CN100336343C - Method for keeping multiple data copy consistency in distributed system - Google Patents

Abstract

The invention discloses a method for maintaining the consistency of multiple data copies in a distributed system. By expanding and defining the Qx interface between the manager and the agent and specifying the operating mechanism of the agent, the background manager and the device side are realized. The information exchange and coordination between agents and MCUs ensures the consistency of data copies during the entire information exchange process, especially when there is a failure in the execution process. The core of the present invention is that the agent first caches the historical data copy, and then deletes the cached copy after the execution is successful, and performs consistent rollback processing according to the cached data copy if the execution fails. The method described in the present invention improves the deficiencies in the one-to-many command message response mechanism in the prior art in principle, and avoids the problem of data copy inconsistency between the network management system and the device side, thus avoiding the problem of the device due to the distributed system The hidden dangers of network operation security caused by the imperfect command transmission mechanism ensure the stable and reliable operation of the equipment.

Description

A method for maintaining the consistency of multiple data copies in a distributed system

technical field

The present invention relates to a distributed communication system, in particular to a distributed communication system which has a complicated front-background division of labor and the front-end must depend on the configuration of the back-end to work.

Background technique

The transmission network (including the network management system) is a complex large-scale distributed system. Its typical network structure model is shown in Figure 1, which is divided into two parts: the network management side and the equipment side; the network management side includes GUI (graphical user interface), MANAGER There are two parts (manager) and the device side includes AGENT (agent) and MCU (micro control unit). Among them, the relationship between MANAGER and AGENT is one-to-many, and the relationship between AGENT and MCU is also one-to-many. In this distributed system, the GUI-MANAGER transmits messages through the F port operation, and the MANAGER-AGENT transmits messages through the Qx port operation, and the AGENT and each MCU (micro control unit) are on the same backplane through the S port command to communicate.

The transmission network works according to the configuration of the network management system. At the same time, in order to work without the network management system, the network has a copy of the configuration data in both the AGENT (agent) and the MCU (micro control unit). In this way, there are actually three A copy of the network configuration data is stored in the MANAGER (manager), AGENT (agent) and MCU (micro control unit) on the network management side respectively. Ensuring the consistency of these three data is the basic prerequisite for the stable and reliable operation of the entire transmission network.

Initially, the working process of the network management and device side is as follows: the graphical user interface sends the F port operation to the manager; the manager sends the corresponding one or more Qx port operations to the agent after analysis and saves a copy of the configuration data ; After the agent receives the Qx operation, it analyzes it, sends the corresponding S port commands to the MCU, and refreshes its own configuration data copy; the MCU also executes and refreshes the data copy according to the command issued by the agent . This way we call the no-response working mode.

The characteristic of the non-response working mode is that the operation message transfer between various functional entities does not require a response, as long as the command is sent, it is considered successful. The advantage is that it is simple to implement; the biggest disadvantage is that the upstream does not know the execution of the downstream. In the actual network configuration process, it is impossible to completely avoid the abnormal situation caused by the execution failure of a certain link, which often causes data inconsistency. In order to improve this shortcoming, the work process of the network management and equipment side is generally a response work mode.

This working method is characterized by response, that is, the upstream entity refreshes the configuration data corresponding to the command and issues the next command only after receiving a correct response, and stops further execution if it receives a wrong response , and consider the entire command to be a failure; the entire command is considered successful only when all decomposed commands are executed successfully. The working mode of the response guarantees the consistency of upstream and downstream information to a certain extent.

After careful analysis of the working mode of the response, there is still a congenital deficiency in the design: when one F port operation issued by the GUI to the manager corresponds to multiple Qx port operations issued by the manager to the agent at the same time, if the agent executes One of the Qx operations in the process fails, and the manager thinks that the entire F port operation fails; but we can find that the Qx operation executed by the agent before the failure and the S port command executed by the corresponding MCU have actually succeeded, then the corresponding The data copy of the device has also been refreshed; and the manager did not perform any data copy update because he believed that the F port operation failed, which resulted in some inconsistencies among the data copies of the manager, agent, and MCU. This data inconsistency caused by partial success and partial failure is a great safety hazard in engineering practice, and may cause serious adverse effects.

Contents of the invention

The problem to be solved by the present invention is the data inconsistency of each data copy caused by partial success and partial failure in the execution process of distributed system commands in the prior art, and a method for maintaining the consistency of multiple data copies in a distributed system is proposed .

