JPH0242523A - Disk controller, difference back-up method and computer system - Google Patents

Abstract

PURPOSE:To add an application propriety flag of an update data control table B to a nonvolatile memory and to reduce the back-up value by preparing an update data control table A for each block using an update presence/absence flag in a volatile memory and adding the table B for each track using the update presence/absence flag to the nonvolatile memory respectively. CONSTITUTION:The higher rank systems 101, 102, 115 and 116 give the register and update requests to a disk controller 103 for the data stored in a disk device 114. These requests are received by a microprocessor 106 of the controller 103 via a channel interface 104. Then the processor 106 sets the contents of the requests at a data buffer 105. The data stored in the device 114 are updated via a device interface control part 113. The bit of an update data control table A108 set in a volatile memory 107 corresponding to the due position in the device 114 is set in an update presence state. In the same way, the corresponding bit of an update data control table B110 set in a nonvolatile memory 109 having a nonvolatile mechanism 112 is set in an update presence state respectively.

Description

[Detailed Description of the Invention] [Industrial Application Field] An economical and backup method that manages a management table of updated portions of data in an external storage device such as a magnetic disk by dividing it into volatile memory and non-volatile memory. The present invention relates to a disk control device for a disk device, typified by a magnetic disk device, which can significantly reduce the amount of storage.

[Conventional technology]

In the prior art, as described in Japanese Patent Laid-Open No. 57-90770, a control memory is provided in a magnetic disk device or the like to record the location of data that has been changed during job execution.
During data backup, the host system reads the contents of the control memory and issues a data read request for the updated portion to a magnetic disk device, etc. h
It was a ceremony.

[Problem to be solved by the invention]

The above-mentioned conventional technology does not consider the management unit of the update portion in the update data management table and the economic efficiency of the memory that stores it. In order to reduce the amount of backup, it is preferable to manage updates in smaller units, such as blocks rather than cylinders, and records rather than blocks. If this is realized, the cost of equipment such as batteries for non-volatization is high, and the capacity cannot be increased economically. For example, in the case of a control memory in a magnetic disk control device that controls eight magnetic disk devices with a capacity of 15 gigabytes, the size of the update data management table is managed by 1 bit of memory per track of 40 kilobytes. This results in a control memory capacity of 3 mechanical pits.

If this is further managed in units of 4 kilobytes per block, 30 megabits will be required.

An object of the present invention is to realize a disk control device that takes into consideration reduction in backup volume and economic efficiency.

[Means to solve the problem]

The above purpose is to provide an update data management table A in volatile memory that manages updated portions in units of blocks, consisting of cylinder numbers, track numbers, block numbers, and update/non-update flags.
An update data management table B for each track consisting of a cylinder number, track number, and an update flag, and a usability flag for the update data management table B are provided and managed in non-volatile memory, and the update data management table B from the host system is managed. This is achieved by providing a means for selecting the updated data management tables A and B based on the above-mentioned flags in response to a transfer request, and transferring the selected update data management tables to the host system.

[Effect]

(1) The first invention can be realized by operating the above means as follows. See Figure 1.

(a) Upper system (101, 102, 115°116
When a request for registering or updating (hereinafter collectively referred to as update) data in the disk device 114 is issued to the disk control device 103 from a computer (generic term for , set the request contents in the data buffer.

Next, data in the disk device 114 is updated via the device interface control unit 113.

Thereafter, an update data management table Al in the volatile memory 107 corresponding to the location of the update data in the disk device (specified by cylinder number, track number, block number)
Set the O3 bit to an updated state. Similarly, a non-volatile memory 1. has a non-volatile mechanism 112 such as a battery.
The corresponding bit in the update data management table 8110 in 09 is also set to an updated state.

(b) When the host system sends a request to transfer the update data management table to the disk controller, the microprocessor checks the availability flag 111 for the update data management table A in the non-volatile memory and determines whether the update data management table A can be used or not. If so, the updated data management table A, in which updated portions are managed in block units, and information indicating the distinction between the management tables are transferred to the higher-level system. If it cannot be used, the update data management table B, in which updated portions are managed on a track-by-track basis, and information indicating the distinction between the management tables are transferred to the higher-level system.

