WO2010072085A1

WO2010072085A1 - Distributed system as well as method and apparatus for managing service data of distributed system

Info

Publication number: WO2010072085A1
Application number: PCT/CN2009/073936
Authority: WO
Inventors: 刘彦夫; 孙永乐; 张丽军
Original assignee: 华为技术有限公司
Priority date: 2008-12-25
Filing date: 2009-09-15
Publication date: 2010-07-01
Also published as: CN101442435A; CN101442435B

Abstract

A distributed system as well as a method and an apparatus for managing service data of the distributed system are disclosed, which solve the problem that the distributed system couldn't balance the load of each service node dynamically in the existing technology. The method comprises obtaining the current load information of service nodes; sending the load information to a central management unit; the central management node receiving the load information sent periodically by the service nodes; searching the overloaded node and the light-load node according to the load information; determining whether the part of the service data in the overloaded service node can be transferred or not; if it can be transferred, sending the transfer controlling signal to the overloaded service node, in which the transfer control signal controls that the overloaded-service node transfers the part of service data to the proper light-load service node.

Description

Business data management method, device and distributed system for distributed system This application claims a service data management method, device and distribution filed on December 25, 2008, with the application number of 200810189103. The priority of the Chinese application of the "System" is incorporated herein by reference.

Technical field

The present invention relates to the field of distributed technologies, and in particular, to a distributed system, a method and apparatus for processing a system when the load is too heavy. Background technique

Every device in a distributed system can be seen as a node that is equal in processing and control capabilities, and these nodes often work together to complete a high-level business data. deal with. The process is as follows: When one of the nodes receives a high-level business data, it can divide the business data into several parts and distribute the parts to other nodes for processing, that is, each node processes the high-level Part of the business data to speed up the completion of this high-level business data.

For each node, it has its own resource configuration, which results in the processing of the processed business data, some nodes processing fast, some nodes processing slow; and according to the processed business data Features: Some nodes process a large amount of business data, heavy load, and some nodes process less business data and load.

When these nodes cooperate with each other to complete the processing of a high-level business data, the inventors found that at least the following problems exist in the prior art: When one of the nodes is insufficiently configured due to resources or has a large amount of processed business data When the load is too heavy, the part of the high-level business data that the node should complete will be affected, so that the entire high-level business data cannot be successfully completed. That is to say: A heavily loaded node (ie, a heavy-duty node) becomes an obstacle to the distributed system's ability to handle business performance. Summary of the invention

Embodiments of the present invention provide a method and apparatus for processing a system when the load is too heavy. Make distributed The system automatically balances the load in each node to reduce the burden on the heavy-duty nodes.

Embodiments of the present invention adopt the following technical solutions:

A business data management method for a distributed system, comprising:

Receiving current load information of the service node periodically sent by the service node;

Finding the overloaded service node and the light load service node according to the load information;

Determining whether part of the service data in the discovered overloaded service node can be migrated;

And if it is determined that part of the service data in the overloaded service node can be migrated, sending, to the overloaded service node, a migration control signal, where the migration control signal controls the overloaded service node to migrate part of the service data to the discovered light On the service node.

A business data management method for a distributed system, comprising:

Regularly obtain the current load information of the service node;

Sending the load information to a central management node;

Receiving a migration control signal sent by the central management node, where the migration control signal includes at least address information of the light service node and an identifier of part of the service data that needs to be migrated;

And cutting part of the service data corresponding to the identifier in the service node to a light load service node corresponding to the address information.

A service data management device for a distributed system, comprising:

a receiving module, configured to receive current load information of the service node periodically sent by the service node; a reload search module, configured to search for a heavy load service node according to the load information received by the receiving module; and a light load search module, configured to receive the module according to the receiving module The received load information is used to find a light load service node; the determining module is configured to determine whether part of the service data in the overloaded service node found by the overloaded search module can be migrated;

a migration control module, configured to: when the determining module determines that part of the service data in the overloaded service node can be migrated, send a migration control signal to the overloaded service node, where the migration control signal controls the overloaded service node to part of the service data Migrate to the light load service node found by the light load lookup module. A service data management device for a distributed system, comprising:

a management module, configured to periodically obtain current load information of the service node;

a sending module, configured to send the load information acquired by the management module to the central management node, where the first receiving module is configured to receive a migration control signal sent by the central management node, where the migration control signal includes at least an address of the light load service node Information and the identification of part of the business data that needs to be migrated;

And a cutting module, configured to cut part of the service data corresponding to the identifier in the service node to a light load service node corresponding to the address information.

A distributed system, including:

a service node entity, configured to periodically acquire current load information of the service node and send the load information to a central management node;

a central management entity, configured to receive current load information of the service node periodically sent by the service node; search for the overloaded service node and the light load service node according to the load information, and determine whether the part of the discovered overloaded service node can be migrated Data, when it is determined that the service data in the overloaded service node can be migrated, sending, to the heavy-duty service node, a migration control signal, where the migration control signal controls the heavy-duty service node to migrate part of the service data to find the light-load service On the node;

The service node entity is further configured to: when receiving the migration control signal, cut part of the service data in the service node to the light load service node according to the migration control signal.

The service data management method, apparatus, and distributed system of the distributed system provided by the embodiments of the present invention have the following beneficial effects: when a heavy-duty service node occurs, the load on the overloaded service node can be dynamically adjusted, and part of the service data is migrated. On the light-duty service node, the beneficial effect of dynamically utilizing the resource configuration on each service node and dynamically balancing the load on each service node is achieved. When each service node in the distributed system processes a high-level service data together, it can avoid the situation that the entire high-level business cannot be successfully completed because the load of a single service node is too heavy.

DRAWINGS

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will be true. The drawings used in the examples or the description of the prior art are briefly introduced. It is obvious that the drawings in the following description are only some embodiments of the present invention, and are not creative to those skilled in the art. Other drawings can also be obtained from these drawings on the premise of labor.

1 is a flowchart of a service data management method of a distributed system on a central management node according to Embodiment 1 of the present invention;

2 is a flowchart of a method for managing service data of a distributed system on a service node according to Embodiment 1 of the present invention;

3 is a flowchart of a service data management method of a distributed system according to Embodiment 2 of the present invention; FIG. 4 is a block diagram of a service data management apparatus of a distributed system on a central management node according to Embodiment 3 of the present invention;

5 is a block diagram of a service data management apparatus of a distributed system on a service node according to Embodiment 4 of the present invention;

Figure 6 is a block diagram of a distributed system according to Embodiment 5 of the present invention.

detailed description

The technical solutions in the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present invention. It is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.

Example 1

This embodiment discloses a service data management method for a distributed system. The method is applicable to a central management node in a distributed system. As shown in FIG. 1, the method includes the following steps:

Let the central management node be A; a, b, and c be the service nodes.

101. Set a central management node in a distributed system, where the central management node receives current load information of the service node periodically sent by each service node.

The current load information of the service node may be that each service node periodically manages the section to the central node. The point is sent in the past, or it can be collected periodically by the central management node.

The specific execution may be: The central management node A receives the current load information al, bl, cl periodically sent from the a service node, the b service node, and the c service node.

102. The central management node searches for overloaded heavy-duty service nodes in each service node of the distributed system according to the received load information.

The specific execution may be as follows: When the central management node A reads the information in al, it is learned that the a service node is currently overloaded and the a service node is classified into the overloaded service node.

103. The central management node searches for light-loaded light-duty service nodes in each service node of the distributed system according to the received load information.

The specific execution may be: When the central management node A reads the information in b 1 and c 1 , it is learned that the b service node and the c service node are currently lightly loaded and the b service node and the c service node are classified into the light load service node. Within the scope.

104. The central management node determines, according to the received load information, whether part of the service data on the searched heavy load node can be migrated to the found light load service node, if the central management node determines part of the service data on the overloaded service node. If the central management node determines that part of the service data on the overloaded service node cannot be migrated to the light load node, step 106 is performed.

The specific executable can be: When the central management node A reads the a, it knows that part of the service data currently occupied by the service node on the a service node is X, and the central management node A judges according to the load information learned when reading the bl information. If the load corresponding to part of the service data X is migrated to the b service node, whether the b service node becomes heavily loaded, and if the b service node becomes heavily loaded, the load learned from reading the cl information is continued. If the load corresponding to the service data X is changed to the c service node, whether the c service node becomes heavily loaded, if the c service node is not heavily loaded, step 105 is performed; otherwise, step 106 is performed. .

105. The central management node sends a migration signal to the overloaded service node, where the migration control signal controls the overloaded service node to migrate part of the service data to the light load service node. The specific execution may be: The central management node A sends a migration control signal to the a service node, where the migration control signal instructs the a service node to migrate part of the data X to the c service node.

106, the process ends.

The 102, 103 in the foregoing embodiment does not limit the order of execution, and may even be performed simultaneously.

The invention described in the above embodiments solves the technical solution of the central management node that can periodically understand the current load information of each service node, and solves the problem in the prior art that the service nodes in the distributed system are in an equal position. Knowing the technical problems of the load information of other service nodes in time, and obtaining the beneficial effect of reducing the load for the heavy-duty service nodes in time according to the information received periodically.

This embodiment continues to disclose a service data management method for a distributed system. The method is suitable for a service node in a distributed system. As shown in FIG. 2, the method includes:

Let the central management node be A; aO be the management unit of a service node, bO be the management unit of the b service node, and cO be the management node of the c service node.

201. Set a management unit on each service node in the distributed system, where the management unit periodically obtains current load information of the service node.

The specific execution may be: aO periodically collects the current load information of a service node. al, bO periodically collects the current load information of the service node bl, and cO periodically collects the current load information cl of the c service node.

202. The management unit sends the current load information of the service node that is periodically acquired to the central management node.

The specific execution may be: aO sends the current load information a of the service node that is periodically collected to the central management node A, and the bO sends the current load information b of the b service node that is periodically collected to the central management node A, and the cO will collect periodically. c The current load information cl of the service node is sent to the central management node A.

The sending process of the load information may be periodically sent by the management unit to the central management node, or the central management node may periodically request the management unit. 203. The service node receives the migration control signal sent by the central management node, where the migration control signal includes at least the address information of the light-duty service node and the identifier of the part of the service data that needs to be migrated, for example: The IP address and the identifier of part of the service data that can be transferred in the overloaded service node.

The specific execution may be: a service node receives the migration control signal sent by the central management node A. The migration control signal indicates that the a service node migrates part of the service data X to the c service node and also contains the address of the c service node in the migration signal.

204. The heavy-duty service node cuts part of the service data corresponding to the identifier to a light-load service node corresponding to the address information according to the migration control signal sent by the central management node.

The specific execution may be: a service node cuts part of the service data X to the c service node according to the indication of the migration control signal.

The service data management method of the distributed system provided in this embodiment has the following beneficial effects: migrating part of the service data on the overloaded service node to the light load service node, and reducing the load of the heavy load service node, effective The use of resources for light-duty business nodes. In a distributed system, for each service node that is involved in processing a high-level service data, timely migration reduces the load of the heavy-duty service node, which is beneficial to improving the processing speed of the entire high-level service data and reducing the processing of the high-level service data. Time spent.

Example 2

This embodiment specifically describes the service data management method of the foregoing distributed system. As shown in FIG. 3, the method includes:

The central management node is set up in the distributed system, and the management unit is set on each service node.

301. The management unit on each node periodically obtains current load information of the node. The load information includes a current total load of the service node and a partial load corresponding to each current service data type. The sub-load corresponding to each service data type is: load information of a service component that processes the service data. 302. The management unit sends the current load information of the service node that is periodically acquired to the central management node.

The sending process of the load information may be periodically sent by the management unit to the central management node, or the central management node may periodically request the management unit.

303. The central management node receives current load information of the service node periodically sent by each service node.

The current load information of the service node may be that each service node periodically sends the past to the central management node, or may be collected periodically by the central management node.

In the distributed system in the prior art, since each service node is in an equal position in processing and control capabilities, it is impossible to timely understand the load information of other service nodes. The solution provided by this embodiment adopts the technical solution of the central management node that can periodically learn the current load information of each service node, so that the central management node can take timely measures according to the learned load information, so as to balance the service nodes. The beneficial effects of the load.

304. The total load of each service node has a preset first reload threshold, and the service node that exceeds the first reload threshold preset by the service node is a heavy load service node; The service component may also have a preset second reload threshold, which is a reload threshold for the service component running on the service node. For the service node where the service component with the second reload threshold is located, the service node where the service component exceeding the preset second reload threshold is located is also a heavy load service node, that is, the load exceeds the preset second weight. The service node with the threshold is also the overloaded service node. The central management node searches for the overloaded service node that meets the above description based on the received load information including the total load and the divided load. For the discovered overloaded service node, some of its business data can be migrated.

305. The total load of each service node may have a preset first light load threshold, and the service node that does not exceed the first light load threshold preset by the service node is a light load node; The service component may also have a preset second light load threshold. For the service node where the service component with the second light load threshold is located, the service component on the service node does not exceed the preset second light load. Threshold, the service node service whose total load does not exceed the preset first light load threshold is a light load node, that is, the load does not exceed the preset second light load threshold, and the total load does not exceed the preset first light load. The service node of the threshold is a light load service node. The central management node searches for the light load service node that meets the above description according to the received load information including the total load and the divided load.

In step 305, the preset first light load threshold may be equal to the preset first overload threshold, and the light load service node is: the service node whose total load does not exceed the preset first overload threshold; In step 305, the preset second light load threshold may also be equal to the preset second heavy load threshold. At this time, the light load service node is: the service node whose load does not exceed the preset second reload threshold, and the total load does not exceed the preset first reload threshold.

306. Determine, on the discovered overloaded service node, part of the service data that needs to be transferred. The overloaded service node may be a heavy-duty service node whose total load exceeds a preset first re-load threshold, or may be a heavy-duty service node whose load exceeds a preset second re-load threshold. The part of the service data that needs to be transferred may be a service group that has the most resources occupied by the overloaded service node, and may also be a service group that exceeds the second overload threshold.

307. When it is the case that the total load exceeds the preset first reload threshold of the overloaded service node, determine whether the light load service node is found, and if the light load service node is not found, determine that the heavy load cannot be migrated. And carrying out part of the service data in the service node and executing step 313; if the light load service node is found, determining whether the total load of the light load service node plus the load corresponding to part of the service data in step 306 exceeds the light load service The preset first reload threshold of the node, if the result of the determination is that the preset first reload threshold is exceeded, determining that part of the service data of the reload service node cannot be migrated in the light load service node and performing steps 313; If the result of the determination is not to exceed the preset first reload threshold, it is determined that part of the service data of the overloaded service node can migrate the light load service node and perform step 308;

When it is a heavy-duty service node whose load exceeds the preset second re-load threshold, it is determined whether the light-duty service node is found, and when the light-duty service node is not found, it is determined that the heavy-duty service node cannot be migrated. Part of the business data and perform step 313; if a light load service node is found, then Determining whether the load corresponding to the part of the service data in the step 306 exceeds the preset second reload threshold and the total load plus the load corresponding to the part of the service data in the step 306 exceeds the preset first weight a load threshold, if the load corresponding to the part of the service data does not exceed the preset second reload threshold and the total load plus the load corresponding to the part of the service data does not exceed the preset number a load threshold, determining that the part of the service data can be migrated to the light load service node and performing step 308; otherwise, determining that the part of the service data cannot be migrated to the light load service node and executing Step 313.

308. The central management node sends a migration signal to the overloaded service node, where the migration control signal controls the overloaded service node to migrate part of the service data to the light load service node. The migration control signal includes at least the address information of the light load service node and the identifier of the part of the service data that needs to be migrated.

Through the technical solution for searching and determining the heavy load service node and the light load service node provided in the above steps, it is possible to obtain that the target light load service node is not received when the part of the service of the overloaded service node is migrated. The load corresponding to the part of the service data becomes a beneficial effect of the overloaded service node.

309. The central management node modifies the address of the part of the service data to determine the address of the target light-load service node that can be migrated.

Modifying the address of part of the migrated service data to the address of the target light-load service node can obtain the part of the service data that cannot be lost because the part of the service data cannot be found when the part of the service data needs to be called for calculation. The benefits of its usability.

310. The overload service node receives the migration control signal sent by the central management node, where the signal includes at least an address of the light load node and part of the service data information that needs to be transferred in the overload service node.

311. According to the migration control signal sent by the central management node, the overloaded service node cuts part of the service data corresponding to the identifier to the light load service node corresponding to the address information, and performs step 316. 312. When the overloaded service node does not receive the migration control signal sent by the central management node, the overloaded service node automatically enters a backpressure state.

31 3 . The service component running on the overloaded service node in the back pressure state sends a "please later" signal to the request for the service data processing sent.

314. The service component running on the overloaded service node discards the unimportant or unreliable service data currently being processed, for example: some service data that is temporarily unused and can be re-acquired from other places when needed. The resources occupied by the discarded business data are recycled.

In this embodiment, by adopting a technical solution of discarding unreliable (such as a user datagram protocol) or unimportant (such as being re-acquired from other places or having a lower priority), it is achieved that migration cannot be performed. The benefits of mitigating the load of the heavy-duty service node can still be achieved by reclaiming resources.

315. Determine whether the load of the overloaded service node where the service component after discarding the unimportant or unreliable service data still exceeds the preset first reload threshold, if the preset first reload threshold is no longer exceeded. Step 316 is performed; if the preset first reload threshold is still exceeded, step 314 is performed.

The service component running on the overloaded service node has a preset second reload threshold. In this case, it is first determined whether the service component after discarding the unimportant or unreliable service data still exceeds the preset second weight. If the threshold is still exceeded, step 314 is performed if the preset second reload threshold is still exceeded. If the preset second reload threshold is no longer exceeded, the method described in step 315 is performed.

316. Stop the backpressure state and preferentially send a "serviceable" signal to the service component with a high priority and sent a service data processing request.

317, end the back pressure process or migration process.

In the existing distributed system, since each service node is in the same position in processing and control capabilities, when they jointly process a high-level service data, the load of other service nodes cannot be known in time. Information, so that a heavy-duty service node, that is, a heavy-duty node, cannot process it in time when it processes some of the high-level business data that it should complete. As a result, the high-level business data of the main item cannot be completed on time. The distributed system provided by this embodiment The service data management method, because the technical solution of the central management node that can periodically understand the load information of each service node is adopted, the obtained part of the service data on the overloaded service node can be timely when the overloaded service node appears. The technical effect of migrating to a light-duty service node. While effectively utilizing the resources of the light-duty service node, the beneficial effects of speeding up the high-level business data processed by each service node are also obtained.

Example 3

The embodiment of the present invention discloses a service data management device for a distributed system, which may be a central management node. As shown in FIG. 4, the device includes: a receiving module 41, a reload search module 42, and a light load searching module 43. Module 44, migration control module 45.

The receiving module 41 is configured to receive the current load information of the service node periodically sent by the service node; the overload search module 42 is configured to search for the overloaded service node according to the load information sent by the receiving module 41; The load information sent by the module 41 is used to find the light load service node; the determining module 44 is configured to determine whether the part of the service data in the overloaded service node can be migrated after the overloaded search module 42 finds the overloaded service node; 45. When the determining module 44 determines that part of the service data in the overloaded service node can be migrated, sending a migration control signal to the overloaded service node, where the migration control signal controls the overloaded service node to migrate part of the service data to The light load service node.

According to the total load information of the service node included in the load information received by the receiving module 41, the overload search module 42 uses the service node whose total load exceeds the preset first reload threshold as the overload service node, and the light load search module 43 A service node whose total load does not exceed the preset first light load threshold is used as a light load service node.

According to the total load information of the service node included in the load information received by the receiving module 41 and the load sharing information corresponding to each service data type, the reload search module 42 takes the service node whose load exceeds the preset second reload threshold as the weight. The service node, the light load search module 43 uses the service node whose load does not exceed the preset second light load threshold and whose total load does not exceed the preset first light load threshold as the light load service node. Further, the device further includes: a modifying module 46, configured to: after the migration module 45 sends the migration control signal to the overloaded service node, the address of the service node where the service data is located is changed to The address of the light load service node.

In the embodiment, the determining module 44 includes: a first determining unit 441, a first determining unit 442, and a second determining unit 443.

The first determining unit 441 is configured to determine part of the service data that needs to be migrated in the overloaded service node that is searched by the reloading search module 42. The first determining unit 442 is configured to determine whether the light load searching module 43 finds the light load service node. When the light load service node is not found, it is determined that the part of the service data in the overloaded service node cannot be migrated. The second determining unit 443 is configured to determine the total load when the light load search module 43 finds the light load service node. Whether the load corresponding to the part of the service data exceeds the preset first reload threshold; if the total load plus the load corresponding to the part of the service data does not exceed the preset first reload threshold, then determining And locating a part of the service data of the overloaded service node to the light load service node; if the total load plus the load corresponding to the part of the service data exceeds the preset first reload threshold, determining not A portion of the service data in the overloaded service node can be migrated to the light load service node.

Further, in the embodiment, the determining module 44 further includes: a second determining unit 444, a third determining unit 445, and a fourth determining unit 446.

The second determining unit 444 is configured to determine part of the service data that needs to be migrated in the overloaded service node that is searched by the reloading search module 42. The third determining unit 445 is configured to determine whether the light load searching module 43 finds the light load service node. When the light load finding module 43 does not find the light load service node, it is determined that the part of the service data in the overloaded service node cannot be migrated. The fourth determining unit 446 is configured to when the light load search module 43 finds the light load service node. Determining whether the load corresponding to the part of the service data exceeds the preset second reload threshold, and whether the total load plus the load corresponding to the part of the service data exceeds the preset first weight a load threshold; if the load corresponding to the part of the service data does not exceed the preset second reload threshold, and the total load plus the load corresponding to the part of the service data does not exceed the preset If the first reload threshold is And migrating part of the service data of the overloaded service node to the light load service node; otherwise, determining that part of the service data of the overloaded service node cannot be migrated to the light load service node.

The service data management device of the distributed system provided by the above embodiment, because of the receiving module

41 The received load information, so that the process of searching for the overloaded service node and the light load service node by the reload search module 42 and the light load search module 43 is facilitated, and in the prior art, in the distributed system, each service node is It is a technical problem that it is in an equal position and cannot know the load information of other business nodes in time.

Example 5

The embodiment of the present invention discloses a service data management device for a distributed system. The device is suitable for being deployed on a service node. As shown in FIG. 5, the device includes: a management module 51, a sending module 52, and a first receiving module 53. Module 54.

The management module 51 is configured to periodically obtain the current load information of the service node; the sending module 52 is configured to send the information periodically acquired by the management module 51 to the central management node, and the first receiving module 53 is configured to receive the central management node. And a migration control signal, where the migration control signal includes at least address information of the light load service node and an identifier of the part of the service data that needs to be migrated; the cutting module 54 is configured to: use the migration control signal received by the first receiving module 53 to send the service node The part of the service data corresponding to the identifier is cut to a light load service node corresponding to the address information.

Further, the apparatus suitable for deployment on the service node in this implementation further includes: a later module 55, a discarding module 56.

When the first receiving module 53 does not receive the migration control signal, the module 55 is later used to send a later signal to the service data request; the discarding module 56 is used to discard the unimportant or unreliable traffic data currently being processed.

The service data management apparatus of the distributed system provided by this embodiment facilitates the implementation of the service data management method of the distributed system. In the distributed system, it is possible to dynamically balance the load of each service node and provide the beneficial effect of the self-regulation capability of the entire distributed system. Example 6

This embodiment specifically provides a distributed system. As shown in FIG. 6, the system includes: a service node entity 61 and a central management entity 62.

The service node entity 61 is configured to periodically obtain current load information of the service node and send the load information to the central management node. The central management entity 62 is configured to receive current load information of the service node periodically sent by the service node, and search for the overloaded service node and the light load service node according to the load information, and determine whether part of the searched overloaded service node can be migrated. Data, when the service data in the overloaded service node can be migrated, sending a migration control signal to the overloaded service node, the migration control signal controlling the overloaded service node to migrate part of the service data to the found light load service On the node.

And the service node entity 61 is further configured to: when receiving the migration control signal, cut part of the service data in the service node to the light load service node according to the migration control signal.

The distributed system of the solution has the following beneficial effects: dynamically balancing the load between the various service nodes, effectively utilizing the resources of each service node in the distributed system, and timely reducing the burden on the heavy-duty service nodes.

The central management entity 62 in this embodiment is further configured to modify an address of the service node where the partial service data is located to an address of the light load service node, and when the service in the overloaded service node cannot be migrated When the data is not transmitted, the migration control signal is not sent to the overloaded service node. The service node entity 61 is further configured to send a later signal to the service data request and discard the unimportant or unreliable service data currently being processed when the migration control signal sent by the central management node is not received.

The embodiments of the present invention are mainly applied to a distributed system to implement load balancing of service nodes in a distributed system. It is of course also possible to use embodiments of the invention in other systems or fields that contain distributed technology and, with the development of network technology, it is possible to apply to other scenarios in the field or to similar or similar technical fields.

Through the description of the above embodiments, those skilled in the art can clearly understand that the present invention can be implemented by means of software plus a necessary general hardware platform, and of course, through hardware. But in many cases the former is a better implementation. Based on such understanding, the technical solution of the present invention, which is essential or contributes to the prior art, may be embodied in the form of a software product stored in a readable storage medium, such as a floppy disk of a computer. , a hard disk or an optical disk, etc., including instructions for causing a device (which may be a router, or a computer, etc.) to perform the methods described in various embodiments of the present invention.

The above is only the specific embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily think of changes or substitutions within the technical scope of the present invention. It should be covered by the scope of the present invention. Therefore, the scope of the invention should be determined by the scope of the claims.

Claims

Claim

A method for managing a service data of a distributed system, comprising:

And if it is determined that part of the service data in the overloaded service node can be migrated, sending, to the overloaded service node, a migration control signal, where the migration control signal controls the overloaded service node to migrate part of the service data to the found light load service. On the node.

The service data management method for a distributed system according to claim 1, wherein, if the service data in the overloaded service node can be migrated, the migration control is sent to the overloaded service node. After the signal, the method further includes:

Modifying the address of the service node where the part of the service data is located to the address of the light load service node.

The service data management method for a distributed system according to claim 1, wherein the load information includes a total load of the service node;

The searching for the overloaded service node according to the load information is: using a service node whose total load exceeds a preset first reload threshold as a heavy load service node;

The searching for the light load service node according to the load information is: using a service node whose total load does not exceed a preset first light load threshold as a light load service node.

The service data management method of the distributed system according to claim 1, wherein the load information includes a total load of the service node and a partial load corresponding to each service data type;

The searching for the overloaded service node according to the load information is: using a service node whose load exceeds a preset second reload threshold as a heavy load service node;

The searching for the light load service node according to the load information is as follows: the service node whose load is not more than the preset second light load threshold and whose total load does not exceed the preset first light load threshold is used as the light load service node.

5. The service data management method of a distributed system according to claim 3, wherein: Determining whether part of the service data in the overloaded service node can be migrated includes:

Determining part of the service data that needs to be migrated in the discovered overloaded service node;

Determine whether the light load service node is found;

If no light load service node is found, determining that part of the service data in the overloaded service node cannot be migrated;

If the light load service node is found, it is determined whether the total load plus the load corresponding to the part of the service data exceeds the preset first reload threshold;

If the total load plus the load corresponding to the part of the service data does not exceed the preset first reload threshold, determining that part of the service data in the overloaded service node can be migrated;

If the total load plus the load corresponding to the part of the service data exceeds the preset first reload threshold, it is determined that part of the service data in the overloaded service node cannot be migrated.

The service data management method of the distributed system according to claim 4, wherein determining whether the part of the service data in the overloaded service node can be migrated comprises:

Determine whether the light load service node is found;

If the light load service node is found, it is determined whether the load corresponding to the part of the service data exceeds the preset second reload threshold, and the total load plus the load corresponding to the part of the service data exceeds The preset first reload threshold;

If the load corresponding to the part of the service data does not exceed the preset second reload threshold, the total load plus the load corresponding to the part of the service data does not exceed the preset first weight. Carrying a threshold, determining that part of the service data in the overloaded service node can be migrated;

Otherwise, it is determined that part of the service data in the overloaded service node cannot be migrated.

A method for managing a business data of a distributed system, comprising:

Regularly obtain the current load information of the service node; Sending the load information to a central management node;

Receiving, by the central management node, a migration control signal, where the migration control signal includes at least an address information of the light-duty service node and an identifier of a part of the service data that needs to be migrated;

Cutting part of the service data corresponding to the identifier in the service node to a light service node corresponding to the address information.

The service data management method of the distributed system according to claim 7, wherein when the migration control signal sent by the central management node is not received, the method further includes:

Sending a later signal to the service data request;

Discard unimportant or unreliable business data that is currently being processed.

9. A service data management apparatus for a distributed system, comprising:

a migration control module, configured to: when the determining module determines that part of the service data in the overloaded service node can be migrated, send a migration control signal to the overloaded service node, where the migration control signal controls the overloaded service node to part of the service data Migrate to the light load service node found by the light load lookup module.

The service data management device of the distributed system according to claim 9, wherein the device further comprises:

And a modification module, configured to modify an address of the service node where the service data is located to an address of the light load service node.

The service data management apparatus of the distributed system according to claim 9, wherein the load information received by the receiving module includes a total load of the service node,

The reloading search module uses the service node whose total load exceeds the preset first reload threshold as the overloaded service node; The light load search module uses the service node whose total load does not exceed the preset first light load threshold as the light load service node.

The service data management apparatus of the distributed system according to claim 9, wherein the load information received by the receiving module includes a total load of the service node and a load corresponding to each service data type;

The reloading search module uses the service node whose load exceeds the preset second reload threshold as the heavy load service node;

The light load search module uses the service node whose load does not exceed the preset second light load threshold and whose total load does not exceed the preset first light load threshold as the light load service node.

The service data management device of the distributed system according to claim 11, wherein the determining module comprises:

a first determining unit, configured to determine part of the service data that needs to be migrated in the overloaded service node found by the overloaded search module;

a first determining unit, configured to determine whether the light load search module finds a light load service node, and when the light load search module does not find the light load service node, determine that part of the service data in the overloaded service node cannot be migrated;

a second determining unit, configured to: when the light load searching module finds the light load service node, determine whether the total load plus the load corresponding to the part of the service data exceeds the preset first reload threshold; If the load corresponding to the part of the service data does not exceed the preset first reload threshold, it is determined that part of the service data in the overloaded service node can be migrated; if the total load is added to the part of the service data If the load exceeds the preset first reload threshold, it is determined that part of the service data in the overloaded service node cannot be migrated.

The service data management device of the distributed system according to claim 12, wherein the determining module comprises:

a second determining unit, configured to determine part of the service data that needs to be migrated in the overloaded service node found by the overloaded search module; a third determining unit, configured to determine whether the light load search module finds the light load service node; when the light load search module does not find the light load service node, determine that part of the service data in the overloaded service node cannot be migrated;

a fourth determining unit, configured to: when the light load searching module finds the light load service node, determine whether the load corresponding to the partial service data exceeds the preset second overload threshold, and the total load is added Whether the load corresponding to the part of the service data exceeds the preset first reload threshold; if the load plus the load corresponding to the part of the service data does not exceed the preset second reload threshold, If the load corresponding to the part of the service data does not exceed the preset first reload threshold, it is determined that part of the service data in the overloaded service node can be migrated; otherwise, it is determined that the load cannot be migrated. Describe some of the service data in the service node.

15. A service data management apparatus for a distributed system, comprising:

a sending module, configured to send the load information acquired by the management module to the central management node; the first receiving module is configured to receive a migration control signal sent by the central management node, where the migration control signal includes at least an address of the light-load service node The information and the identifier of the part of the service data that needs to be migrated; the cutting module is configured to cut part of the service data corresponding to the identifier in the service node to the light load service node corresponding to the address information.

The service data management device of the distributed system according to claim 15, wherein the device further comprises:

a module later, for transmitting a later signal to the service data request;

A discarding module is used to discard unimportant or unreliable business data that is currently being processed.

17. A distributed system, comprising:

a service node entity, configured to periodically acquire current load information of the service node and send the load information to the central management node;

a central management entity, configured to receive current load information of the service node periodically sent by the service node; search for the overloaded service node and the light load service node according to the load information, and determine whether the migration can be found Part of the service data in the overloaded service node, when it is determined that part of the service data in the overloaded service node can be migrated, sending a migration control signal to the overloaded service node, the migration control signal controlling the overloaded service node to Part of the business data is migrated to the discovered light load service node;

The service node entity is further configured to, when receiving the migration control signal, cut part of the service data in the service node to the light load service node according to the migration control signal.

The distributed system according to claim 17, comprising:

The central management entity is further configured to modify an address of the service data to an address of the light load service node, and not to the heavy load service when the service data in the overloaded service node cannot be migrated The node sends a migration control signal.

The distributed system according to claim 17, comprising:

The service node entity is further configured to: when the migration control signal sent by the central management node is not received, send a later signal to the service data request and discard the unimportant or unreliable service data currently being processed.