JPH08235044A - Journal retrieval systm - Google Patents

Abstract

PURPOSE: To confirm the presence or absence of the relevant journal information without retrieving an entire journal file. CONSTITUTION: The journal information includes a time stamp showing the writing time point of the journal information and a transaction identifier. This identifier includes a time stamp that shows the starting time point of a transaction. When the journal information is retrieved, the corresponding transaction identifier is used as a retrieval key and the journal information is successively read out at and after its final end. Then the time stamp of the journal information is compared with that of the retrieval key. If the former stamp is later than the latter one and no coincidence is confirmed between both transaction identifiers, the next journal information is read out to undergo the same check. Then the retrieval of journal information ends when the time stamp of the read journal information is earlier than that of the retrieval key.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a system processing history (journal) search method.

[0002]

2. Description of the Related Art In a transaction processing system or the like, journal information of processed transactions is accumulated in a journal file on a non-volatile storage medium for use in failure recovery processing and the like. When retrieving journal information of a specific transaction in recovery processing or the like, it is important how to retrieve the target journal from a huge amount of journal files at high speed.

As a journal search method for the purpose of speeding up the search, for example, the one disclosed in Japanese Patent Laid-Open No. 3-202956 is known. In this conventional method, the date and time of each journal information is collected in a small size index file separately from the main body of the journal file, and the target journal information is searched at high speed by using this index file. To do.

A similar technique is disclosed in Japanese Patent Laid-Open No. 4-190454. In this method, a file that stores time information and the start pointer of the journal information group corresponding to this time information is provided in the journal file, and the target journal information is searched at high speed by using this time information file. It is what

[0005]

Among transaction processing methods, there is a method of saving only journal information of completed (commit) transactions and not saving journal information of backward recovery (rollback) transactions. In this type of transaction processing method, when the journal information of a specific transaction is searched and it is not found anywhere in the journal file, the transaction is determined to have been rolled back, and the transaction is Removed from recovery targets.

When the journal information which does not exist in the file is to be searched as described above, the fact that the journal information does not exist can be found only by searching the entire file. Therefore, even if the above conventional technique is used, if the target time does not exist in the index file or the time information file,
You have to search the entire range of files. As a result, there is a problem that the search time becomes long.

Further, since the entire range of the file is searched, the load on the disk device in which the file is stored becomes high. In addition, when a transaction is executed at the same time while searching for a file using such a search method, access to the disk in which the file is stored is significantly restricted, so the transaction response time and The throughput will also drop considerably.

Therefore, an object of the present invention is to quickly confirm that the journal information does not exist without searching the entire journal file when the journal information that does not exist in the journal file is to be searched. To provide a journal search method.

[0009]

According to the journal search method of the present invention, a journal file storing journal information having time information in the order of time information and a journal search request having designated time information are received, The former time information is compared with the latter time information by comparing the time information read means for reading the time information of each journal information from the searchable range in the file with the specified time information. The search range limiting means limits the searchable range in the journal file only to a range having time information later than the former time information when it is not later than the information.

The present invention also provides a novel journal search method performed by the above method.

[0011]

According to the method of the present invention, each journal information is stored in time series in the journal file together with the time information. The time information in this journal file indicates, for example, the time when each journal information was written in the journal file. Further, the journal information search (= read) request also includes time information related to the journal information to be searched. The time information of the search request indicates, for example, the processing time of the transaction that generated the journal information to be searched.

When a journal information search request arrives under the above conditions, first, the time information of one journal information is read from the searchable range in the journal file. Next, the time information is compared with the time information of the search request. As a result of this comparison, if the time information of the journal information is later than the time information of the search request,
Since the journal information may correspond to the search target, it is checked whether or not it is the search target by an appropriate method. As a result, if it is found that the read journal information is not the search target, then the time information of another journal information is read from the searchable range of the journal file, and the time check is similarly performed.

In this way, the search is advanced within the searchable range of the journal file. In this process, if the read time information is not later than the search request time information, it is clear that the search target does not exist in the file range having the time information earlier than the read time information. Therefore, the searchable range is limited to the range of the time information that is later than the read time information, and the above-described search processing is performed on the limited narrow search range thereafter.

