WO2013000135A1

WO2013000135A1 - Locking method and system for distributed messages

Info

Publication number: WO2013000135A1
Application number: PCT/CN2011/076570
Authority: WO
Inventors: 崔文生
Original assignee: 中兴通讯股份有限公司
Priority date: 2011-06-29
Filing date: 2011-06-29
Publication date: 2013-01-03

Abstract

The present invention discloses a locking method and system for distributed messages, in both of which locking templates for the messages can be saved unifiedly, and when carrying out operations needing coordination of multiple service modules, all the correlative distributed nodes perform locking on the distributed messages according to the locking templates for the messages. The distributed messages locking technology of the present invention can perform distributed locking on correlative user operations (i.e. based on messages), so as to implement coordination among different modules to execute operation-exclusive locking, and consequently guarantee the consistency of the data; at the same time, the message-based locking mechanism can avoid the locking-invasive impact on the service modules.

Description

Method and system for locking distributed message

The present invention relates to the field of communications, and in particular, to a method and system for locking a distributed message. Background technique

The network equipment management software bears the heavy responsibility of the operation and maintenance of the network communication equipment. It has high requirements on data reliability and data consistency. Once the data is inconsistent, it will cause serious consequences such as business interruption. Network device management software has multiple types of operational user interfaces, such as: GUI (Graphical User Interface) client (multi-user), northbound interface, and southbound interface. In this multi-user, multi-interface concurrent operation mode, a certain mechanism is required to ensure data consistency caused by concurrent access operations; and, when distributed deployment of network device management software, it is required to cooperate on different hosts. Module to ensure data consistency.

The current solution is that each distributed service module applies for locking to a lock management arbiter, and solves the distributed reliability problem by using a heartbeat mechanism between distributed nodes, but the solution is more invasive to the service module, when added When a cooperative multi-service module processes a scenario, the related modules not only need to be explicitly locked, but also because the locking modules are relatively scattered, which is not conducive to management and analysis of the locked scenario. Summary of the invention

In view of this, the main object of the present invention is to provide a method and system for locking distributed messages, which are mutually exclusive and locked in cooperation with different modules to avoid data inconsistency.

In order to achieve the above object, the technical solution of the present invention is achieved as follows:

A method for locking a distributed message, which has a message locking template stored in a unified manner, and the method further includes:

When performing operations that require multi-service module assistance, all relevant distributed nodes are Message locking template, locking for distributed messages.

The method further includes:

When each distributed node is started, the message locking template is obtained as a basis for determining the lock. The distributed node receives the lock request, and the process of locking the distributed message includes:

The distributed node queries all distributed nodes to the distributed system and issues a lock request to all distributed nodes, and the distributed node that receives the lock request locks the related distributed messages.

The method further includes:

The message scheduling center of each distributed node listens to other distributed node states; and/or, during the locking process, the message scheduling center of the distributed node performs a lock determination on each message to be scheduled, and according to a predefined scheduling policy Process related distributed messages.

The method further includes:

After the operation of the multi-service module assistance is completed, an unlock request is initiated to the distributed node that initiates the lock, and the distributed nodes are notified to unlock according to this; after the unlocking is successful, each distributed node schedules the message according to the normal logic.

A locking system for a distributed message, the system comprising a message locking template maintenance unit, a message locking template acquiring unit, and a locking execution unit; wherein

The message locking template maintenance unit is configured to uniformly store a message locking template;

The message locking template acquiring unit is configured to acquire a message locking template when the distributed nodes are started, as a judgment basis for locking;

The lock execution unit is configured to control all distributed nodes to lock the distributed message according to the message locking template when performing operations requiring multi-service module assistance.

The distributed node is further configured to:

At startup, the message locking template is obtained as a basis for determining the lock. The distributed node will receive a lock request when the operation of the multi-service module is required, and the lock execution unit is configured to:

The distributed node is controlled to query all distributed nodes to the distributed system, and issues a lock request to all distributed nodes, and triggers the distributed node that receives the lock request to lock the related distributed message.

The distributed node is further configured to:

Listen to other distributed node states; and / or,

Each of the messages to be scheduled is locked and judged, and related distributed messages are processed according to a predefined scheduling policy.

The distributed node is further configured to:

After the operation of the multi-service module assistance is completed, the unlock request is received, and the distributed nodes are notified to unlock according to this; after the unlocking is successful, each distributed node schedules the message according to the normal logic.

The distributed message locking technology of the present invention can perform distributed locking on associated user operations (ie, based on messages), so as to implement mutual exclusion and locking of different modules, thereby ensuring data consistency; The locking mechanism of the message avoids the intrusive impact on the business module. DRAWINGS

1 is a schematic diagram of acquiring a lock template by each distributed node according to an embodiment of the present invention; FIG. 2 is a schematic diagram of initiating a lock according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of initiating unlocking according to an embodiment of the present invention; FIG.