(2) In the second invention (1) (b), in response to a request to transfer only update data from the host system, the update data management table A is stored in the volatile memory if it is available, and in the non-volatile memory if it is not available. This can be achieved by adding means for reading update data from the disk device and transferring it to the higher-level system based on update management table B in the table.

(3) The third invention is (1) (b), (2)
This can be achieved by adding a means to write the update data read based on the same judgment as above to the disk device specified by the host system.

(4) The fourth invention is (1) (b), (2)
This can be achieved by adding a means to overwrite the update data read with the same judgment as above on the data of the same cylinder number, track number, and block number of the disk device specified by the host system.

(5) In the fifth invention, when there is a request to save update data management table A or B from the higher level system, the higher level system saves the updated data management table A in the volatile memory or the updated data management table B in the nonvolatile memory. This can be realized by adding a writing means to the specified disk device. Similarly, a recovery request can be realized by adding a means for reading an update data management table from a disk device specified by a host system and setting it in volatile or nonvolatile memory.

(6) In the sixth invention, in (5), when the update data management table A is restored, the update data management table B in the nonvolatile memory is
This is achieved by adding and operating a means to automatically regenerate the update data management table B for each track from the update data management table A that manages the updated part in units of blocks if the update data management table A has been initialized. can.

(7) In the seventh invention, the differential backup processing program existing in the host system identifies update data from the update data management table A based on the type of management table transferred from the disk control device according to claim 1. This can be achieved by selectively using two processes: the process of identifying the update data from the update data management table B and generating the read command.

(8) The eighth invention provides update data management from a host system in the process of bringing the disk control device online according to claims 1, 2, 3, and 4, which is instructed by a user via a console or the like. Issuing the recovery request (5) or (6) for table A to the disk controller; similarly, (5) for update data management table A from the upper system in response to an offline instruction from the user. This can be achieved by issuing an evacuation request to the disk controller.

〔Example〕

An embodiment of claim 1 of the present invention will be described below with reference to FIG.

FIG. 1 is a diagram showing the overall configuration of the present invention.

101 is a CPU (Central Process).
ing Unit: central processing unit), 102 is a channel, 103 is a disk control device represented by a magnetic disk control device, and 104 is a channel for exchanging data between the channel 102 and the disk control device 103 represented by a magnetic disk device. The interface control unit 105 is a data buffer that temporarily stores data.

106 is a disk controller i! 107 is a volatile memory that stores an update data management table AlO3 that manages updated portions of data in the disk device 114 after a certain point in block units;
Similarly, 09 is an update data management table 8110 for each track.
112 is a nonvolatile memory for storing the availability flag 111 of update data management table A, and 112 is a nonvolatile mechanism for the nonvolatile memory 109 composed of a battery, etc.;
113 is a disk control device 103 and a disk device 114
115 is the main storage, 116 is a program that issues requests to update data in the disk device, requests to read the updated data management table, etc. In this embodiment, 101
, 102, 115, and 116 are collectively called the upper system.

The microprocessor 106 in the disk controller 103 operates as shown in FIG. 2 in response to a data update request, an update data management table read request, and an update data management table initialization request from the host system.

If it is a data update request (201), it requests the device interface control unit 113 to write update data to the disk device 114 (202).

It is assumed that the cylinder number, track number, and block number indicating the write location of the update data are passed along with the update data from the host system, as in the prior art. next,
Update data management table BII corresponding to the cylinder number, track number, and block number in the disk device of update data
The bit of O is set to the updated state (203), and the contents of the usability flag 111 of the update data management table A are determined (2
04) If the status is usable, update data management table A
The bit of lO3 is also set to an updated state (205).

If it is in an unusable state, the process ends without updating the update data management table AlO3.

If it is an update data management table read request (206).

The content of the usability flag 111 of the update data management table A is determined (207), and if the update data management table A is usable, the update data management table A and the information indicating that the management table is the update data management table A are transferred to the upper level. Transfer to the system (208). If it is in an unusable state, update data management table B and information indicating that the management table is update data management table B are transferred to the higher-level system (209).

