WO2003003244A1

WO2003003244A1 - Method of rapidly eliminating different information in databases

Info

Publication number: WO2003003244A1
Application number: PCT/CN2002/000248
Authority: WO
Inventors: Zhijian Lu; Xiaozheng Guo; Jing Liu
Original assignee: Huawei Technologies Co., Ltd.
Priority date: 2001-06-28
Filing date: 2002-04-09
Publication date: 2003-01-09
Also published as: CN1192308C; CN1393771A

Abstract

A method of eliminating different information in databases in a multi-database application system, in which when different information is generated in operating these databases, the different information is recorded in a memory table and a different information file by a management background in a management configure module. Receiving retry request from the management foreground, extracting flag of the different information and database SQL instruction or storage procedure to be executed based on the received ID, and re-performing these SQL instruction or storage procedure using corresponding databases. When retry-operation is fully successful, different information records are deleted from the memory table and the different information file. When retry-operation partially successful, the different information in the memory table and the different information file is updated.

Description

Method for quickly eliminating database difference information

The invention relates to a method for quickly eliminating difference information of a database. In particular, a method for eliminating database difference information generated in multiple related databases that are processed in parallel in a large-scale distributed system to ensure consistency of operations on multiple databases. Background technique

With the development and popularization of computer technology, large databases have quickly entered various industries such as telecommunications and finance. In the era of information explosion, in many cases, it is increasingly difficult for a single database to meet the needs of performance and information volume. Therefore, distributed databases are put into more use.

Taking the call management center (CMC) as an example, the main service of the call management center is dial-up Internet authentication service. A dial-up user's dial-up request arrives at a certain access server (NAS), and the access server sends the request to the call management center, and the call management center decides to accept or reject the request according to some authentication policy. The dial-up Internet authentication service of the call management center is specifically implemented by the protocol processing module. After the protocol processing module runs, it always monitors the packets sent by the access server, and processes them accordingly, and sends the result packets back to the access server. .

The processing of the packets sent by the access server by the protocol processing module depends on the stored procedures provided by the database. That is, the database provides storage procedures for processing various messages. The protocol processing module constantly calls the storage procedures of the database during the interaction with the access server. A large number of dial-in users' online information generated by the interaction process is also stored in the database.

Therefore, the dial-up Internet authentication service of the call management center is implemented by combining the protocol processing module and the database module (these two modules can run on the same machine, or they can run on one machine each). With the rapid growth of business volume, the operation mode of a single call management center cannot meet user needs. In order to improve the system's wholesale port processing capacity, multiple distributed call management centers (DCMC) are used. The distributed call management center uses distributed and parallel processing IP to access the public platform system, and implements a mechanism for multiple databases to operate independently. Each database corresponds to a call management center, and different databases manage call control with different access codes. Therefore, each database must be configured with the same port, routing management center (RMC) and other network-wide resource information. In a multi-database application system, the management desk software often needs to synchronize multiple databases and call the same storage.

1

Replacement page (Article 26) The process is responsible for the configuration query of all databases in the system. When configuring (adding, deleting, and modifying) resources on the entire network, the same stored procedure needs to be called for all databases in the system. Due to various reasons (such as network failure), some databases may execute stored procedures successfully, and some databases may fail to execute stored procedures. At this time, database difference information appears. The database difference information will affect the port wholesale processing capacity of the entire system and cause unnecessary call loss. Therefore, how to quickly and reliably eliminate database difference information is a problem that must be solved in a multi-database application system.

In a multi-database application system, the rollback method is generally adopted to eliminate the database difference information. The strategy adopted for the rollback method is to execute the stored procedure in the reverse direction on the database that has been successfully configured once the difference information is generated. Considering that when the difference information is generated, each database has finished executing the stored procedure, so it cannot use the database's own rollback mechanism. In addition, the rollback method has the following defects −

1. Increased the overhead of writing the reverse stored procedure, and when some stored procedures need to update multiple tables, it is difficult to write the corresponding reverse stored procedure;

2. Reverse stored procedures cannot guarantee successful execution and may result in failure to eliminate discrepant information.

Therefore, in view of the above problems of a multi-database application system, an object of the present invention is to provide a method capable of quickly and reliably eliminating database difference information. Summary of the Invention

The method for eliminating difference information of a database provided by the present invention uses a retry mechanism. The retry mechanism is to record the difference information (including the stored procedure command and the identification of the database that failed the operation) once the database difference information is generated, and trigger or manually control the re-execution of the stored procedure that failed the operation on the database that failed the operation.