4 is a schematic diagram of a locking process of a distributed message according to an embodiment of the present invention;

FIG. 5 is a diagram of a locking system of a distributed message according to an embodiment of the present invention. detailed description

When locking a distributed message, you can do the following: The first step: storing a message locking template in a central database, each distributed node can access the database to obtain the locking information in the message locking template.

The message locking template can contain at least one of the following information:

Lock ID, used to uniquely identify the lock use;

a service module ID, a unique identifier of a service module that performs message processing; the service module is capable of processing a message, and when a lock is locked, a processing policy of the responsible message ID needs to be defined; the message ID is a unique identifier of the message The message is a message to be processed by the service module. The message scheduling policy indicates that the policy for processing a message in the case of locking may include at least one of the following policies:

Discard and prompt an error to indicate that the operation is forbidden at the current time to ensure data consistency;

Waiting, used to indicate waiting during the lock, until it is unlocked, can be scheduled;

Direct scheduling, used to indicate that processing can still be performed during locking;

Other custom policies, such as: Message IDs and modules not included in the lock template information are not affected by the lock.

Step 2: When each distributed node starts, each distributed node obtains the message-locked template from the central database. When the message dispatching center in the distributed node dispatches the message to each service module, the obtained message locking template is used as the locking judgment basis, as shown in FIG.

Step 3: When performing operations that require multi-service module assistance, the message dispatching center in the distributed node will receive a lock request (indicating the lock ID), and the message dispatching center queries the distributed environment management server in the distributed system. All distributed nodes, and issue a lock request to all distributed nodes, the distributed nodes that receive the lock request will lock the related distributed messages, as shown in Figure 2.

It should be noted that the message scheduling center of each distributed node needs to monitor other distributed node states: If a new distributed node joins, the distributed node that initiates the lock notifies the newly added distributed node to lock;

If there is a distributed node locking failure, when the distributed node is restarted, the distributed node that initiates the locking notifies the distributed node to lock;

If the distributed node that does not initiate the locking fails to detect the failure of the distributed node that initiates the locking, the locking of the distributed node is cancelled;

In the meantime, the distributed node that initiated the lock can query the lock status to handle the abnormal situation. Step 4: During the locking period, the message dispatching center of the distributed node needs to lock the message to be scheduled:

Get the service module ID of the processed message;

Find the corresponding lock ID from the message locking template according to the service module ID+message ID;

If the lock corresponding to the found lock ID is in the locked state, the related distributed message is processed according to the predefined scheduling policy.

Step 5: After the operation of the multi-service module assistance is completed, an unlock request (indicating a lock ID) is initiated to the message dispatching center in the distributed node that initiates the lock, and the message dispatching center notifies each distributed node to unlock. After the unlocking is successful, each distributed node schedules messages according to normal logic, as shown in Figure 3.

Specifically, Figure 1 depicts the process by which each distributed node acquires a locked template. Among them, distributed nodes include:

Distributed node management (E1), responsible for providing status information of each distributed node;

The message dispatching center (E2) is responsible for message scheduling for the service processing module (E3) in the distributed node, processing the service processing module (E3) to apply for locking or unlocking, and monitoring the status of each distributed node to the service processing module. (E3) providing lock status information;

The service processing module (E3), which is responsible for processing the message, can initiate an application for locking or unlocking to the message dispatching center (E2), and can monitor the locked state. After each distributed node is started, execute S1: Read the lock template information to obtain the lock information in the message lock template.

Figure 2 depicts the process by which a distributed node initiates a lock, including the following steps:

S1 lock: The service processing module (E3) initiates a lock request to the message dispatch center (E2); S1.1 lock state: the service processing module (E3) starts to monitor the lock state from the message dispatch center (E2);

S2 queries the distributed node: The message dispatching center (E2) manages all distributed node information in the system to the distributed node management (E1);

S2.1 node status: The message scheduling center (E2) starts to monitor the status of each distributed node; S3 notifies the lock: The message dispatch center (E2) sends a lock request to the message dispatch center (E2) in other distributed nodes. After receiving the lock request, the message dispatching center (E2) of each distributed node enters the locked state;

S4 scheduling message: During the locking period, the message scheduling center of the distributed node needs to lock the message to be scheduled.

Figure 3 depicts the process by which a distributed node initiates an unlock, including the following steps:

S1 unlocking: The service processing module (E3) initiates an unlock request to the message dispatching center (E2);

S1.1 lock status: The service processing module (E3) monitors the lock status from the message dispatch center (E2);

S2.1 node status: The message scheduling center (E2) monitors the status of each distributed node;

The S3 notification is unlocked: The message dispatching center (E2) sends an unlock request to the message dispatching center (E2) in the other distributed nodes. After receiving the unlocking request, the message dispatching center (E2) of each distributed node unlocks to enter the unadded Lock state

S4 scheduling message: During the unlocked period, the message scheduling center of the distributed node is normally scheduled to cancel Interest.

The invention will be further described below with specific application examples.

