CN103918239A - Load balancing method, device, system and computer readable medium - Google Patents
Load balancing method, device, system and computer readable medium Download PDFInfo
- Publication number
- CN103918239A CN103918239A CN201280001746.3A CN201280001746A CN103918239A CN 103918239 A CN103918239 A CN 103918239A CN 201280001746 A CN201280001746 A CN 201280001746A CN 103918239 A CN103918239 A CN 103918239A
- Authority
- CN
- China
- Prior art keywords
- node
- load
- processing
- distributed
- dps
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
Abstract
Embodiments of the present invention provide a load balancing method, device, system and computer readable medium. The load balancing method comprises: obtaining a first load of a first processing node in a distributed processing system and an average load of all processing nodes in the distributed processing system; and if the first load is greater than the average load, migrating at least one virtual node distributed on the first processing node to at least one processing node whose load is less than the average load in the distributed processing system. By means of the embodiment of the present invention, at least one virtual node corresponding to a processing node whose load is greater than an average load in a distributed system can be migrated to at least one processing node whose load is less than the average load in the distributed processing system, so that the load distribution on processing nodes in the distributed processing system becomes more balanced, thereby enhancing the parallel processing capability of the distributed processing system and improving the efficiency of the distributed processing system.
Description
The present embodiments relate to the communication technology, more particularly to a kind of load-balancing method, device, system and computer-readable medium for load-balancing method, device, system and computer-readable medium technical field.Background technology
With the high speed development of internet, requirement more and more higher of the enterprise to distributed treatment ability.Distributed processing system(DPS) is by a large amount of cheap server groups into cluster, and each server is a processing node, is handled by handling nodal parallel, to provide an entirety high performance cluster disposal ability.
At present; in distributed processing system(DPS); one processing task would generally be broken down into multiple processing and be requested and allocated to multiple processing nodal parallel processing, it is hereby achieved that higher parallel processing capability, this is accomplished by multiple processing requests balancedly distributing to each processing node.
During the embodiment of the present invention is realized, inventor has found that the unbalanced problem of processing request distribution occurs often in processing method of the prior art, and then cause the processing node of load weight to turn into the bottleneck of whole distributed processing system(DPS), and load light processing node and waste its disposal ability so that the treatment effeciency of distributed processing system(DPS) declines.The content of the invention embodiment of the present invention provides a kind of load-balancing method, device, system and computer-readable medium, with realize the processing node in distributed processing system(DPS) in the distribution of load it is more balanced.
First aspect there is provided a kind of load balance process method, including：
Obtain the average loads of all processing nodes in the in distributed processing system(DPS) first the first load for handling node and the distributed processing system(DPS)；
If first load is more than the average load, it will be distributed at least one dummy node on the first processing node and move in the distributed processing system(DPS) load less than at least one processing node of the average load.
In the first possible implementation of first aspect, described at least one dummy node that will be distributed on the first processing node moves to load in the distributed processing system(DPS) and is less than institute
State at least one processing node of average load, including：
Load total amount on the first processing node is will be distributed over to be less than or equal at least one dummy node of overload quantity and move to underload total amount in the distributed processing system(DPS) to be more than or equal at least one processing node of the load total amount, the overload quantity is the load capacity that the described first load exceedes the average load, and the underload total amount is less than the load capacity sum of the average load for the load of each processing node at least one described processing node.
With reference to the first possible embodiment of first aspect or the first aspect, in second of possible embodiment, before described at least one dummy node that will be distributed on the first processing node is moved at least one the processing node for being loaded in the distributed processing system(DPS) and be less than the average load, in addition to：
The load information of each processing node in the distributed processing system(DPS) is obtained, the load information includes the load information that each dummy node on node is throughout managed in the distribution；
According to the load information of each processing node, it is determined that at least one described processing node；Migration request message is sent at least one described processing node, the migration request message carries the information of at least one dummy node；
Receive the migration request response message of at least one processing node；
Described at least one dummy node that will be distributed on the first processing node moves to load in the distributed processing system(DPS)：
If the migration request response message is success response message, it will be distributed at least one dummy node on the first processing node and move at least one described processing node.
