JPH0652053A - Fail safe type computer system - Google Patents

Abstract

PURPOSE:To reduce the memory capacity of an external memory device by accumulating roll-back information by distributing to the external memory device at every generation, and also switching the external memory device simultaneously when the generation is updated at a roll-back point. CONSTITUTION:The roll-back and restart of processing can be performed by using a disk of either generation in the past even when a fault occurs in one disk by switching the disk to be designated by a memory managing unit(MHO) at every roll-back point and deciding the disk to be used at every generation. In such a case, a virtual address instructed by an EPU 1 is changed to a real address by MMO hardware 2, and the read/write of the EPU 1 on a main storage part 3 is performed. Also, when no virtual address exists on the main storage part 3, a page fault occurs, and the unrequired content of the main storage part 3 is saved once to the disk similarly as an ordinary virtual storage system, and the content of an instructed virtual address is loaded from the disk.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a failure recovery type computer system which employs a rollback system for automatically retrying from a past checkpoint when an abnormality occurs in computer information processing.

[0002]

2. Description of the Related Art Conventionally, as a computer system having a rollback point, for example, there is one shown in Japanese Patent Laid-Open No. 3-265951.

FIG. 8 illustrates the rollback method disclosed in the above publication, showing the concept of generation update, in which rollback data is stored in an auxiliary storage area (file) so as not to be destroyed. , Generation management is performed.

As shown in FIG. 8A, the rollback point update processing (steps 81 and 83) and the application program execution processing (steps 82 and 84) are performed in time series, and the application program is executed. If an error occurs in the execution processing process (steps 82 and 84), the process returns to the rollback point (steps 81 and 83) immediately before that and the rollback data execution processing (steps 82 and 84).
To recover from the abnormal condition.

The update of the rollback point is to change the generation of rollback data at the rollback point, and is performed as shown in FIG. 8B, for example.

When the rollback point is detected during the execution of the application program, the new generation (85, 88), which is the latest data at that time, is indicated by the solid line arrow in the figure.
Is transferred to the file as rollback data of the previous generation (86, 89), and the program proceeds toward the next rollback point. If no abnormality has occurred, the update process for making the generation one generation old for each rollback point is performed by the above procedure.

In this way, a plurality of generations such as the second generation (new generation 85, previous generation 86) and the third generation (new generation 88, previous generation 89, previous generation 90) are updated, and rollback data ( New generations 86, 89, 90 before last, etc.) are formed.

On the other hand, when the data of the new generation (85, 88) is executed, the subsequent data is processed by the current generation (87, 9).
1) However, when an error occurs, the process returns to the immediately preceding rollback point and the restoration processing at the time of abnormality is performed. That is, as indicated by the broken line arrow in the figure, in the case of two-generation management, the previous generation 86 of the file is transferred to the main memory as the new generation 85, and the copied data of the new generation 85 is executed.

In the case of three-generation management, the pre-previous generation 90 on the file is the new pre-generation 89 and the old pre-generation 89.
Is transferred to the main memory as a new new generation 88. in short,
The process of returning the rollback data to the first generation is performed.

Further, as one of the methods for realizing the generation update shown in FIG. 8, the above-mentioned publication discloses MMU (Memory Ma).
A generation update method using the Pipping Unit) is shown.

FIG. 9 shows a generation updating method using this MMU, and shows an updating process for one logical page number "x".

In FIG. 9, as the contents of the logical page number "x" of the MMU conversion map immediately before the update, a flag is set in the write flag area of the physical page number "a" of the main storage unit and changed to the stored data. It is shown that there was.

Therefore, the physical page number "a" of the nth generation (old current generation) at the time of updating is set to the n + 1th generation (new current generation), and the contents of the physical page number "a" are used as rollback data for the file page. The file page number "u" is set as the nth generation (new previous generation) by transcribing to the area of the number "u". The flag is reset at an appropriate time after the transfer.