If it is an update data management table initialization request (210), update data management tables A and B are created and initialized on volatile memory and nonvolatile memory, respectively (211), and update data is stored on the nonvolatile memory. A usability flag is set for management table A, and it is set in a usable state (212).

It is assumed that the flag becomes unusable when the contents of the update data management table A in the volatile memory disappear due to a power cut in the disk control device or the like.

An example of the logical configuration of the update data management table will be explained based on FIGS. 3 and 4.

FIG. 3 is an example of the logical structure of the update data management table A.

It is composed of a cylinder number, track number, block number, and update information indicating whether or not there has been an update so that the updated portion can be managed in units of blocks. In this example, if there is an update, the bit is set to 1. If there is no update, the bit is assigned as OFF. Cylinder number 123. This indicates that the data corresponding to block number 2.3 of track number 4 has been updated. In addition, on 1. The meaning of OFF 0 may be assigned in reverse.

FIG. 4 is an example of the logical configuration of the update data management table B.

It consists of a cylinder number, a track number, and update information indicating whether or not there has been an update so that the updated portion can be managed on a track-by-track basis. The contents of the update information are the same as in FIG. 3. In this example, track number 4 of cylinder number 123,
6.7 has been updated.

Note that the physical configuration example of the update data management table may be the same as the logical configuration example, or the management table may not include cylinder numbers, track numbers, and block numbers, and may consist only of update information and correspond to the updated parts. When determining the update information, an example implementation may be to calculate the number of cylinders, tracks, blocks, etc. of the disk device and specify the update information corresponding to the target cylinder, track, or block. A physical configuration example is omitted because it is not the purpose of the present invention.

An embodiment of claim 2 of the present invention will be described.

The overall configuration and the configurations of update data management tables A and B are similar to the embodiment of claim 1.

Further, the operation of the microprocessor within the disk controller 104 in response to a data update request or an update data management table initialization request from the host system is similar.

Here, the operation in response to a request to transfer only update data will be explained based on FIG.

If the request is to transfer only update data from the upper system (501), the availability flag 11 of update data management table A is set.
Determine the contents of 1 (502) L. If available, update data management table A that manages update data in blocks.
Based on lO3, disk device i! 114 and transfers it to the upper system (503). If the update data is not available, the update data is read from the disk drive ft1 based on the update data management table B110 that manages the update data on a track-by-track basis.
14 and transfers it to the higher-level system (504).

Hereinafter, one embodiment of claim 3 of the present invention will be described based on FIG. 6.

FIG. 6 is a diagram showing the overall configuration of the present invention.

601 is (Central Processing U
nit: central processing unit), 602 is a channel, 60
3 is a disk control device represented by a magnetic disk control device; 604 is a channel interface control unit that exchanges data between the channel 602 and the disk control device f1603, which is represented by a magnetic disk device; and 605 is a data storage unit that temporarily stores data. A buffer 606 is a microprocessor 60 that controls the disk controller 711603.
7 is an update data management table A6 that manages updated portions of data in the disk device 604 after a certain point in block units.
08 is stored in a volatile memory, 609 is a non-volatile memory in which a track-by-track update data management table 8610 and an update data management table A usability flag 611 are similarly stored, and 612 is a non-volatile memory including a battery. 613 is a device interface control unit that exchanges data between the disk controller 603 and the disk device 614; 615 is a disk device that stores only the updated part specified by the host system; 616 is a The main storage device 617 is a program that updates data in the disk device and issues requests to transfer only updated data.
02.616 and 617 are collectively called the upper system.

Update data management table A608 and update data management table 86
10 is the third configuration shown in one embodiment of claim 1.
It is the same as Fig. 4.

Further, the operation of the microprocessor within the disk controller 604 in response to a data update request or an update data management table initialization request from the host system is also similar.

Here, the operation in response to a request to copy only updated data will be explained based on FIG.