With reference to the first possible embodiment of first aspect or the first aspect, in the third possible embodiment, described at least one dummy node that will be distributed on the first processing node moves to load in the distributed processing system(DPS) and is less than at least one processing node of the average load, including：
Update the mapping table between the first processing node and at least one described dummy node；
Other processing nodes into the distributed processing system(DPS) send fresh information, so that other processing nodes in the distributed processing system(DPS) update the mapping table between the processing node being locally stored and dummy node according to the fresh information.
With reference to the first possible embodiment of first aspect or the first aspect, in the 4th kind of possible embodiment, after described at least one dummy node that will be distributed on the first processing node is moved at least one the processing node for being loaded in the distributed processing system(DPS) and be less than the average load, in addition to：
By on the corresponding data syn-chronization of at least one dummy node at least one described processing node.
Second aspect there is provided a kind of load balance process device, including：
Acquisition module, the average loads for obtaining all processing nodes in the in distributed processing system(DPS) first the first load for handling node and the distributed processing system(DPS)；
Processing module, if first load is more than the average load, will be distributed at least one dummy node on the first processing node and moves in the distributed processing system(DPS) load less than at least one processing node of the average load.
In the first possible implementation of second aspect, the processing module, it is less than or equal at least one dummy node of overload quantity and moves to underload total amount in the distributed processing system(DPS) to be more than or equal at least one processing node of the load total amount specifically for will be distributed over load total amount on the first processing node, the overload quantity is the load capacity that the described first load exceedes the average load, and the underload total amount is less than the load capacity sum of the average load for the load of each processing node at least one described processing node.
With reference to the first possible embodiment of second aspect or the second aspect, in second of possible embodiment, in addition to determining module, sending module and receiving module；
The acquisition module, is additionally operable to obtain the load information of each processing node in the distributed processing system(DPS), and the load information includes the load information for each dummy node being distributed on each processing node；
The determining module, for the load information according to each processing node, it is determined that at least one described processing node；
The sending module, for sending migration request message at least one described processing node, the migration request message carries the information of at least one dummy node；
The receiving module, for receiving the migration request response message that at least one described processing node feeds back according to the migration request message；
If the processing module is success response message specifically for the migration request response message,
At least one dummy node on the first processing node is then will be distributed over to move at least one described processing node.
With reference to the first possible embodiment of second aspect or the second aspect, in the third possible embodiment, in addition to：
Update module, for updating the mapping table between the first processing node and at least one described dummy node；
The sending module, other processing nodes being additionally operable into the distributed processing system(DPS) send fresh information, so that other processing nodes in the distributed processing system(DPS) update the mapping table between the processing node being locally stored and dummy node according to the fresh information.
With reference to the first possible embodiment of second aspect or the second aspect, in the 4th kind of possible embodiment, in addition to：
Synchronization module, for inciting somebody to action the corresponding data syn-chronization of at least one dummy node at least one described processing node.
With reference to the first possible embodiment of second aspect or the second aspect, in the 5th kind of possible embodiment, described device is processing node or Master Control Center.
The third aspect there is provided a kind of distributed processing system(DPS), including：Processing node described at least two above-mentioned 5th kind of possible implementations.
Fourth aspect, also provides a kind of distributed processing system(DPS), including：Master Control Center described at least two processing nodes and above-mentioned 5th kind of possible implementation.
5th aspect, node is handled there is provided one kind, including processor and memory, the memory storage execute instruction, when described device is run, communicated between the processor and the memory, execute instruction described in the computing device causes the processing node to perform the method described in first aspect.
6th aspect, there is provided a kind of Master Control Center, including processor and memory, the memory storage execute instruction, when described device is run, communicated between the processor and the memory, execute instruction described in the computing device causes the Master Control Center to perform the method described in first aspect.
7th aspect, additionally provides a kind of distributed processing system(DPS), including the processing node described at least two the 5th aspects.
Eighth aspect, provides a kind of distributed processing system(DPS), including at least two processing nodes again
And the Master Control Center in terms of the 6th.
9th aspect is there is provided a kind of computer-readable medium, comprising computer executed instructions, and the computer executed instructions are used to make processing node perform the method described in first aspect.
Tenth aspect, additionally provides a kind of computer-readable medium, comprising computer executed instructions, and the computer executed instructions are used to make Master Control Center perform the method described in first aspect.