By the way, in the virtual storage system, basically, the storage contents of the entire software are placed in an external storage device such as a disk in correspondence with a virtual address, and a part of the main storage unit is copied from the external storage device as an actual storage. Basically, it is loaded (Roll-In) and returned to the external storage device (Roll-Out) when unnecessary. The MMU is a conversion table mechanism that indicates where in the external storage device or the main storage unit the information indicated by the virtual storage address is located.

In the generation management based on the rollback point, an appropriate checkpoint is set in the process of processing in preparation for the occurrence of an abnormality, and when an abnormality occurs, the method is to return to that point and restart the processing from that point. R
It was called "oll-Back-Point". Conventional M
The MU-based management method allows the MMU to manage multiple generations, and if an error occurs, the contents of the MMU are returned by one generation after performing error processing such as hardware switching and automatically restarting from the previous rollback point. I was able to do it.

[0016]

However, in the above-mentioned conventional MMU-based management method, the external storage device also has a redundant configuration, and in the case of 3-generation management, 3 generations × 2 units = one virtual page = Since an external storage device for 6 pages is required and a storage capacity of 6 times is required, improvement thereof has been desired.

The present invention has been made in view of the above points, and can reduce the storage capacity of an external storage device in a failure recovery type computer system of a rollback point system in which an MMU has generation management. The purpose is to obtain a failure recovery computer system.

[0018]

In order to achieve the above object, a failure recovery computer system according to claim 1 of the present invention is a rollback point system in which an MMU (memory management unit) has generation management. In a failure recovery type computer system, rollback information is distributed and accumulated in an external storage device for each generation, and the external storage device is also switched at the same time when a generation is updated at a rollback point.

A failure recovery type computer system according to a second aspect is the failure recovery type computer system according to the first aspect, when the generation is updated in rollback point processing,
By sending only the contents of the page changed from the end of the previous n-1 generation to the current n generation to the external storage device used in the next n + 1 generation, and performing the processing of the next n + 1 generation based on that. , DISK used in the nth generation and the n + 1th generation
The feature is that the continuity of the treatment is maintained even if the values are different.

A failure recovery type computer system according to a third aspect is a rollback point type failure recovery type computer system in which an MMU (memory management unit) has generation management. When the generation management is realized by the MMU conversion method in which the conversion is divided into a plurality of generations and managed, a disk to be used is determined in advance for each generation.

A failure recovery type computer system according to a fourth aspect is the failure recovery type computer system according to the third aspect, wherein only the conversion map of the current generation is stored in the hardware MM.
It is characterized in that it is placed in the U conversion mechanism, and MMU conversion information other than this, such as past and old generations, is stored in an external storage device.

A failure recovery computer system according to a fifth aspect of the present invention is the failure recovery computer system according to the third aspect, wherein the nth generation MMU conversion information is initially n-1 generation of the external storage device n-1 generation. The page at the end is specified, and the page is loaded from here only when the page is rolled in for the first time, and it is stored in the external storage device of the nth generation at the time of subsequent rollout. .

According to a sixth aspect of the present invention, there is provided a failure recovery type computer system according to the third aspect, wherein in the failure recovery type computer system, at the time of rollout in the nth generation, another external storage device also has a future generation (n + 1th generation). Etc.) for simultaneous writing.

According to a seventh aspect of the present invention, there is provided a failure recovery type computer system according to the third aspect, wherein in the failure recovery type computer system, contents between n-1 and n-2 generations are generated in the MMU generation management of three generations. In the case of the same, the n-2 generation page contents are rolled in from the n-2 generation external storage device side.

Further, the failure recovery type computer system according to claim 8 is the failure recovery type computer system according to claim 3, wherein during the MMU generation management, whether or not there is a change between generations of each page of the external storage device, n / n-1, n-1 / n-2
As described above, the change status between the two previous generations is stored.

According to a ninth aspect of the present invention, there is provided the failure recovery type computer system according to the third aspect, wherein in the MMU generation management, whether or not there is a change between generations of each page of the external storage device is determined. It is characterized in that the change status is stored as a difference from the nth generation, such as / n-1 and n / n-2.