If the request is to copy only updated data from the higher-level system (701), the update data management table A availability flag 61
Determine the contents of 1 (702) L. If available, read the update data from the disk device 614 based on 608 into the update data management table, and write it sequentially to the disk device 615 specified by the host system (703). . If the update data cannot be used, similarly based on the update data management table 8610, the update data is read from the disk drive 614 and sequentially written to the disk drive 615 (704).

An embodiment of claim 4 of the present invention will be described.

The overall configuration is the same as the embodiment of claim 3. However, the disk device 615 is used as a destination for overwriting update data.

The structure of the update data management tables A and B and the operation of the microprocessor in the disk controller 604 in response to data update requests and update data management table initialization requests from the host system are also similar.

Here, the operation in response to a request to overwrite only update data will be explained based on FIG. 8.

If it is a request to overwrite only update data from the host system (801), the content of the usability flag 611 of update data management table A is determined (802).

If available, update data is read from the disk device 614 based on the update data management table A608.

Overwrites the same cylinder, track, and block of the disk device 615 specified by the host system (803
). If the data cannot be used, update data is similarly read from the disk device 614 based on the update data management table B and overwritten on the disk device 615 (804).

An embodiment of claim 5 of the present invention will be described.

The configurations of the overall configuration diagram and update data management tables A and B are similar to claims 1, 2, and 3.4.

Here, the operation of the microprocessor within the disk control device in response to a request to save update data management table A or B from the host system and a request to restore it will be explained based on FIG.

If it is a request to save update data management table A (901), determine the contents of the unusable flag of update data management table A (902).
), and if it is usable, the management table is written into the disk device designated by the higher-level system (903). If it is not available, return error information to the above system (904)
.

If the request is for recovery of update data management table A (905), update data management table A is read from the disk device specified by the host system and set in volatile memory (906).
). Next, the usability flag of the update data management table A in the nonvolatile memory is set to the usable state (907).

If it is a request to save update data management table B (908), update data management table B is written into the disk device specified by the higher-level system (909).

If it is a recovery request for update data management table 8 (910), update data management table B is read from the disk device specified by the host system and set in nonvolatile memory (91
1).

Note that the request to save and restore update data management tables A and B at the same time can be achieved by executing step 909 after step 903 when saving, and executing step 911 after step 907 during recovery. Omitted.

An embodiment of claim 6 of the present invention will be described.

The configurations of the overall configuration diagram and update data management tables A and B are similar to claims 1, 2, 3, and 4.5.

Here, the operation of the microprocessor in the disk controller in response to a request to restore update data management table A from a higher system will be explained based on FIG. Read update data management table A from the specified disk device and set it in volatile memory (100
2). Next, the availability flag of the update data management table A in the nonvolatile memory is set to the usage state (1003). Determine whether the bit of the update data management table A in the non-volatile memory corresponding to the track containing the block that has been updated in the read update data management table A is updated (10
04) L, if there is no correspondence, it is assumed that update data management table B has been initialized, and update data management table 8 is recreated based on the read update data management table A (l
OO5). If the response is correct, the process ends.

In all of the above embodiments, the hardware described as a disk control device or a disk device is assumed to be a magnetic disk control device or a magnetic disk device.

However, this may be a control device and a storage device such as an optical disk, a magnetic drum, or a floppy disk.

Also, although these storage devices do not have the concept of a cylinder, there are equivalents to tracks and blocks, so
The present invention can be applied.

Hereinafter, one embodiment of claim 7 of the present invention will be described based on FIG. 11.

FIG. 11 is a diagram showing the overall configuration of the present invention.

1101 is CPU (Central Process
sing Unit (central processing unit), 1102 is a channel, 1103 is a disk control device represented by a magnetic disk control device, and 1104 is a channel for exchanging data between the channel 1102 and the disk control device 1103 represented by a magnetic disk device. An interface control unit, 1105 is a data buffer that temporarily stores data, 1106 is a microprocessor that controls the disk control device 1103, and 1107 is update data that manages updated portions of data in the disk device 1104 after a certain point in block units. A volatile memory 1109 stores a management table A 1108, a non-volatile memory 1109 stores a track-by-track update data management table BI 110 and an availability flag 1111 for the update data management table A, and 1112 a battery. 1113 is a disk controller 603;
and a device interface control unit 1 that exchanges data with a disk device [1114] having update data.
115 is a magnetic tape device ii! that stores updated data of the disk device 1114! This is a magnetic tape control device that controls 1116.