The load balance process method that the present embodiment is provided, device, system and computer-readable medium, it is less than by that will load at least one dummy node being more than corresponding to the processing node of average load in distributed system and move to load in distributed processing system(DPS) at least one processing node of average load, the processing node in distributed processing system(DPS) can be made more balanced in the distribution of load, and then strengthen the parallel processing capability of distributed processing system(DPS), improve the efficiency of distributed processing system(DPS).Brief description of the drawings
The accompanying drawing used required in embodiment or description of the prior art is briefly described, apparently, drawings in the following description are some embodiments of the present invention, for those of ordinary skill in the art, without having to pay creative labor, other accompanying drawings can also be obtained according to these accompanying drawings.
The flow chart for the load balance process embodiment of the method one that Fig. 1 provides for the present invention；
The flow chart for the load balance process embodiment of the method two that Fig. 2 provides for the present invention；
The flow chart for the load balance process embodiment of the method three that Fig. 3 provides for the present invention；
The structural representation for the load balance process device embodiment one that Fig. 4 provides for the present invention；The structural representation for the load balance process device embodiment two that Fig. 5 provides for the present invention；The structural representation for the load balance process device embodiment three that Fig. 6 provides for the present invention；The structural representation for the load balance process device embodiment four that Fig. 7 provides for the present invention；The structural representation for the processing node embodiment one that Fig. 8 provides for the present invention；
The structural representation for the Master Control Center embodiment one that Fig. 9 provides for the present invention；
The structural representation for the distributed processing system(DPS) embodiment one that Figure 10 provides for the present invention；The structural representation for the distributed processing system(DPS) embodiment two that Fig. 11 provides for the present invention.Embodiment
To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with the accompanying drawing in the embodiment of the present invention, technical scheme in the embodiment of the present invention is clearly and completely described, obviously, described embodiment is a part of embodiment of the invention, rather than whole embodiments.Based on the embodiment in the present invention, the every other embodiment that those of ordinary skill in the art are obtained under the premise of creative work is not made belongs to the scope of protection of the invention.
The flow chart for the load balance process embodiment of the method one that Fig. 1 provides for the present invention, as shown in figure 1, load balance process method includes in the present embodiment：
The average load of all processing nodes in step 101, the first load of the first processing node obtained in distributed processing system(DPS) and distributed processing system(DPS)；
For example, the mode for obtaining the first load of the in distributed processing system(DPS) first processing node can set a load counter on the first processing node, first processing node receives a processing request, then Jia 1 to the progress of above-mentioned load counter to operate, the processing request amount of above-mentioned load counter record is the above-mentioned first load for handling node；It is another it may is that：Load counter is set on each corresponding dummy node of the above-mentioned first processing node, then the load sum of each dummy node is the load of the above-mentioned first processing node.
The acquisition modes of the average load of all processing nodes of distributed processing system(DPS) can be：Each processing node in distributed processing system(DPS) elects a main process task node by consistency algorithms such as pasox, one software timer is set on main process task node, its load information is sent to other processing nodes in distributed processing system(DPS) by the timing of main process task node, other above-mentioned processing nodes are received after the load information of main process task node, the load information of itself is sent to main process task node, main process task node calculates the average load of all processing nodes of whole distributed processing system(DPS) according to the load information of all processing nodes in distributed processing system(DPS), again to other processing above-mentioned average loads of node broadcasts.
The acquisition modes of the average load of all processing nodes of distributed processing system(DPS) can also be：Master Control Center in distributed processing system(DPS) calculates the average load for all processing nodes for obtaining distributed processing system(DPS), then handle the above-mentioned average load of node broadcasts to other by obtaining all load informations for handling nodes in distributed processing system(DPS).
In distributed processing system(DPS), a kind of distributed hash for being called dummy node is employed.Assuming that there is N processing node, the spatial loop of keyword is randomly divided into by Q interval (Q N) by algorithm, an interval is known as a dummy node.Then this Q dummy node is divided at random
It is fitted on N number of processing node, storage is about Q/N dummy node on each processing node.Wherein, N, Q are positive integer here.Distributed processing system(DPS) is also suitable the distributed processing system(DPS) described in following specific embodiment in the embodiment of the present invention.