Furthermore, the failure recovery type computer system according to a tenth aspect is the failure recovery type computer system according to the third aspect, wherein the MMU hardware has only a conversion mechanism between the current generation virtual address and the real address. , The old generation MMU mapping information is stored in an external storage device,
The rollback process when an abnormality occurs is characterized by loading this into the MMU hardware.

[0028]

By adopting the above-mentioned structure, according to each claim, it operates as follows. 1) According to claim 1, the capacity of the required external storage device is reduced. 2) According to claim 2, processing continuity is maintained even if the disks used in the nth generation and the n + 1th generation are different. 3) According to claim 3, the amount of MMU management information is reduced. 4) According to claim 4, the amount of hardware of the MMU is reduced. 5) According to claim 5, the transfer amount to the external storage device related to the rollback point process is reduced. 6) According to claim 6, the same operation as that of claim 5 is exerted. 7) According to claim 7, the load on the external storage device can be made uniform and the reliability can be improved. 8) According to claim 8, the process at the rollback point is facilitated. 9) According to claim 9, the same action as that of claim 8 is exerted. 10) According to claim 10, the MMU hardware is reduced.

[0029]

Embodiments of the present invention will be described below with reference to the drawings.

According to the present invention, in order to reduce the storage capacity of the external storage device, each generation is distributed / stored in a plurality of disks, instead of placing all the plurality of generations in the active system of the conventional external storage device, and the required disks are stored. FIG. 1 shows an embodiment in the case of the second generation management for reducing the capacity.

As shown in FIG. 1, the disks designated by the MMU are switched for each rollback point, and the disks to be used are determined for each generation. Using generation disks, processing can be pulled back (rollback) and restarted.

FIG. 2 shows a hardware configuration diagram thereof. That is, this embodiment shows an application example of a computer system having a redundant configuration, and this failure recovery computer is an EPU.
1, the MMU hardware 2, the main storage unit 3, the external storage device control unit 4, and the main system / redundancy management unit 5 are basically provided.
The virtual address designated by the PU 1 is converted into a real address by the MMU hardware 2, and the EPU for the main memory 3 is converted.
1 is read and written.

Since the virtual memory system is used here,
If the virtual address designated by the EPU 1 does not exist in the main memory unit 3, a page fault will occur, and as with the normal virtual memory system, unnecessary contents of the main memory unit 3 will be temporarily escaped to the disk (roll out) and designated. The contents of the virtual address will be loaded from the disk (roll-in).

Note that the MMU hardware 2 of FIG. 2 may have only the current-generation provisional-to-actual conversion mechanism. The contents of the old-generation MMU are escaped to the disk, and the MMU hardware is not processed until the rollback processing at the time of an abnormality occurs. M as a method to load 2
It is also possible to reduce the MU hardware 2.

In the processing of the rollback point after the normal processing of each generation is completed, the contents of the page changed in that generation are escaped to the disk used in that generation, and the software register at the rollback point is registered. It is necessary to write all contents such as contents, work contents, status information, and programs to the disc. At the same time, since the processing of the next generation is handled by a different next disk, it is necessary to ensure that the internal state used in the previous n generations is inherited to the next disk when updating the generation. is there.

In order to reduce the inheritance of this internal state,
By using the property that the program part does not normally change between generations, and that there is only a small change between generations, the current generation and the past generation are compared on a page-by-page basis. FIG. 3 shows a rollback point processing method in which only a certain page is transferred between disks or from the main memory to the next-generation disk.

In FIG. 3, the nth generation is the generation that has been processed so far, and the n + 1th generation is the next generation after this rollback point.

In the MMU conversion map shown in FIG. 3, the presence / absence of change indicates whether or not there is writing in the nth generation and the content of the page is different from the n-1th generation. When the write flag of each page in the main memory unit of FIG. 9 is write enabled at the time of rollout (including rollout at the rollback point), the “change flag” of this MMU conversion map is changed. By setting it, it means that the page is different between the nth generation and the n-1th generation.

In the MMU conversion map, only the current MM is used.
The MMU information on the backup side is placed in the U hardware and is escaped / stored while being generationally managed in the MMU escape area of FIG.

