JPS6257030A - History information managing system - Google Patents

Abstract

PURPOSE:To attain high speed system recovery by allowing a job being object of recovery to write information on a virtual storage device into an exclusive external storage device at a prescribed interval and allowing an alternate job to read the information from the exclusive external storage device. CONSTITUTION:If a fault of an execution on-line job is detected in a standby on-line job processing after the fault takes place, whether or not a table recovery by a check point CD at a prescribed interval according to the check point CD processing task is checked. In case of the table recovery processing, the end of the table recovery processing is awaited. Then in order to recover the table revision after the acquisition of the CD, the systmem recovery using a journal JNL is applied. Further, the CPU mode is switched from the standby on-line system to the execution on-line system to restart the on-line service.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to a history information management method suitable for realizing particularly high-speed system recovery in an information processing system.

[Background of the invention]

As a method for performing high-speed system recovery by installing two central processing units, for example, the IBM programming announcement (February 12.
1985) Imnis/Buinis Virgin 2 Release 1 Analyst with General Availability Brand for First Quarter 1
986
Arvning Annourcgmant (Fch
(wruary 12.1985).

IMS/VS Vgrzion2 FLalgasa
l knnnouncgdwitnGtngval A
Availability Planned for Fi
rst Q)bψview1986 XRF 5upport
Planned for kavailability
As shown in Third Qwartgr 19B6), there is a known method that reads out the history information output by the recovery target job from the replacement job before an error occurs, and makes the replacement job in the same state as the recovery target job. ing.

Although this method was effective in shortening the system recovery time, it required an increase in the usage rate of the external storage device because both central processing units had to access the external storage device that stores history information, and There is a problem in that the utilization rate of the processing device increases significantly due to the replacement job.

[Purpose of the invention]

The purpose of the present invention is to provide a method for reading history information output from a zip to be recovered from a replacement zip before an abnormality occurs in order to shorten recovery time in an information processing system that installs two central processing units to achieve high-speed processing. The purpose of the present invention is to provide a means for solving the problem of an increase in the utilization rate of an external storage device and an increase in the utilization rate of a central processing unit in which a replacement job exists due to accessing historical information by both central processing units.

[Summary of the invention]

Reading the history information is necessary to bring the replacement job into a state as close as possible to the recovery target job for rapid system recovery. In this case, reading out the history information output by the job to be recovered requires both central processing units to access the external storage device that stores the history information, and the usage rate of the external storage device increases. Problems include performance deterioration due to the storage device wait K, and the utilization rate of the central processing unit in which the replacement zip exists increases due to the replacement zip, which affects other zips.

The acquisition of history information includes a large amount of information, as it is used to record the contents of business processes other than those performed to recover the management information of the recovery target zip at the time when an abnormality occurred. Information unnecessary for the recovery process is also read out by the alternate zipper, so the utilization rate of the central processing unit in which the alternate zipper exists increases. It is preferable to read out historical information in an accumulated form during the shift zip. Furthermore, the load caused by reading can be reduced by reading the history information at regular intervals instead of reading the history information every time.

The concept of the history information management method according to the present invention will be explained below. ◎ Two types of external storage devices for acquiring history information are provided. Conventionally, there are external storage devices called journals that acquire change information when certain management information is changed, and external storage devices called checkpoints that acquire management information on virtual storage devices that should be recovered at regular intervals. Business processing contents that are unnecessary for recovery processing can be recorded in the journal.

When the recovery target zip changes the management information to be recovered, the changes are acquired in the journal and checkpoints are acquired at regular intervals. Checkpoints reduce the amount of information acquired by compressing and storing the information.

In the replacement zip, before an abnormality occurs, checkpoints are read at regular intervals and the management information on the virtual storage device is restored to the same state as at the time when the checkpoint of the job to be restored is obtained. If the job or central processing unit to be recovered becomes abnormal, if the checkpoint read process is in progress, wait for the read to complete, and if the read process is not in progress, immediately read the journal and restore the management information on virtual memory to the state at the time of the error. to recover.