According to an aspect of the present invention, a method for eliminating database difference information in a multi-database application system by using a management configuration module is provided. The management configuration module includes a management background and a management foreground. The method is characterized in that the method includes the following steps:

(a) Setting the management background of the management configuration module to generate a difference information memory table and a difference information file,

(b) execution of structured query language (SQL) instructions or stored procedures in the corresponding database,

(c) checking the processing results of the structured query language instructions or stored procedures of the database to determine whether the database operation was successful,

(d) if it is judged in step (c) that the database operation is partially successful, generating difference information with a retry identification code (RedoID) for the database in which the operation failed, and recording the difference information in the difference information memory table And the difference information file, returning a partial success indication to the management front, (e) the management background receives the retry command carrying the retry identification code sent by the management frontend, and finds the difference information corresponding to the retry identification code in the difference information file according to the received retry identification code,

(f) obtaining the tag field of the difference information record and the stored procedure or structured query language instruction to be executed from the corresponding difference information, and calling the stored procedure or structure to a database or some databases according to the pointer of the difference information Translate query language instructions and perform retry operations,

(g) check the processing result of the retry operation, and judge whether the retry operation is all successful,

(h) If it is judged in step (g) that all the retry operations are successful, delete the difference information recorded in the difference information memory table and the difference information file, and return all success instructions to the management front desk

(i) if it is determined in step (g) that the retry operation is not all successful, determine whether the retry operation is partially successful or all failures,

(j) If it is determined in step (i) that the retry operation is partially successful, update the difference information recorded in the difference information memory table and the difference information file, and return a partial success indication to the management front.

In addition, before step (b) of executing the structured query language instruction or stored procedure in the corresponding database, the method further includes receiving a command request from the front desk and converting the command request into a corresponding stored procedure or structured query language instruction. A step of.

In addition, the step (c) of judging whether the database operation is successful further includes judging whether the database operation is all successful or all failures. If the database operation is all successful, returning an indication of all successes to the management foreground, and if the database operation is all failing, returning to the management The foreground returns all failed instructions.

In addition, the step (i) of updating the difference information recorded in the difference information memory table and the difference information file further includes deleting the difference information corresponding to the database in which the retry operation is successful.

In addition, if it is determined in step (i) that all the retry operations have failed, it further includes returning an indication to the management foreground that all the retry operations have failed.

The difference information in the present invention includes a retry identification code field, a flag field, and a query statement or stored procedure executed by a database when the difference information is generated. BRIEF DESCRIPTION OF THE DRAWINGS

The method for eliminating the difference information in the database of the present invention is described in detail below with reference to the accompanying drawings, so as to better understand the purpose, characteristics and advantages of the present invention.

Figure 1 is an overall block diagram of a distributed call management center;

2 is a flowchart of generating difference information in a database; 3 is a difference information recording format according to the present invention;

FIG. 4 is a flowchart of eliminating database discrepancy information according to the present invention. detailed description

The structure and working principle of the distributed call management center will be described with reference to FIG. 1. In the figure, three call management centers are taken as an example to form a distributed call management center. The distributed call management center is mainly composed of a protocol processing module 2, a database 3, and a management configuration module 4. In operation, the access server 1 receives a dial request from the dial-up user to access the data, and sends the access request to the call management center. The protocol processing module 2 of the call management center sends the access server 1 to the access server 1 according to the stored procedure provided by the database 3. Analyze the packets and check the validity of the access request. Decide whether to accept or reject the access request based on the specific authentication method. After the protocol processing module 2 runs, it always listens to the packets sent by the access server 1. During the interaction with the access server 1, it constantly calls the stored procedure of the database, processes the packets accordingly, and sends the result packets back. Give access server 1.

Database 3 mainly has two functions: first, providing an interface for the main process protocol processing part and ensuring the main process processing efficiency; second, providing an interface for the management configuration module 4 and ensuring the data consistency of the management configuration module. The main functions of the management configuration module are: first, providing call management management and configuration functions for system operators, including customer management, resource management, operator management, performance management, and security management; second, providing real-time network for port lease customers Information on operating conditions and the ability to perform statistical analysis, allowing customers to understand performance indicators related to them.

The role of the management configuration module 4 is to provide an interface for system administrators to configure database information (such as which access servers are in the system) and query information on the Internet. The management configuration module 4 is further divided into a management background and a management foreground. The management background is the processing core. Only one copy of the entire call management center system is configured. The management foreground provides a human-computer interaction interface. There may be multiple management foreground programs running in the system at the same time. The management foreground and management background communicate through the CORBA (Common Object Request Agent Architecture) standard and the II0P protocol (Internet Target Request Agent Protocol). The ΠOP protocol can realize the interoperability between the Internet client and the CORBA server. On the other hand, the interoperability between the CORBA client and the Internet server can be achieved, so that even if different types of computers or various programming languages are connected, it is written as As long as the objects meet the CORPA standard, they can communicate with each other. The management front end responds to the configuration or query request input by the user, and converts the configuration or query request into a command request to the management back end. After receiving the command request, the management background executes the related database stored procedure or query command, and delivers the final processing result to the corresponding management foreground. The management front end finally displays the received results on the interface. The operators in the management front desk only deal with the management front desk and do not feel the existence of the management back office. In the management background, corresponding to each database 3 in the system, there is a corresponding database worker thread (worker thread for short), and a worker thread can be understood as a connection to a corresponding database. There are three databases in the system shown in Figure 1, and there are three working threads in the management background. All worker threads are controlled by the main thread in the management background. When the management background does not receive a command request from the management foreground, all the worker threads in the management background are in a waiting state;

The generation process of the database difference information is described below with reference to FIG. 2. First, in step S200, a management background of the management configuration module is set to generate a difference information memory table and a difference information file. The difference information memory table and the difference information file are used to record all the difference information currently existing in the multi-database application system. In this way, even if the management background exits abnormally (such as power failure), the difference information can still be recovered from the physical file of the difference information. In step S201, the management background of the management configuration module in the call management center receives a command request sent by the management front desk, where the command request includes a database identification code (DBID) representing each database, and the management backstage analyzes the request, and The command request is converted into a corresponding database SQL (Structured Query Language) command or stored procedure. In step S202, the management background extracts the DBID from it. DBID instructs to operate on that database. For example, there are three databases in the system, and their serial numbers are ₀ and ΐ 2, respectively. DBID-2 means that this command only operates on the database with serial number 2. In addition, a case is set. When DBID = -1, it means that this command requests to operate all databases in the system. In step S203, the management background sets the initial environment of the worker thread corresponding to the DBID according to the value of the extracted DBID. Thereafter, in step S204, the management background unlocks the mutex of one or all worker threads corresponding to the DBID. After unlocking, the worker thread enters the working state and executes the SQL command or stored procedure of the corresponding database. The management background waits for all the worker threads to finish processing in step S205. After the worker threads execute the database SQL commands or stored procedures, the mutex is locked again by the main thread, and the worker threads return to the waiting state. In step S206, the processing result of the SQL command or stored procedure of the database is checked to determine whether the database operations are all successful. If it is determined in step S206 that all database operations are successful, then all success indications are returned to the management front desk in step S207. If it is determined in step S206 that the database operations are not all successful, then it is determined in step S208 whether the database operations are partially successful. If it is determined in step S208 that the database operation is partially successful, then in step S209, a difference information record is generated in the difference information memory table and the difference information file for the database whose operation failed, and records a query statement or stored procedure executed by the database when the difference information is generated At step S210, a partially successful 'indication' is returned to the management front desk. If it is determined in step S208 that the database operation is not partially successful, then all failure indications are returned to the management front desk in step S211.

The structure of the difference information file is described below with reference to FIG. 3. The difference information file is stored in a binary format. 'Each difference information corresponds to a record. To ensure fast reading and writing of records, the difference information record is set to a fixed-length structure. The difference information includes a retry identification code (RedoID) field that represents a unique identifier of the difference information record, and the retry identification code indicates which part of the difference information file the management backstage extracts the related difference information. The Retry ID field is followed by a Flag field, which identifies which databases need to retry the difference information. The length of the tag field can be, for example, four bytes, a total of 32 bits, and each bit corresponds to a database. In the present invention, it is set as follows: if the value of a bit of the mark is "1", it means that the database corresponding to the bit has generated difference information; if the value of a bit of the mark is "0", it means that the The database corresponding to the bit does not generate difference information, that is, the database is successfully executed. The tag field is followed by the command string (C ₀ mmandString). The command string refers to the query statement or stored procedure executed by the database when the difference information is generated. Once the management background finds that a command is partially successful, it knows which database or databases failed to execute, and it knows what the command was then. Therefore, the management background can accurately add a difference information record to the memory difference information memory table and the difference information file. Then, a partial success indication is returned to the front desk, and the partial success indication carries a retry identification code field.

After receiving the partial success indication from the management background, the management front end outputs the difference information due to the partial success and alerts the operator. The operator locates the database that generates the difference information. The operator judges whether the database is operating normally, whether the connection is normal, and so on. If the operator finds that the network cable of the host where the database is located is disconnected, reconnect it, and then return to the management front desk, enter the retry command, the operator sends a "retry" command to the management front desk, and the management front desk retry carrying The command of the identification code is sent to the management background.