In the processing of the rollback point in FIG. 3, first, from the MMU working map used in the nth generation, the real address is the content in the main memory and the write "YES" in FIG.
The content of the page is rolled out to the disk # 1 and the "changed or not" flag of the corresponding MMU virtual address is changed to "present". After all the escape processing of the main memory is completed, this working MMU map is also saved in the disk # 1 and the saving in the internal state disk # 1 is completed.

Next, in preparation for starting the n + 1 generation,
In the example of FIG. 3, the disk # 1 is used for the nth generation, the disk # 2 is used for the n−1th generation, and the disk is backed up for the nth generation. In the next n + 1 generation, on the contrary, the disk # 2 is used and the disk # 1 is used as a backup. At this time, the disk # 2 needs to take over the final internal state of the n generation and proceed with the processing.

By the way, in this embodiment, the disk # 1,
Since # 2 is used / spare alternately, pages of the nth generation that have no change can be used as they are at the start of the n + 1th generation. For example, in FIG. 3, if there is no change in the page content of the address Y of the disk # 1 used in the nth generation, the page content of the address x of the disk # 2 of the n-1 generation corresponding to this is the same, For the n + 1 generation, it is sufficient to use the page of the disk # 2 address x, and if the MMU map is updated, the disk #
There is no need to transfer between 1 and # 2.

Therefore, in the example of FIG. 3, only the page portion of "changed" in the "changed" flag of the MMU conversion map needs to be transferred from the disk # 1 or the main memory to the disk # 2. It is possible to reduce the transfer amount.

The above-described embodiment of FIG. 3 is a method of transferring the contents of the page written in the generation at the time of processing the rollback point to the next-generation disk.
The example shown in does not transfer all the changed parts at the time of rollback point processing, but at the time of roll-in, first identifies where the content is and loads it from there, and transfers the data at the time of rollback point processing. The amount is reduced and the speed is increased.

Therefore, in the embodiment shown in FIG. 4, the MMU conversion map is expanded so that it is possible to identify which disk has contents in the n-th generation. # 2 address U is specified, at the time of roll-in, it is loaded from the disk U of n-1 generation of disk # 2, and at the time of roll-out when the page is no longer needed in the main memory, it will be the first time in n generation The data is stored in the address U of the disk # 1.

It is to be noted that in FIG. 4, ~ indicates the flow of time.

Indicates the contents of the original MMU map in the n-th generation, indicates the state immediately after roll-in, and indicates the change once in the roll-out process when there is a write in the memory. , Indicate the states after the roll-out processing.

The embodiment of FIG. 5 is an embodiment in which transfer between disks is not required in the processing at the rollback point of FIG. 3 as in the case of FIG. 4, and data is simultaneously written to disks of future generations during rollout. It is something that ends up. That is, when the initial MMU contents in the nth generation in FIG. 5 are different from the n-1th generation, they are already written in w of the disk # 1 which is the disk in the nth generation, and at the time of roll-in at →, The main memory is read from w of this disk # 1.

At the time of rollout, the data is written in U of the disk # 1 for the nth generation as in
It is also written in x of the disk # 2 for the n + 1 generation.

The contents of the MMU for the n + 1-th generation are initially undecided, or the disk for the n + 1-th generation (here, V of the disk # 2) is designated because it is not changed.

As a result, at the beginning of each generation, the MMU is always prepared with the MMU data on the disk used in that generation, thus improving efficiency.

Next, FIG. 6 shows an example of the third generation management.

In this embodiment, the MMU is provided with the difference information between the nth generation and the n-1th generation, and the n-1th generation and the n-2th generation.

If there is no change (difference) between the nth generation and the n-2th generation, instead of loading from the n-1th generation in the embodiment of FIG. 4, the n-2th generation as shown in FIG. (In this example, the disk # 1 is loaded), and the backup disk is not used as much as possible to improve the recoverability in the event of a disk failure.

In FIG. 7, the difference between the generations is n.
It is held in the MMU as a difference from the generation to facilitate the roll-in process.

[0056]

As described above, according to the present invention, the required capacity of the external storage device is reduced by reducing the redundancy of the escape information of each generation to the external storage device in the conventional rollback point system. There is an effect that can be.