Note that 1115 and 1116 may be other external storage devices such as a magnetic disk control device or a magnetic disk device.

1117 is the main storage device, 1118 is the disk system [111
This is a differential backup program that copies updated data in 1116 to another medium 1116, and in this embodiment, 11
01, 1102, 1117, and 1118 are collectively called the upper system.

Update data management table A608 and update data management table B6
10 is the third configuration shown in one embodiment of claim 1.
It is the same as Fig. 4.

Here, the operation of the differential backup program will be explained based on FIG. 12.

When the program is started, the disk controller 11
03 to read the update data management table and receives it (1201). Next, the type of the updated data management table that has been read is determined (1202), and if the updated data is the updated data table A, the updated data is identified in block units and a read command is generated for it (1203), and the updated data management table B is If so, update data is specified for each record and a read command is generated for it (1204).

The above command is issued to the disk controller 1103 to receive the update data and store it in another medium 1116 (1204).

An embodiment of claim 8 of the present invention will be described below.

In the overall configuration, a human power device such as a console for instructing online and offline is added to each of the components of claims 1, 2, and 3.4. The case of claim 1 is illustrated in FIG. In FIG. 1, 1317 consoles are added. Furthermore, the configurations of update data management tables A and B are the same.

Here, we will explain the upper system program 116 in Figure 1 that brings the disk controller online and offline.
The processing related to the present invention in the operation 617 in FIG. 6 will be explained based on FIG. 14.

When the program is started, it determines whether it is an online processing request (1401), and if it is an online processing request, one of the processes is to recover the update data management table A from the disk drive [1314]. The request is issued to the disk controller 13o3 (1402).

Similarly, in the case of offline processing, one of the processes is to issue a request to the disk controller 11303 to save the updated data management table A from the disk controller 1303 to the disk device (1403).

In this embodiment, only update data management table A is saved and
Although an example has been shown in which a request for recovery is issued to the disk control device, both update data management table A and update data management table B may be saved and restored.

Further, in step 1302, as a method for restoring the updated data management table A, any of the means described in claim 5 or 6 may be used.

〔Effect of the invention〕

(1) According to the present invention, it is possible to economically manage updated parts in a disk device on a track-by-track basis in order to improve reliability, and also to manage on a block-by-block basis in order to further reduce the backup amount.

The effects of the present invention will be explained using a database of a financial online system as an example.

For example, each account occupies 1 block, and 10 blocks equals 1 block.
It is assumed that it is managed as a truck.

If there are 5 million accounts, there will be 5 million blocks and 500,000 tracks. - If we assume that there are 100,000 updated records per day, and we take a backup of only the updated parts every day, it can be assumed that updated accounts on Financial Online occur almost randomly, so we can manage the updated parts on a track-by-track basis. This results in a daily backup volume of 100,000 tracks, which is 11 times more than the total data backup.
5 is enough. Furthermore, if we manage updated parts in block units, the amount of backup will be 100,000 blocks every day,
It costs 1,150 yen compared to backing up all the data, and 1/10 compared to backing up each track. Volatile memory can be incorporated much more cheaply than non-volatile memory and its non-volatile mechanisms, and by combining the two as in the present invention, it is possible to economically reduce the amount of backup (several tens of times less than a full dump). -) can be realized. Furthermore, even if the contents of the volatile memory are erased due to a power outage, etc., the updated data management table for each track is retained in the non-volatile memory, so even in that case, the amount of dump is only a few minutes less than the total dump. be able to.

(2) According to the present invention, the updated data management table can be saved and restored to the disk device, so even if the operation of the computer is stopped or interrupted, the updated data management table for each block or track will not disappear.