If a search target exists in the journal file, the search target will be discovered at some stage when the above-described search processing is repeated. On the other hand, if the search target does not exist in the file, the search target will not be found forever. However, since the searchable range is narrowly restricted in the search process, the search outside the restricted searchable range is omitted. Therefore, even if the search target does not exist, the search of the file range in which the search target may not exist is omitted, and the entire file is not searched, so that the search end time is shortened.

In the preferred embodiment, a simple sequential search method is adopted as a search method for a journal file, and the journal information having the latest time is read in order from the end to the beginning of the journal file. Then, it is checked whether or not it corresponds to the search target. When this process is repeated without finding the search target and the time information of the read journal information is no later than the time information of the search request, it is determined that there is no search target at that time. , The search process is terminated. Therefore, no search is performed on the file range having the time information earlier than the time information of the search request. Therefore, the search time is shortened.

In the preferred embodiment, the transaction identifier for each transaction includes a time stamp indicating the start time of that transaction. Further, the journal information in the journal file includes the transaction identifier of the transaction that generated the journal information and the time stamp indicating the writing time of the journal information.

When retrieving certain journal information, the transaction identifier of the transaction that produced the journal information is used as a retrieval key. First, the journal information is read in order from the end of the journal file, and the time stamp (write time) is compared with the time stamp (transaction start time) of the search key. As a result, if the former is later than the latter, then the transaction identifier of the read journal information is compared with the transaction identifier of the search key, and if both transaction identifiers match, the journal information is output as the search result. .
If the transaction identifiers do not match, the next journal information is read and checked in the same way.

If the time stamp of the read journal information is earlier than that of the search key without the transaction identifiers being matched, the search target is retried at that stage when the above processing is repeated. And the search process is terminated.

In this embodiment, a simple sequential search method is adopted as the search method, but it is also possible to implement the present invention by adopting another search method. For example, 2
According to the present invention, the searchable range is narrowed as the search processing progresses even when the minute search method or various balanced tree search methods developed from the minute search method are adopted, so that the search time is shortened. . Further, although the search of the embodiment directly accesses the journal information itself in the journal file, an index file and a time information file are provided separately from the main body of the journal information as in the search of the above-mentioned patent publication. The present invention can also be applied to the case of accessing these index files and time information files.

[0020]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 shows the configuration of an embodiment of a journal processing apparatus to which the journal search method of the present invention is applied.

The journal processor 1 manages a journal file stored in a disk device (not shown). The journal information of the transactions processed in the past is accumulated in the journal file in time series. It should be noted that the journal processing device 1 is shown as one independent hardware device for the sake of convenience of description, and of course it may be configured as such, but in reality, it is actually a part of the transaction processing system. It is a group of functional elements realized by a program, and may be installed in a single hardware device or distributed to a plurality of hardware devices.

In the journal processing device 1, the journal processing request input unit 3 accepts a journal writing request and a journal reading request from an external transaction processing device (not shown). Here, the journal write request is a request to write the journal information of a newly processed transaction in a journal file, and at least this request has the identifier of the transaction and the content of the journal information (performed in the transaction processing). The updated contents of the database etc.) are included.
The journal read request is a request for retrieving the journal information of a specific transaction from the journal file and reading the journal information to the outside, and this request includes at least the identifier of the transaction.

The received journal writing request and journal reading request are sorted into types by the journal processing request sorting unit 5, the journal writing request is sent to the journal writing request processing unit 7, and the journal reading request is sent to the journal reading request processing. Sent to department 9.

The journal write request processing unit 7 prepares journal information in a predetermined format based on the journal write request and writes the journal information in a journal file. The journal read request processing unit 9 performs a process of searching and reading the specified journal information from the journal file based on the journal read request.

The journal processing result output unit 11 includes a journal writing request processing unit 7 and a journal reading request processing unit 9.
It receives the respective processing results from and sends them back to the transaction processing device that made the request request.

FIG. 2 shows the configuration of the journal write request processing unit 7.