According to the present invention, the increase in the utilization rate of the external storage device for storing history information can be solved by accessing only checkpoints with compressed information at regular intervals by the restoration target jib and the replacement job. The influence on the utilization rate of the central processing unit in which the replacement jib exists is suppressed because the number of times of processing is reduced by compressing and processing information at regular intervals.

The problem is that the process of acquiring checkpoints in the recovery target job increases at regular intervals, the performance deterioration of the recovery target zip due to K, and the system recovery time due to the increase in history information read by the replacement jib when an error occurs. , is the deterioration of . First, the process of acquiring checkpoints in the recovery target zip at regular intervals makes it impossible to recover the system using a replacement zip, and the amount of history information that needs to be read when starting another zip to perform system recovery. This process is naturally necessary to suppress this, and has no impact. The problem of the amount of history information read by the replacement jumper when an abnormality occurs can be solved by acquiring checkpoints at appropriate intervals because the amount of history information read can be kept to a certain amount [Embodiment of the invention] Below , an embodiment of the present invention is shown in FIG. 1, FIG. 2, and WXs
This will be explained using figures. CCP is communication control processing unit, CPU
JNL is the central processing unit, JNL is the journal, %CD is the checkpoint, DV is various devices, the solid line indicates a thread that is connected and accessed, and the dotted line indicates a thread that is connected but not accessed. I'll do it. FIG. 1 shows an example of the configuration of an online system. central processing unit CPU
One is used to start a recovery target job (execution online job) that is executed online, and the other is used to start an online replacement job (waiting online job).

A CPU in which an executing online job exists is called an executing online system, and a CPU in which a waiting online job exists is called a waiting online system. You should be able to start recovery target jobs for both the execution online system and the standby online system.

The journal JNL, checkpoint CD, and some DVs of various devices are connected from both the active online system and the standby online system, but the C is the one that is accessed from both.
Only D.

FIG. 2 shows the processing of an executing online job and a waiting online job before an abnormality occurs. The execution-on-life database consists of a transaction processing task that processes transactions such as input from a terminal, and a CD processing task that compresses table information and edits and outputs it at regular intervals. The periodic online processing task consists of a CD processing task that reads the CD output from the online system and restores the table at regular intervals. The transaction processing task of the execution online processing section is not operating before the abnormality occurs because there is no transaction processing task in the online processing section. The transaction processing task of the executed online job performs various device accesses, journal output, and table updates as online processing.

Figure $5 shows the processing of a standby online job after an abnormality occurs. When an abnormality in an executing online job is detected, a check is made to see if table recovery processing using CDs at regular intervals by a CD processing task is in progress. If the table recovery process is in progress, wait for the table recovery process to complete. after that,
Perform system recovery using journal to recover table update after CD acquisition. Thereafter, the CPU mode is switched from the standby online system to the active online system and the online service is restarted.

According to this embodiment, the journal is accessed only by the execution online system, and the transaction performance is desired to be influenced by the standby online system, so that the performance of the terminal system can be improved.

〔Effect of the invention〕

According to the present invention, only checkpoints are accessed from the two central processing units before an abnormality occurs, and moreover, they are only accessed at regular intervals. This has the effect of improving the performance of the recovery target jib and reducing the processing of the replacement jib, thereby improving the performance of other jobs.

[Brief explanation of the drawing]

Fig. 1 is a configuration diagram of an online system showing an embodiment of the present invention, Fig. 2 is an explanatory diagram of the processing method before the occurrence of an abnormality in Fig. 1, and Fig. 5 is an illustration of the processing method after the occurrence of an abnormality in Fig. 1. It is a diagram. CCP...Communication control processing unit, CPU...Central processing unit, JNL...Journal, CD...Checkpoint, DV...Each device. ,−“−

Claims (1)

[Claims] 1. In order to achieve high-speed system recovery, two central processing units are installed, one to start the job to be recovered, and the other to start the job that will take over in the event of an error. In an information processing system that continues processing with a replacement job when a situation occurs, the job to be recovered writes information on the virtual storage device to a dedicated external storage device at regular intervals, and the replacement job writes information from the dedicated external storage device. A history information management method characterized by reading.