There are three distributed nodes in the distributed operation of the network management software: NodeA, NodeB, and NodeC. The expansion module is deployed on the NodeA. Module 1 and module 2 are deployed on the NodeB and the NodeC. It is required that module 1 and module 2 prohibit execution of messages M1, M2.

When locking a distributed message, you can do the following:

Step 1: Define a message lock template, which includes:

Lock ID

Business module ID;

Message ID;

Scheduling strategy;

The message locking template can be stored in a database, and the specific content can be as shown in Table 1:

Table 1

The second step: NodeA, NodeB, and NodeC respectively obtain the above message locking template from the central database.

Step 3: Before the expansion module on the NodeA performs the expansion, apply for locking, and the message scheduling center in the NodeA initiates the locking request (expansion lock). The message scheduling center queries all the distributed environment servers in the distributed system. Distributed nodes: NodeA, NodeB, and NodeC, and send a lock request to the NodeB and NodeC. The NodeB and NodeC that receive the lock request will be directed to the phase. The closed distributed message is locked.

The fourth step: During the lock period, if the message scheduling center of the NodeB receives the message M1 that needs to be scheduled to the module 1, the lock determination is performed:

Get the business module ID of the processed message: module 1;

According to the modules 1 and M 1 , the corresponding lock ID is found in the message locking template: expansion lock; because the found expansion lock is in the locked state, the message M1 is discarded according to the predefined scheduling policy and an error is prompted. .

Step 5: After the expansion is completed, the expansion module in the NodeA initiates an unlock request to the message dispatching center in the distributed node, and the message dispatching center notifies the NodeB and the NodeC to unlock. After the unlocking is successful, each distributed node schedules messages according to normal logic.

According to the foregoing technical description, the operation of the present invention for locking a distributed message may represent a process as shown in FIG. 4, and the process includes the following steps:

Step 410: Unified storage of the message locking template.

Step 420: When each distributed node starts, obtain a message locking template, and determine the basis for locking.

Step 430: When performing operations requiring multi-service module assistance, all distributed nodes lock the distributed messages according to the message locking template.

It should be noted that step 410 is a storage operation performed in advance, and does not need to be repeated every time the lock is performed. Step 420 is a pre-acquisition operation that does not need to be repeated each time the lock is performed; of course, if the message lock template sends a change, it needs to be re-acquired to implement the update.

In order to ensure that the above technical description and operation ideas can be smoothly implemented, the settings shown in Figure 5 can be performed. Referring to FIG. 5, FIG. 5 is a diagram of a locking system for a distributed message according to an embodiment of the present invention. The system includes a connected message locking template maintenance unit, a message locking template acquiring unit, and a locking execution unit. The message locking template maintenance unit may be a central database or other The function entity is set independently or set in the distributed node; the message lock template acquisition unit and the lock execution unit can be set in the distributed node.

In practical applications, the message locking template maintenance unit can uniformly store the message locking template. The message locking template obtaining unit can acquire the message locking template when the distributed nodes are started, as the basis for determining the locking. The lock execution unit can control all distributed nodes to lock the distributed message according to the message locking template when performing operations requiring multi-service module assistance.

In summary, the method of locking the distributed message of the present invention can perform distributed locking on the associated user operation (ie, based on the message), so as to achieve mutual exclusion between different modules. Locked to ensure data consistency; At the same time, based on the message locking mechanism, the intrusive impact on the business module is avoided.

The above are only the preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalents, and improvements made within the spirit and scope of the present invention should be included in the scope of the present invention. Inside.

Claims

Claim

A method for locking a distributed message, which has a message locking template stored in a unified manner, and the method further comprises:

When an operation requiring multi-service module assistance is performed, all related distributed nodes lock the distributed message according to the message locking template.

2. The method according to claim 1, wherein the method further comprises:

When each distributed node is started, the message locking template is obtained as a basis for determining the lock.

3. The method according to claim 1, wherein, when performing an operation requiring assistance from a multi-service module, the distributed node receives a lock request, and the process of locking the distributed message includes:

4. The method according to claim 1, wherein the method further comprises:

The method according to any one of claims 1 to 4, wherein the method further comprises: after the operation requiring the assistance of the multi-service module ends, initiating an unlock request to the distributed node that initiates the locking, according to which The distributed nodes are unlocked; after the unlocking is successful, each distributed node schedules messages according to normal logic.

6. A locking system for a distributed message, the system comprising a message locking template maintenance unit, a message locking template acquiring unit, and a locking execution unit; wherein

The message locking template acquiring unit is configured to acquire a message when each distributed node starts Locking the template, as the basis for determining the lock;

The system according to claim 6, wherein the distributed node is further configured to: acquire the message locking template at the time of startup, as a locking determination basis.

8. The system according to claim 6, wherein, when an operation requiring multi-service module assistance is performed, the distributed node receives a lock request, and when the lock execution unit locks a distributed message, Used for:

9. The system according to claim 6, wherein the distributed node is further configured to: monitor other distributed node states; and/or,

10. The system according to any one of claims 6 to 9, wherein the distributed node is further configured to: