JPH0926857A - Disk write guarantee method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To shorten a time exclusively used by disk write by judging the case that the write of all the data on a disk device is completed as normal completion and writing new data of a remaining part by changing the address of start data when the write of only a part of data is completed. SOLUTION: A write target range group is decided from file areas managed by host processors 10, 20, and write guarantee processes 11, 12 or 12, 21 are operated. The write target range group is started to write on the disk device 40 from a cache part 32 starting from a first request range sequentially. When the write operation of the disk device 40 set for one time of processing is completed. the case that the write of all the data on the disk device 40 is completed is judged as the normal completion, and in the case of partial completion in which the write of all the data is incomplete, the write operation of the new data of the remaining part is performed by changing the address of the start data.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk write guarantee method in a computer system including a disk processing device equipped with a cache memory, and more particularly to writing data written in the cache memory of the disk processing device into the disk device. At this time, the present invention relates to a disk write guarantee method for guaranteeing that data writing is surely performed.

[0002]

2. Description of the Related Art In a computer system including a disk processing device equipped with a cache memory, data to be written from the host processor to the disk device is not directly written from the host processor to the disk device, but is temporarily cached by the disk processing device. An operation of writing data in the cache memory of the disk processing device to the disk device is performed when the disk processing device and the disk device are not operating according to the input / output command. The data writing operation (writing guarantee operation) to such a disk device is performed at the time of switching from the online processing operation of the computer system to the badge processing operation, and at the time of saving the data from the update badge processing operation that performs the data update processing. It is generally performed at the time of switching to the maintenance-type badge processing operation to be performed, and the write guarantee operation is prioritized over the data write or read operation for a normal disk device.

A first method as a conventional disk write guarantee method for ensuring that the operation of writing the write data to the disk device from the cache memory of the disk processing device to the disk device as described above is surely performed. As a result, the host processor issues a command to the disk processing device in a batch to the write guarantee target range group (the range of addresses that specify the data to be written to the disk device from the disk processing device), and the disk processing device receives the write guarantee target range according to the command. A method is adopted in which after writing all the data in the group to the disk device, the command is ended.

As a second method, the host processor divides the write guarantee target range group into a plurality of ranges and gives an instruction to the disk processing apparatus, and the disk processing apparatus individually sets each of the plurality of divided ranges as one unit. A method is adopted in which a write operation is performed, and finally all the data within the write guarantee target range group is written to the disk device, and then the command is ended.

[0005]

In the above-described conventional disk write guarantee method, in the first method, since the occupation time of the disk processing device and the disk device (resource) associated with the execution of the command becomes long, the same method is used. However, there is a disadvantage that the response to a normal read command or write command using the resources of the above is delayed.

In the second method, the host processor can adjust the occupation time of the resource accompanying the execution of a command. However, when one resource is connected to a plurality of host processors, commands from the plurality of host processors are equally distributed. However, there is a problem that it is inappropriate to dynamically optimize the load on resources to be executed.

An object of the present invention is to eliminate the above-mentioned drawbacks of the conventional disk write guarantee method, shorten the resource occupation time associated with the execution of a disk write command, and perform dynamic optimization. It is an object of the present invention to provide a disk write guarantee method which can be performed.

[0008]

According to the present invention, there is provided a disk write guarantee method comprising the steps of:
The guarantee process of the host processor is operated to specify a write target range group, and a write guarantee command is issued to the disk processing device, and the disk processing device sequentially caches from the first requested range of the write target range group. Section starts writing to the disk device, and the write operation time of the disk device set for a write operation of a data amount adjusted according to the load status of the disk processing device and the disk device or for one-time processing When all the data in the requested range have been written to the disk device at that time, the processing ends normally, and only part of the data in the requested range has been written to the disk device. In the case of abnormal termination, the abnormal termination is due to a malfunction of the disk processing device or the disk device. When it is the end, it ends as an abnormality of the computer system, and when the abnormal end is a partial end in which only part of the data in the requested range has been written to the disk device, the requested range of the write guarantee command This includes changing the address of the start data and performing a data write operation with the remaining portion as a new request range.

[0009]

Next, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a flowchart showing an embodiment of the present invention, and FIG. 2 is a block diagram showing an example of a computer system to which the embodiment of FIG. 1 is applied.

The computer system shown in FIG. 2 has two host processors 10 and 20 with a disk processing unit 30.
Is connected, and the disk processing device 30 controls the operation of the disk device 40.

The host processors 10 and 20 can simultaneously operate a plurality of write assurance processes (write assurance operation programs) 11 and 12 or write assurance processes 21 and 22, respectively. A command is sent to the disk processing device 30 via the input / output control unit 13 and the input / output interface 14. Similarly, the write assurance process 21
And 22 send commands to the disk processing device 30 via the input / output control unit 23 and the input / output interface 24. The disk processing device 30 sends an execution result of the command to the host processor 10 via the input / output interface 14 and the input / output control unit 13, and executes the command to the host processor 20 via the input / output interface 24 and the input / output control unit 23. Send results.

When the second input / output command is issued during execution of the first input / output command for a specific disk device, the input / output control units 13 and 23 hold the second input / output command. After the execution of the first input / output command is completed, the second input / output command being held is sent to the disk processing device 30.

The disk processing unit 30 includes an input / output execution unit 3
1, a cache unit 32, and a load monitoring unit 33, and the input / output execution unit 31
Or, when another input / output command is issued via another input / output interface 24 or 14 while executing the input / output command issued via 24, the other input / output command is suspended. After the execution of the previous input / output command is completed, another pending input / output command is executed.

The cache unit 32 includes the host processor 1
0 or 20 is issued to the disk device 4
It has a write cache function for terminating 0 without operating it. The write operation by the write cache function is performed by the disk processing device 30 when the latest write data held in the cache unit 32 is not operated by the disk processing device 30 and the disk device 40 according to the normal input / output command. This is an operation of writing to the disk device 40.

The load monitoring unit 33 includes a disk processing unit 30
And the service time per unit time in the disk device 40 is monitored, and the load status of each unit in the disk processing device 30 and the disk device 40 is notified to the input / output execution unit 31 in a timely manner.

Next, the operation of this embodiment will be described with reference to FIG.

A user job is input to the computer system or the host processor 10 or 20
When the server voluntarily starts the operation, first, the write target range group (1 to N groups) is determined from the file area specified by the user job or the file area managed by the host processor 10 or 20 (step 51), and the write guarantee is performed. The process 11 or 12 or 21 or 22 is operated (step 52).

The write guarantee process 11 or 12 or 21 or 22 sets the index n of the write target range group to n.
= 1 (step 53), and the range up to the n-th writing target range group is set as the required range (writing range) (step 54). Next, a write assurance command using the requested range as a parameter is issued to the disk processing device 30, and the end of the input / output operation (I / O end) in the disk processing device 30 is waited for (step 55).

Next, the input / output execution unit 31 of the disk processing unit 30 inputs the load status of each unit in the disk processing unit 30 from the load monitoring unit 33. If the load status is light, the processing is performed in one processing operation. The amount of data to be processed is increased, and when the load condition is heavy, adjustment is performed so that the amount of data to be processed in one processing operation is reduced, or the write operation time in the disk device 40 for one processing. To determine.
The input / output execution unit 31 terminates the data processing of the processing amount determined in this way, or when the write operation time has elapsed (step 56), the cache unit 3
When the writing of all the data in the request range 2 to the disk device 40 is completed, it is assumed that the processing of the command is completed normally (normal end) (step 57), and only a part of the data in the request range is set. When the writing from the cache unit 32 to the disk device 40 is not completed, it is assumed that the processing of the command is abnormally ended (abnormal end) (step 58). If this abnormal termination occurs, step 56
The partial end is used as information for identifying whether the input / output operation end (I / O end) in (1) is due to a partial end or due to a malfunction of the disk processing device 30 or the disk device 40. At the end, the address of the data that has been written to the disk device 40 is included. Such I / O end information is returned to the write assurance process 11 or 12 or 21 or 22 that issued the write assurance instruction via the input / output control unit 13 or 23.

If the I / O end in step 56 is a normal end (step 57), the write assurance process 11 or 12 or 21 or 22 proceeds to step 60, sets the index n to n = n + 1, and sets the index n to n. The operation from step 54 to step 60 is repeated until N becomes N (step 61). I / O in step 56
If the end is due to a malfunction (step 5
8) The processing ends abnormally assuming that an abnormality has occurred in the computer system (step 63). I in step 56
If the / O end is an end due to a partial end (step 58), the process proceeds to step 59, and the address of the start data of the range requested by the command is set to the address of the data for which writing to the disk device 40 is completed in step 56. By replacing the value with (+1), the range obtained by removing the already processed data from the requested range is set as a new requested range and the process returns to step 54.

[0022]

As described above, according to the disk write guarantee method of the present invention, when the computer system is started, the guarantee process of the host processor is operated to designate a write target range group and write the data into the disk processing device. The disk processing device issues a guarantee command, and the disk processing device starts writing from the cache unit to the disk device sequentially from the first requested range of the write target range group, and performs a write operation of the data amount adjusted according to the load condition or 1 When the write operation time of the disk device set for the same process ends, if writing of all data in the requested range to the disk device has been completed at that time, it is regarded as normal end, and a part of the requested range is If only data has been written to the disk drive, the start data address of the command request range is Further and by performing the data write operation to the remaining portion as a new request range, there is an effect that it is possible to dynamically optimize the occupation time of a resource associated with the execution of the disk write instruction.

[Brief description of drawings]

FIG. 1 is a flowchart showing one embodiment of the present invention.

FIG. 2 is a block diagram illustrating an example of a computer system to which the embodiment of FIG. 1 is applied.

[Explanation of symbols]

 10.20 Host Processor 11.12.21.22 Write Assurance Process 13.23 I / O Control Unit 14.24 I / O Interface 30 Disk Processing Unit 31 I / O Execution Unit 32 Cache Unit 33 Load Monitoring Unit 40 Disk Units 51-63 Steps

Claims (2)

[Claims] 1. When a computer system is started, a guarantee process of a host processor is operated to specify a write target range group and a write guarantee command is issued to a disk processing device, and the disk processing device is configured to write the write target command. Writing from the cache unit to the disk device is sequentially started from the first requested range of the range group, and a write operation of a data amount adjusted according to the load status of the disk processing device and the disk device or one time When the write operation time of the disk device set for processing is completed, if all the data in the requested range has been written to the disk device at that time, it is regarded as a normal end and part of the requested range. In the case of the partial end in which only the data of is written to the disk device,
A disk write guarantee method comprising changing the address of the start data of the request range of the write guarantee command and performing a data write operation with the remaining part as a new request range. 2. When a computer system is started, a guarantee process of a host processor is operated to specify a write target range group and a write guarantee command is issued to a disk processing device, and the disk processing device is configured to write the write target command. Writing from the cache unit to the disk device is sequentially started from the first requested range of the range group, and a write operation of a data amount adjusted according to the load status of the disk processing device and the disk device or one time When the write operation time of the disk device set for processing is completed, if all the data in the requested range has been written to the disk device at that time, it is regarded as a normal end and part of the requested range. In the case of abnormal termination in which only the data of the above has been written to the disk device,
When the abnormal end is due to a malfunction of the disk processing device or the disk device, the abnormal end of the computer system ends, and the abnormal end ends the writing of only a part of the data in the requested range to the disk device. The disk writing is characterized in that, at the end of the portion which has not been written, the address of the start data of the request range of the write guarantee command is changed to perform the data write operation with the remaining part as the new request range. Guarantee method.