Further, by reducing the amount of writing / reading to / from the external storage device, there is an effect that the speed of the system and the responsiveness can be improved.

[Brief description of drawings]

FIG. 1 is a basic conceptual diagram of a rollback system of the present invention.

FIG. 2 is a configuration diagram according to an embodiment of hardware of the present invention.

FIG. 3 is a conceptual diagram of rollback point processing of the present invention.

FIG. 4 is an explanatory diagram in which roll-back point processing is reduced by loading from an n-1 generation side disk when rolling in for the first time according to the present invention.

FIG. 5 is an explanatory diagram in which the write rollback point processing is reduced even in the n + 1-generation side disk at the time of rollout of the present invention.

FIG. 6 is an explanatory diagram of an example of 3-generation management.

FIG. 7: When n rolls in for the first time in 3 generation management
It is explanatory drawing which shows the Example loaded from the -2nd generation side.

FIG. 8 is an explanatory diagram of a conventional rollback method realized by an application program.

FIG. 9 is an explanatory diagram of conventional rollback point update processing.

[Explanation of symbols]

 1a, 1b EPU 2a, 2b MMU hardware 3a, 3b Main storage unit 4a, 4b External storage device control unit

Claims (10)

[Claims] 1. An MMU (memory management unit)
In a rollback point failure recovery type computer system that has generation management for each generation, rollback information is distributed and stored in the external storage device for each generation, and the external storage device is switched at the same time when the generation is updated at the rollback point. Failure recovery type computer system. 2. The failure recovery computer system according to claim 1, wherein when a generation is updated in rollback point processing, only the contents of the page changed between the end of the previous n-1 generation and the current n generation, The DI used in the nth and n + 1th generations is sent to the external storage device used in the next n + 1th generation, and the processing of the next n + 1th generation is performed based on the DI.
A failure recovery computer system characterized by maintaining processing continuity even if SKs are different. 3. An MMU (memory management unit)
In a rollback point failure recovery type computer system having generation management in each, an MM that manages conversion of virtual memory addresses and real memory addresses by dividing into multiple generations
A failure recovery type computer system characterized in that a disk to be used is determined in advance for each generation when the generation management is realized by the U conversion method. 4. The failure recovery computer system according to claim 3, wherein only the conversion map of the current generation is placed in the MMU conversion mechanism of the hardware, and the other MMs of the past and old generations etc.
A failure recovery type computer system characterized in that U conversion information is stored in an external storage device. 5. The failure recovery computer system according to claim 3, wherein at the beginning of the n-th generation MMU conversion information, a page at the end of the n-1 generation of the n-1 generation external storage device is designated, and the page is designated. Load from here only at the first roll-in to n
A failure recovery type computer system characterized by storing in an external storage device on the generation side. 6. The failure recovery type computer system according to claim 3, wherein at the time of rollout in the nth generation, it is simultaneously written in another external storage device for future generations (n + 1 generation etc.). Recovery computer system. 7. The failure recovery computer system according to claim 3, wherein n-1 and n-2 are used for MMU generation management of three generations.
A failure recovery computer system characterized in that, when the contents of each generation are the same, the page contents of the (n-2) th generation are rolled in from the external storage device side. 8. The failure recovery computer system according to claim 3, wherein during MMU generation management, it is determined whether or not there is a change between generations of each page of the external storage device, n / n-1, n-1 / n.
-2, a failure recovery type computer system characterized by storing the change status between two generations before and after. 9. The failure recovery computer system according to claim 3, wherein when managing the MMU generation, whether or not there is a change between generations of each page of the external storage device is determined as n / n-1, n / n-.
A failure recovery type computer system characterized in that the change status is stored as a difference from the nth generation as in 2. 10. The failure recovery computer system according to claim 3, wherein the MMU hardware has only a conversion mechanism between the current-generation virtual address and the real address, and the old-generation MMU mapping information is stored in an external storage device. A failure-recovery type computer system that is stored and loaded into the MMU hardware for rollback processing when an abnormality occurs.