A method for maintaining the consistency of multiple data copies in a distributed system includes the following processes:

Manager's process:

Step 1, the graphical user interface sends the F port operation to the manager;

Step 2, the manager analyzes after receiving it, and splits the operation into one or more Qx port operations to the agent;

Step 3. The manager sends the Qx port operations with serial numbers to the agent in turn;

Step 4: The agent receives and executes the Qx port operations in turn, and feeds back the operation response information to the manager after each Qx port operation is executed;

Step 5, if the manager receives the agent's operation failure response, then return an error message to the graphical user interface, and end this F port operation;

Step 6. Otherwise, update the corresponding data copy of this F port operation at the manager after receiving all the successful responses of this F port operation, and return a successful execution response to the graphical user interface, and end this F port operation ;

Proxy process:

Step A, the agent receives the Qx port operation issued by the manager;

Step B, before the agent executes the Qx port operation, it caches a copy of the data that will be changed by this Qx port operation;

Step C, the agent analyzes the Qx port operation, and splits it into one or more corresponding S port commands;

Step D, the agent sends the S port commands to the micro control unit in sequence;

Step E, the micro control unit receives and executes the S port commands in sequence, and feeds back the execution response information to the agent after each S port command is executed;

Step F. If the agent receives the successful response of all S port commands corresponding to this Qx port operation, update the data copy corresponding to this Qx port operation at the agent, and return this Qx port operation to the manager successful response;

Step G, if all Qx port operations corresponding to this F port operation are successfully operated at the agent, delete the data copy cached by the agent;

Step H, if the agent receives the execution failure response from the microcontroller unit, then

H1: The agent returns a response to the failure of this Qx port operation to the manager;

H2: According to the serial number of this Qx port operation, determine all Qx port operations corresponding to this F port operation on the agent, and determine all data copies cached by the agent during this F port operation;

H3: The agent restores all Qx port operation data copies cached as actual data copies;

H4: The agent constructs the corresponding S port command according to the cached data copy, and sends it to the micro control unit, and restores the data copy of the micro control unit to the state before the F port operation;

H5: The agent deletes the cached data copy;

The processing of the microcontroller unit:

Step 1: the micro control unit receives the S port command issued by the agent;

Step II: The micro-control unit analyzes the received S port command, and performs corresponding operations on the device according to the analysis result;

Step III: If the execution is successful, update the data copy corresponding to the command, return a successful execution response to the agent, and end the execution of the S port command;

Step IV: If the execution fails, return an execution failure response to the agent, and end the execution of the S interface command.

The above sequence number is identified in the message header of the Qx port operation.

The method described in the present invention improves the deficiencies in the one-to-many command message response mechanism in the prior art in principle, and avoids the problem of data copy inconsistency between the network management system and the device side, thus avoiding the problem of the device due to the distributed system The hidden dangers of network operation security caused by the imperfect command transmission mechanism ensure the stable and reliable operation of the equipment. Obviously, the present invention is not limited to the transmission network field, and the method described in the present invention can also be applied to other distributed systems with similar properties.

Description of drawings

FIG. 1 is a schematic diagram of a system structure of a distributed transmission network;

Fig. 2 is the processing flowchart of the administrator in the proposed method of the present invention;

Fig. 3 is the processing flowchart of agent in the proposed method of the present invention;

Fig. 4 is the working flow chart when agent executes abnormality in S port order among the present invention;

Fig. 5 is a processing flowchart of the micro control unit in the method proposed by the present invention.

Detailed ways

The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments.

FIG. 1 is a schematic diagram of a system structure of a distributed transmission network. As shown in Figure 1, the distributed transmission network system includes four parts: graphical user interface, manager, agent, and micro-control unit; the relationship between the manager and the agent is one-to-many, and the relationship between the agent and the micro-control unit It is also a one-to-many relationship. Generally speaking, the graphic user interface and the manager are called the background network management system, and the agent and the micro-control unit are the main body of the communication equipment. Data copies are stored in the manager, the agent and the micro-control unit respectively.

