JPH0241052B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to a data management system, and particularly to a data management system that inputs sorted log data and performs recovery at high speed.

[Conventional technology and problems]

Recovery in the event of a data set failure is performed by restoration using backup data and actual update proxy using updated log data after backup creation. In the conventional technology, the updated log data was processed in the order in which it was acquired, so even if a buffering function was introduced at the time of recovery, input/output required a large amount of time.

[Purpose of the invention]

The present invention aims to eliminate the above-mentioned drawbacks and to provide a data management system that enables high-speed recovery.

[Structure of the invention]

Therefore, the data management method of the present invention, prior to the start of processing by the application program,
When acquiring backups of multiple datasets used in an application program and updating the datasets during the processing process, the updated data is written to the log file as log data, and a failure occurs in the dataset during the processing process. At this time, the log data of the above log file is separated into each data set, the log data separated into each data set is sorted into blocks, and the sorted log data sorted into each data set is sorted into blocks.
Recovery of the failed data set is performed using sort log data corresponding to the failed data set and backup data corresponding to the failed data set.

It is characterized by this.

[Embodiments of the invention]

Hereinafter, the present invention will be explained with reference to Examples.
Figure 1 is a diagram showing the relationship between application programs and log data, Figure 2 is a diagram for explaining an operating method that takes media failure into account, and Figure 3 is a process of one embodiment of high-speed recovery according to the present invention. It shows the flow.

In Figure 1, 1 is an application program, 2 is
AIM control unit, 3 is history log file, 4-1
4-5 represent backups of history log files, A1 through AN and B1 through BM represent user datasets, respectively. Also,
, , indicate the order of processing.

When the application program 1 requests to refer to the data set A1, the AIM control unit 2 passes the designated data a to the application program 1. in this case,
The AIM control unit 2 can be considered as a database management unit. Application program 1 updates data a to a'. Application program 1 is data set A1
When a request is made to write the updated data a' to the data set A1, the AIM control unit 3 writes the updated data a' to the history log file 3 before writing the updated data a' to the data set A1.
Write as data. After writing the updated data a' to the history log file 3, the AIM control unit 2 writes the updated data a' to the data set A1.

The updated log data is sequentially written to the history file 3 in block units. When the history log file 3 is full, its contents are copied to magnetic tape for backup, and the history log file 3 is used in circulation. In the figure, "NO=4" is written in history log file 3, but
This means that the history log file has already been backed up 3 times.
This indicates that the history log file 3 has been used for the fourth time. The same applies to others. Also, in the figure, "NO=1" is written in backup 4-1, which indicates that this backup is the first one.

FIG. 2 explains the operation method and recovery in consideration of media failure. In Figure 2, TA
1 to TAN and TB1 to TBM indicate the backup of the user data set.
Note that the same reference numerals as in FIG. 1 indicate the same parts.

In preparation for a media failure, user data sets A1, . . .
AN, B1,...BM backup TA1,...
TAN, TB1 or TBM are taken. If a failure occurs in dataset A1 at time t1, recovery is performed using the following procedure.

(b) Load backup TA1 and initialize data set A1.

(b) Extract valid log data regarding dataset A1 from the history log file backups 4-1 to 4-5 and history log file 3, update dataset A1, and write it out.

FIG. 3 shows a processing flow of one embodiment of high-speed recovery according to the present invention. In FIG. 3, 5 and 6 indicate sort log files, respectively. Note that the same reference numerals as in FIG. 1 indicate the same parts.

Assume that a failure has occurred in dataset A1 and the medium has been converted to a normal medium. First, the log data is sorted and then the recovery process is performed. The log data sorting process is performed as follows.

READ the log data from the history log file backup 4-1 or 4-5.

Separate the READ log data into data set groups. That is, data set A1
Every data set AN, every data set B1
Separate each data set BM. At this time, only the log data for the same block that generates the final result is left. For example, if the log data is the update result of a block that has been updated in units of blocks, then for the same block, the one whose acquisition order is the youngest as determined from the information about the order of log data. The (latest) log data will produce the final results.

Sort/merge log data in dataset according to block number.

If the sort/merge within each data set is done for all data sets, the sort/merge
Sort log data into log files 5 and 6
WRITE to.

After the log data is sorted as described above, recovery processing as described below is performed.

' Read backup data TA1 and write it to data set A1.

' Read only valid log data regarding dataset A1 from sort log file 5 and history log file 3, and update dataset A1. Valid log data is determined by the duration and extent of the failure.
The period is the period from when backup data TA1 is created until data set A1 becomes a failure. The failure is Dataset A
This refers to the range of failure. Data set A1
(for example, corresponds to a failure of the entire volume), the data set A for the above period is
It is sufficient to update all the log data of 1.
Furthermore, in the case of a partial failure in dataset A1 (for example, corresponding to a failure in some trucks), it is sufficient to update the log data corresponding to the failed part during the above-mentioned period.

' Write the updated data to dataset A1. When exporting, select the output medium
Utilizes the IO characteristics to output in units of track groups and cylinder groups.

〔Effect of the invention〕

As is clear from the above description, according to the present invention, recovery can be performed more efficiently and in a shorter time than in the conventional method.

[Brief explanation of drawings]

Figure 1 is a diagram showing the relationship between application programs and log data, Figure 2 is a diagram for explaining an operating method that takes media failure into account, and Figure 3 is a process of one embodiment of high-speed recovery according to the present invention. It shows the flow. 1...Application program, 2...AIM control section, 3...
History log files, 4-1 and 4-5...backup of history log files, A1 to AN and B1 to BM...user datasets,
TA1 to TAN and TB1 to TBM...user dataset backup, 5 and 6...sort log files.

Claims (1)

[Claims] 1. Prior to the start of processing by an application program, a backup of a plurality of datasets used by the application program is acquired, and when updating the datasets in the process of processing, the updated data is used as log data. When data is written to a log file and a failure occurs in a dataset during processing, the log data in the log file is separated into each dataset, and the log data separated by dataset is sorted into blocks. and sorted logs sorted by dataset.
A data management method characterized in that a data set in which a failure has occurred is recovered by using sort log data corresponding to the data set in which the failure has occurred and backup data corresponding to the data set in which the failure has occurred.