The process of eliminating database difference information is described with reference to FIG. 4. In step S401, the management background receives the retry command carrying the retry identification code sent by the management front desk, and the management background finds the specific corresponding to the retry identification code in the difference information file according to the received retry identification code in step S402. The difference information is obtained from the corresponding difference information, a mark field of the difference information record, a stored procedure or SQL command to be executed, and information about which database or databases need to execute the stored procedure or SQL command. Then, in step S403, the management background sets the initial environment of the corresponding worker thread according to the indication of the flag field. In step S404, the corresponding worker thread is unlocked according to the flag field, and the stored procedure or SQL command is called to a database or some databases according to the pointer of the difference information, and a retry operation is performed. In step S405, wait for all the worker threads to finish processing and return to the waiting state. In step S406, the processing result of the database is checked to determine whether all the retry operations are successful. If it is determined in step S406 that all the retry operations are successful, the difference information recorded in the difference information memory table and the difference information file is deleted in step S408, and all success indications are returned to the management front desk in step S408. If it is determined in step S406 that the retry operation is not all successful, then it is determined in step S407 whether the retry operation is partially successful. If it is determined in step S407 that the retry operation is partially successful, then the difference information memory table and the difference information file are updated in step S409. The difference information recorded in the file is deleted, and the difference information corresponding to the database with successful retry operation is deleted, and a partial success indication is returned to the management front desk at step S410. If it is determined in step S407 that all the retry operations have failed, an indication of all failures is returned to the management front desk in step S411.

According to the present invention, a polling thread is set in the management background, and the polling thread can periodically retry all the difference information in the system without inputting a retry command in the management foreground to prevent the cumulative growth of the difference information. Industrial applicability

In the multi-database application system of the present invention, the management station provides an interactive operation interface. Whenever a new piece of difference information appears, the management background can promptly notify the management front desk, so that the operator can locate the fault problem as quickly as possible, and respond to the difference. Message and try again.

The difference information generated by multiple databases is uniformly managed by the only management background. The management front desk simply enters a retry request for a piece of difference information, and the management back desk implements the retry operation and returns the operation result to the management front desk. Obtain the difference information generated by all the databases in the system, so that the operator can quickly and comprehensively locate the problem and eliminate the difference information as soon as possible.

So far, the specific implementation process of the present invention has been described in conjunction with the embodiments. The method for eliminating database difference information of the present invention has obtained satisfactory results when it is run in a distributed multi-database application system, and provides a method for universally and rapidly eliminating database difference information for a distributed multi-database application system.

Claims

Claim

A method for eliminating database difference information in a multi-database application system by using a management configuration module, wherein the management configuration module includes a management background and a management foreground, and is characterized in that the method includes the following steps:

(b) executing structured query language (SQL) instructions or stored procedures in the corresponding database,

(c) check the structured query language instructions of the database or the processing results of stored procedures to determine whether the database operation was successful,

(d) if it is judged in step (c) that the database operation is partially successful, generating difference information with a retry identification code for the database where the operation failed, and recording the difference information in the difference information memory table and difference information In the file, a partial success indication is returned to the management front desk.

(e) the management background receives a retry command carrying a retry identification code sent by the management front desk, and finds the difference information corresponding to the retry identification code in the difference information file according to the received retry identification code,

(f) obtaining the mark field of the difference information record and the stored procedure or structured query language instruction to be executed from the corresponding difference information, and calling the stored procedure or a certain database according to the pointer of the difference information or Structured query language instructions and perform retry operations,

(h) If it is determined in step (g) that all the retry operations are successful, delete the difference information recorded in the difference information memory table and the difference information file, and return all success instructions to the management front desk

2. The method for eliminating database discrepancy information according to claim 1, characterized in that before executing step (b) of the structured query language instruction or stored procedure in the corresponding database, further comprising receiving a command request from the foreground and The step of converting the command request into a corresponding stored procedure or structured query language instruction.

3. The method for eliminating database discrepancy information according to claim 1, wherein the step (c) of determining whether the database operation is successful further comprises determining whether the database operation is all successful or not. In the failed step, if the database operations are all successful, all the indications of success are returned to the management foreground; if the database operations are all failed, the indications of all failures are returned to the management foreground.

4. The method for eliminating difference information in a database according to claim 1, wherein the step (i) of updating the difference information recorded in the difference information memory table and the difference information file further comprises deleting and retrying successful operations. The difference information corresponding to the database.

The method for eliminating database difference information according to claim 1, characterized in that, if it is determined in step (i) that all retry operations have failed, further comprising the step of returning an indication that all retry operations have failed to the management front desk.

6. The method for eliminating database difference information according to any one of claims 1 to 6, wherein the recorded difference information includes a retry identification code field, a flag field, and a database execution when the difference information is generated Structured query language instructions or stored procedures.