The journal writing instruction receiving unit 13 receives the journal writing request, adds a unique journal number to the journal information, and passes it to the time stamp adding unit 15. The time stamp adding unit 15 prepares journal information in a predetermined format based on the journal writing request, and at that time, adds a time stamp indicating the current time to the journal information.

The journal writing processing unit 17 writes the journal information prepared by the time stamp adding unit 15 in the journal file. At this time, since the past journal information is arranged in time series in the journal file, the new journal information is added to the end of the arrangement. Journal writing process end part 1
Reference numeral 9 confirms that the writing of the journal information is completed, outputs the completion result, and terminates the operation of the journal writing request processing unit 7.

FIG. 3 concretely explains the processing of the time stamp giving section 15.

FIG. 3A shows the transaction identifier 100 included in the journal write request.
This transaction identifier is created by an external transaction processing device, and includes a transaction number unique to the transaction and a time stamp indicating the start time of the transaction.

FIG. 3B shows the time stamp adding section 15.
Information prepared by the journal write request 2
00 is shown. The journal information 200 includes the contents of the journal information acquired from the journal write request (update information of the database and the like, not shown) and the copy 203 of the transaction identifier 100 (FIG. 3A), as well as the time stamp portion. The journal information identifier 201 subscribed by 15 is included.

The journal information identifier 201 includes the journal number unique to the journal information given by the journal writing instruction receiving unit 13 and the time stamp given by the time stamp giving unit 15 at the time of processing. There is. The time stamp in the journal information identifier 201 substantially indicates the writing time of the journal information, and naturally, the transaction identifier 2
The time is slightly later than the time stamp (transaction start time) in 03 (100).

The journal information in the format as shown in FIG. 3B is accumulated in the journal file in time series.

FIG. 4 shows the configuration of the journal read request processing unit 9.

The journal read instruction receiving unit 21 receives the journal read request and notifies the journal read processing unit 23 of it.

The journal read processor 23 first reads the journal information at the end of the journal file (that is, the writing time is latest). When a notification that the transaction identifiers do not match is received from the transaction identifier determination unit 27, which will be described later, the journal read processing unit 2
3 reads the next slowest journal information. In this way, the journal read processing unit 23 reads the journal information from the end of the journal file toward the beginning, in the order of the latest writing time.

The time stamp determination unit 25 compares the time stamp of the journal information identifier 201 in the journal information read by the journal read processing unit 23 with the time stamp of the transaction identifier 100 in the journal read request. As a result, when the time stamp of the journal information identifier 201 (journal information writing time) is later than the time stamp of the transaction identifier 100 (transaction start time), the transaction identifier determining unit 27 is requested to perform transaction identifier determination processing. If the journal write time is not later than the transaction start time, the journal read processing end unit 29 is notified to that effect.

Upon receipt of the request from the time stamp determination unit 25, the transaction identifier determination unit 27 compares the transaction identifier 203 in the journal information read by the journal read processing unit 23 with the transaction identifier 100 in the journal read request. To do.
As a result, if the transaction identifiers 203 and 100 do not match, the fact is notified to the journal reading processing unit 23, and if they match, the journal reading processing ending unit 29 is notified.

When the journal read processing termination unit 29 receives the notification from the time stamp determination unit 23 that it is not late, it determines that the journal information to be searched does not exist in the journal file and outputs the determination result. At the same time, the operation of the journal read request processing unit 9 is terminated. On the other hand, when the notification of the above-mentioned coincidence is received from the transaction identifier determination unit 27, the journal information read by the journal read processing unit 23 is output as a search result and the operation of the journal read request processing unit 9 is ended. .

FIG. 5 shows an operation flow of the journal read request processing unit 9 having the above configuration.

When the journal read request is received, the following search operation is performed using the journal read request as a search key.

First, the last journal information of the journal file is read (S1). At that time, if the data could not be read normally (S2), the search ends in failure and the process ends.

When the journal information is normally read, next, the transaction identifier 100 in the search key
Inside time stamp (transaction start time),
Journal identifier 201 in the read journal information
The time stamp (journal writing time) is compared (S3). Since the transaction start time is always earlier than the write time of the journal information, if the comparison result indicates that the transaction start time is not earlier than the journal write time, the process ends as a search failure.

