary storage device) are A data multiplex write control method for an auxiliary storage device, characterized in that when a write operation cannot be performed normally (N≧M), the auxiliary storage control device reports an abnormality in the auxiliary storage device to the central processing unit. 4. In the data multiplex write control method for an auxiliary storage device according to claim 1, when a read request is issued from the central processing unit and the data of the read request exists in the data buffer in the auxiliary storage control device, The data in the data buffer is transferred to the central processing unit, and when the target data does not exist in the data buffer, the target data is written normally by referring to the information on the fault location of the memory in the auxiliary storage control device. 1. A data multiplex write control method for an auxiliary storage device, characterized in that one auxiliary storage device is selected, data is read from the auxiliary storage device, and the data is transferred to the central processing unit. 5. In the data multiplex writing control method for the auxiliary storage device according to claim 4, a failure occurs while reading data from the selected auxiliary storage device due to a data read request from the central processing unit, and therefore the normal operation is not performed. If the failure is not completed, the location of the failure is recorded for each auxiliary storage device in the memory in the auxiliary storage control device, and the auxiliary storage device in which the read data has been correctly written is selected from among the combinations, and the auxiliary storage device is A data multiplex write control method for an auxiliary storage device characterized by reading target data from the device. 6. In the data multiplex writing control method for an auxiliary storage device according to claim 4 or 5, in response to a data read request from the central processing unit, normal read data from among the N auxiliary storage devices constituting the logical auxiliary storage device is detected. A data multiplex write control method for an auxiliary storage device, characterized in that when there are M auxiliary storage devices (N≧M) that do not have an auxiliary storage device, an abnormality in the auxiliary storage device is reported to the central processing unit. . 7. In the data multiplex write control method for an auxiliary storage device according to claim 1 or 4, the writing-time failure occurrence position and the reading-time failure occurrence position are written in a storage medium of the auxiliary storage device, and the storage medium is attached. A data multiplex write control method for an auxiliary storage device, characterized in that when a failure occurs in the memory in which the failure location information is written in the auxiliary storage control device, the failure information written in the storage medium is read. 8. In the data multiplex write control method for an auxiliary storage device according to claim 1 or 4, the memory in the auxiliary storage control device for recording the location of failure during writing and the location of failure during reading is a non-volatile memory. A data multiplex write control method for an auxiliary storage device characterized by comprising a memory. 9. In the data multiplex writing control method for an auxiliary storage device according to claim 1, 4, 5 or 7, the memory in the auxiliary storage control device includes a cylinder and a track forming the cylinder as an area of the auxiliary storage device. By referring to the table, it is possible to determine whether the data on the auxiliary storage device exists on the data buffer in the auxiliary storage control device and whether the data is normal or not. A multiple write control method for an auxiliary storage device, characterized in that it determines.