Fig. 2 is a flow chart of the administrator's processing in the method proposed by the present invention. As shown in Figure 2, the manager's processing process includes: 1. The graphical user interface sends an F port operation to the manager; 2. The manager analyzes and splits the command into one or more Qx to the agent 3. The manager sends the Qx port operations to the agent in turn, and indicates the serial number of the operation in the Qx operation message. For the last command, the serial number is 0xfefe; 4. First, judge whether all Qx All operations have been sent? 5. If it is not finished, then send the next Qx operation to the agent; 6. If all of them are finished, then update the corresponding data copy of this F port operation at the manager, and return a successful execution response to the graphical user interface, and end This F port operation. 7. The manager receives the response from the agent; 1) if it is a successful response, then go to step 4; 2) if it is a failed response, then directly return an error message to the graphical user interface, and end this F port operate. Wherein, the sequence number can be identified in the header of the Qx port operation, and the Qx interface command description format in the prior art is: command source address command destination address command code command unique identifier command execution result command body length command body

The Qx interface command description format after identifying the serial number among the present invention is: command source address command destination address command code command unique identifier Operation sequence number command execution result command body length command body

Fig. 3 is a process flowchart of the agent in the proposed method of the present invention. As shown in Figure 3, the processing process of the agent includes: 1. The agent receives the Qx port operation issued by the manager; 2. Before the agent executes the Qx port operation, it first caches the data that will be changed by the Qx port operation 3. The agent analyzes the operation of the Qx port and splits it into corresponding one or more S port commands; 4. The agent sends the S port commands to the micro control unit in turn; 5. First judge whether all S port commands All operations are sent? 6. If it is not finished, send the next S port command to the micro control unit; 7. If all of them are finished, update the corresponding data copy of this Qx port operation at the agent, and then return a successful execution response to the manager; 8. If all the Qx port operations corresponding to the F port operation are completed, delete the data corresponding to the Qx port cached by the agent. 9. If the agent receives the execution failure response of the micro-control unit, then execute the work flow when the S port command is executed abnormally as shown in Figure 4: 1. At first return the response to the manager's failure to execute the Qx port operation; 2. According to the operation sequence number in the Qx operation message, determine all the Qx operations corresponding to the F port operation on the agent, and determine all data copies cached by the agent during this F port operation; 3. The agent Restore all Qx port operation data copies in the cache as the actual data copies of the system; 4. The agent constructs the corresponding S port command according to the cached data copy, and sends it to the micro control unit to restore the data copy of the micro control unit as The state before this F port operation; 5. The agent deletes the cached data.

Fig. 5 is a processing flowchart of the micro control unit in the method proposed by the present invention. As shown in Figure 5, the processing process of the micro-control unit includes: 1. The micro-control unit receives the S-port command issued by the agent; 2. The micro-control unit analyzes the S-port command, and performs corresponding actions on the device according to the analysis result. Operation; 3. If the execution is successful, update the data copy corresponding to the command, and return a successful response to the agent, ending the execution of the S-port statement. 4. If the execution fails, return a failure response to the agent, and end the execution of the S port command.

The detailed implementation of the method of the present invention will be described below in conjunction with a specific embodiment.

Combined with the transmission network system shown in Figure 1, it is assumed that one F port operation issued by the GUI corresponds to three Qx port operations issued by the manager to the agent, and each Qx port operation corresponds to three S port commands. , in order to ensure the consistency of the front and back data copies when an exception occurs, the main working process of the present invention is as follows:

1. The graphical user interface sends an F port configuration operation to the manager, and the manager analyzes, splits and converts it into three Qx operations to the agent;

2. The manager identifies the sequence number of the operation in the message header, and sends the first Qx operation to the agent;

3. The agent receives the Qx operation issued by the manager, (assuming that the data copy corresponding to the Qx operation is A, and the data copy is A1 before this operation is executed), and first caches A1;

4. Analyze the Qx operation, split and convert it into three S port commands to the MCU;

5. The agent sends the first S port command to the corresponding MCU in sequence;

6. The corresponding micro-control unit receives the S port command issued by the agent and executes it. After success, it changes its own data copy and returns a successful response to the agent;

7. The agent receives the successful response returned by the micro control unit, and continues to issue the second S port command;

8. The MCU receives the second S port command and successfully updates the data copy and returns success;

9. The agent receives the successful response returned by the micro control unit, and continues to issue the third S port command;

10. The MCU receives the third S port command and successfully updates the data copy and returns success;

11. The agent receives the successful response returned by the micro control unit, judges that the three S port commands corresponding to this Qx operation have been executed successfully, refreshes its own data copy, and then the data copy is updated from A1 to A2; Answer and end this Qx operation;

12. After receiving the successful response of the first Qx operation from the agent, the manager will continue to issue the second Qx port operation to the agent;