On the other hand, if the comparison result shows that the transaction start time is earlier than the journal write time, then the transaction identifier 1 in the search key
00 and the transaction identifier 203 in the read journal information are compared (S4). As a result, if they match, it is determined that the search is successful, the journal information is output, and the process ends.

If the transaction identifiers do not match, the previous journal information is read from the journal file (S5). At this time, if the data cannot be read normally, the process ends as a search failure, and if the data can be normally read, the journal information is checked again in steps S3 and S4.

By the above operation, the journal information is searched only in the range of the journal information of which the writing time is later than the transaction start time of the reading object in the journal file, and the writing time is earlier than the transaction start time. The search for a range of information is omitted. Therefore, even if the read target is journal information that does not exist in the file, it is not necessary to search the entire range in the journal file, and the search is completed in a short time.

Further, since the search range of the journal file is limited to only a part of the range, the load on the disk device storing the journal file is also reduced. Furthermore,
When a transaction is executed during the journal search, the response time and throughput of the transaction are also improved.

Although one embodiment of the present invention has been described above, the present invention can be implemented in various modes other than this embodiment. For example, a search using an index file as disclosed in Japanese Patent Laid-Open No. 3-202956 and Japanese Patent Laid-Open No.
Also in the search in the time information file disclosed in No. 190454, if only the time range later than the time indicated by the search key is searched according to the present invention, the search time can be further shortened particularly when there is no search target.

[0049]

According to the present invention, even if non-existent journal information is a search target, only the range limited by the time need be searched, so the search time is shortened.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of a journal processing apparatus to which the present invention is applied.

FIG. 2 is a block diagram showing the configuration of a journal write request processing unit of the embodiment.

FIG. 3 is a data configuration diagram showing a configuration of a transaction identifier and journal information used in the embodiment.

FIG. 4 is a block diagram showing the configuration of a journal read request processing unit of the embodiment.

FIG. 5 is a flowchart showing the operation of a journal read request processing unit.

[Explanation of symbols]

 1 Journal Processing Device 7 Journal Write Request Processing Unit 9 Journal Read Request Processing Unit 15 Time Stamp Adding Unit 23 Journal Read Processing Unit 25 Time Stamp Determination Unit 27 Transaction Identifier Determination Unit

Claims (5)

[Claims] 1. A journal file in which journal information having time information is stored in the order of time information, and a journal search request having designated time information is received, and individual journals are searched from the searchable range in the journal file. Time information reading means for reading time information of information, and the read time information and the specified time information are compared, and when the former time information is not later than the latter time information, the journal A journal search method comprising a search range limiting means for limiting a searchable range in a file to a range having time information later than the former time information. 2. The method according to claim 1, wherein the time information of the journal information in the journal file indicates the time when the journal information is written in the journal file, and the time information of the journal search request is A journal search method characterized by indicating the processing time of the transaction that generated the journal information to be searched. 3. The method according to claim 1, wherein each journal information in the journal file includes, in addition to the time information, a transaction identifier unique to a transaction that generated each journal information, and the journal search request is A journal including a transaction identifier unique to the transaction that has generated the journal information to be searched, and the transaction identifier of each transaction includes time information indicating a processing time of each transaction as the specified time information. Search method. 4. The method according to claim 3, wherein when the time information reading means reads the time information of the individual journal information, the transaction identifier of the individual journal information is also read. When the result of the comparison by the search range limiting means indicates that the former time information is later than the latter time information, the read transaction identifier is compared with the transaction identifier included in the journal search request. However, when both transaction identifiers match, it is determined that the search is successful, and further comprises a success determination means for outputting the corresponding journal information. 5. A process of preparing a journal file in which journal information having time information is stored in the order of time information, and a journal search request having specified time information is received, and the journal file within the searchable range in the journal file is received. If the former time information is not later than the latter time information by comparing the time information read process of reading the time information of each journal information with the specified time information. A search range limiting step of limiting the searchable range in the journal file to only a range having time information later than the former time information.