(3) Even if the update data management table in non-volatile memory is not saved during the intended operational stop/interruption, the track-based management table can be restored from the block-based update data management table that is being saved. Therefore, the amount of data saved to the disk device can be reduced.

In addition, there is no need for the non-volatile memory to retain its contents in the non-volatile mechanism during operation stoppage or interruption, and battery consumption of the non-volatile mechanism can be avoided.

[Brief explanation of the drawing]

FIG. 1 is an overall configuration diagram of the disk control device of the present invention, and FIG.
The figure is a flow diagram of the operation of the microprocessor in response to requests from the upper system, Figure 3 is a configuration diagram of update data management table A, Figure 4 is a configuration diagram of update data management table B, and Figure 5 is a diagram of the upper system. 6 is an overall configuration diagram of another disk control device according to the present invention, FIG. 7 is a flow diagram of the operation of the microprocessor in response to requests from the microprocessor, and FIG. FIG. 9 is a flow diagram of the operation of the microprocessor; FIG. 10 is a flow diagram of the operation of the microprocessor; FIG. 11 is an overall configuration diagram of another disk control device according to the present invention; The figure is a processing flow diagram of a differential backup program, FIG. 13 is an overall configuration diagram of another disk control device according to the present invention, and FIG. 14 is a diagram showing the overall configuration of another disk control device according to the present invention. If I'm 421, I'll have 3! lθ 口χ Figure Figure Foil Round Defense

Claims (1)

[Scope of Claims] 1. In a disk control device having an update data management table for managing a portion of data in a directly accessible disk storage device that is updated by execution of a job in a host system, the updated portion is stored in blocks. An update data management table A to be managed is provided in a volatile memory, and an update data management table B for managing updated portions on a track-by-track basis and a usability flag for the update data management table A are provided in a non-volatile memory for management. and, in response to an update data management table read request from the upper system, based on the usability flag of update data management table A, if possible, update data management table A is used, and if not, update data management table B is used as a management table. What is claimed is: 1. A disk control device comprising means for transmitting information indicating the distinction of the information to a host system. 2. In claim 1, in response to a data transfer request for only the updated part from the host system, if the update data management table A is usable, the update data management table A is used, and if it is not available, the update data management table B is used. 1. A disk control device comprising means for reading update data from a disk storage device based on the information and transferring the updated data to a host system. 3. In claim 1, in response to a data copy request from an upper system for only an updated part, if update data management table A is available, the update data management table A is used, and if not, update data management table B is used. 1. A disk control device comprising means for reading updated data from a disk storage device based on the information and copying it to a disk storage device specified by a host system. 4. In claim 1, in response to a data overwrite request for only the updated portion from the host system, if update data management table A can be used, then update data management table B is used, and if not, update data management table B is used. What is claimed is: 1. A disk control device comprising means for reading update data from a disk storage device based on the above information, and overwriting the updated data at the same position in the disk storage device specified by a host system. 5. In any one of claims 1 to 4, in response to a request to save updated data management table A or B from a host system, the updated data management table A or B in the disk control device is saved. , means to save the updated data management table A or B in the specified disk storage device in the disk controller in response to a request for restoring the updated data management table A or B from the upper system. What is claimed is: 1. A disk control device characterized by comprising: a means for recovering. 6. In claim 5, when the updated data management table A in the specified disk storage device is restored into the disk controller in response to a request for restoring the updated data management table A from the host system, the disk A disk control device comprising: means for restoring an update data management table B based on an update data management table A if an update data management table B in the control device is in an initial state. 7. In a differential backup system that backs up only updated data in a disk storage device controlled by a disk control device according to claim 1, based on the type information of the updated data management table read from the disk control device. A differential backup method comprising: generating an instruction to read update data from an update data management table A; and generating an instruction to read update data from an update data management table B. 8. The disk control device according to claim 6, means for recovering the updated data management table A from the disk storage device when the disk control device is brought online, and the updated data management table B from the updated data management table A. 1. A computer system comprising: means for recovering the updated data management table A; and means for saving the updated data management table A to a disk storage device when the disk control device is taken offline.