13. The agent receives the second Qx operation issued by the manager, (assuming that the data copy at the agent corresponding to the Qx operation is B, and the data copy is B1 before this operation is executed), and first caches B1;

14. The second Qx operation also corresponds to three S port commands; the agent sends the first S port command to the micro control unit, the micro control unit executes successfully, updates the data copy and returns a successful response;

15. The agent receives a successful response and sends the second S port command to the corresponding MCU;

16. The micro-control unit executes the S port command. Due to the failure of the device, the command fails to execute, so an error response is returned to the agent;

17. The agent receives the reply that the MCU executes an error, and returns the second reply that the execution of the Qx operation fails to the manager; at the same time, it performs rollback processing to ensure the consistency of the data copy:

1) The agent rolls back the data corresponding to the first Qx port operation from copy A2 to A1 before execution;

2) Based on the data A1, construct three S port commands corresponding to the first Qx port operation, and send them all to the micro control unit, so as to restore the data copy on the micro control unit to the state before this F port operation;

3) Based on the data copy B1 corresponding to the second Qx port, construct the corresponding S port command, and send the first one to the corresponding micro-control unit to restore the data copy on the micro-control unit;

18. When the manager receives the agent's execution error response message, he does not update any data, and directly regards the F port operation as execution failure, and returns it to the graphical user interface.

Claims (2)

1, keep the method for a plurality of data copy consistencies in the distributed system, it is characterized in that comprising following processing procedure:

Gerentocratic processing procedure:

Step 1, graphical user interface issue the F mouth and operate to the manager;

Step 2, manager analyze after receiving, and this operation is split into the one or more Qx mouth operation to the succedaneum;

Step 3, manager are handed down to the succedaneum successively with serialized Qx mouth operation;

Step 4, succedaneum receive and carry out Qx mouth operation successively, after the Qx mouth operation that whenever is finished all to manager's feedback operation response message;

If step 5 manager receives succedaneum's operation failure and replys, then return wrong message and give graphical user interface, and finish this F mouth operation;

Step 6 otherwise after whole operation successful respond of receiving the operation of this F mouth, upgrade this F mouth and operate in the manager and locate corresponding data trnascription, and return replying of running succeeded to graphical user interface, finish this F mouth operation;

Succedaneum's processing procedure:

The Qx mouth operation that steps A, succedaneum's receiving management person issue;

Step B, succedaneum carry out before this Qx mouth operation, the copy of the data that this Qx mouth operation of buffer memory can be changed;

Step C, succedaneum analyze this Qx mouth operation, and split into the order of corresponding one or more of S mouth;

Step D, succedaneum are handed down to micro-control unit successively with the order of S mouth;

Step e, micro-control unit receive and carry out the order of S mouth successively, all feed back the execution response message to the succedaneum after the S mouth order that whenever is finished;

If step F succedaneum receives whole S mouth command execution successful respond of this Qx mouth operation institute correspondence, then upgrade this Qx mouth and operate in the succedaneum and locate corresponding data trnascription, return this Qx mouth to the manager and operate successful respond;

If this F mouth of step G is operated pairing whole Qx mouth operation and all located to operate successfully the succedaneum, then delete succedaneum's data in buffer copy;

If step H succedaneum receives the execution failure of micro-control unit and replys, then

H1: the succedaneum returns this Qx mouth operation failure to the manager and replys;

H2: according to the sequence number of this Qx mouth operation, determine that this F mouth operates in pairing all the Qx mouth operations of succedaneum, and all data trnascriptions of definite succedaneum institute's buffer memory in this F mouth operating process;

H3: all Qx mouth operating data copies that the succedaneum recovers buffer memory are actual data trnascription;

H4: the succedaneum constructs corresponding S mouth order, and is handed down to micro-control unit according to the data in buffer copy, and the data trnascription that recovers micro-control unit is the state before this F mouth operation;

H5: the succedaneum deletes the data in buffer copy;

The processing procedure of micro-control unit:

Step I: micro-control unit receives the S mouth order that the succedaneum issues;

Step II: micro-control unit is analyzed the S mouth order that receives, and according to analysis result equipment is carried out corresponding operation;

Step II I: if run succeeded, then update command institute corresponding data copy returns replying of running succeeded to the succedaneum, finishes this S mouth command execution;

Step IV: if carry out failure, then return and carry out replying of failure, finish this S mouth command execution to the succedaneum.

2, method according to claim 1 is characterized in that: described sequence number identifies in the heading of Qx mouth operation.

Images