If step 102, the first load are more than average load, it will be distributed at least one dummy node on the first processing node and move in distributed processing system(DPS) load less than at least one processing node of the average load.
For example, if the first load of the first processing node is more than average load, first load is referred to as overload quantity more than the load capacity of average load, one overload quantity threshold value can be according to circumstances set, that is, above-mentioned transition process is just carried out in the case where the overload quantity of the first processing node exceedes threshold value.After above-mentioned dummy node migration, the part processing request corresponding to the above-mentioned dummy node being migrated can then be handled less than at least one processing node of average load by receiving the load of the above-mentioned dummy node being migrated, and then reduce the load of above-mentioned first processing node.
The load balance process method that the present embodiment is provided, it is less than by will be distributed over load in distributed system and move to load in distributed processing system(DPS) more than at least one dummy node on the processing node of average load at least one processing node of average load, the processing node in distributed processing system(DPS) can be made more balanced in the distribution of load, and then strengthen the parallel processing capability of distributed processing system(DPS), improve the efficiency of distributed processing system(DPS).
Alternatively, on the basis of Fig. 1 correspondence embodiments, above-mentioned at least one dummy node that will be distributed on the first processing node moves to load in distributed processing system(DPS) and is less than at least one processing node of average load, including：
It will be distributed over and total amount is loaded on the first processing node be less than or equal at least one dummy node of overload quantity and move to underload total amount in distributed processing system(DPS) to be more than or equal at least one processing node of load total amount, overload quantity is load capacity of first load more than average load, and underload total amount handles load capacity sum of the load less than average load of each processing node in node at least one.
Specifically, if at least one above-mentioned dummy node is a dummy node, then above-mentioned load total amount is the load capacity of a dummy node, if at least one above-mentioned dummy node is multiple dummy nodes, then above-mentioned load total amount is exactly the load sum of multiple dummy nodes, same, if at least one above-mentioned processing node is a processing node, then above-mentioned underload total amount is load capacity of the load less than average load of this processing node, if at least one above-mentioned processing node is multiple processing nodes, then above-mentioned underload total amount is less than for the load of each processing node at least one processing node
The load capacity sum of average load.
For example, if the overload quantity that node A is handled in distributed processing system(DPS) is 50, the load for a dummy node A1 being distributed on processing node A is 50, the processing node of another in distributed processing system(DPS) B underload amount is 50, then dummy node A1 is moved on processing node B, it is another it may is that：If the overload quantity that node A is handled in distributed processing system(DPS) is 50, two dummy nodes A1 and A2 being distributed on processing node A load are respectively 20 and 30, the processing node of another in distributed processing system(DPS) B underload amount is 60, then moves to dummy node A1 and A2 on processing node B.
If the overload quantity that node A is handled in distributed processing system(DPS) is 50, two dummy nodes A1 and A2 being distributed on processing node A load are respectively 20 and 30, the processing node of two other in distributed processing system(DPS) B and C underload amount are respectively 20 and 30, then dummy node A1 is moved on processing node B, dummy node A2 is moved on processing node C, it is another it may is that：If the overload quantity that node A is handled in distributed processing system(DPS) is 50, the load for three dummy node Al and A2 and A3 being distributed on processing node A is respectively 15,20 and 10, the processing node of another in distributed processing system(DPS) B and C underload amount are respectively 25 and 45, then dummy node A1 is moved on processing node B, dummy node A2 and A3 are moved on processing node C.The load balance process method that the present embodiment is provided, it is less than or equal at least one dummy node of overload quantity and moves to underload total amount in distributed processing system(DPS) to be more than or equal at least one processing node of load total amount by the way that load total amount in node will be handled, the processing node in distributed processing system(DPS) can be made more balanced in the distribution of load, strengthen the parallel processing capability of distributed processing system(DPS), improve the efficiency of distributed processing system(DPS).
The flow chart for the load balance process embodiment of the method two that Fig. 2 provides for the present invention, as shown in Fig. 2 load balance process method includes in the present embodiment：
Step 201, the load information for obtaining each processing node in distributed processing system(DPS), load information include being distributed the load information of each dummy node on throughout reason node.
For example, obtaining the method for the load information of each processing node in distributed processing system(DPS) can be：Each processing node in distributed processing system(DPS) elects a main process task node by consistency algorithms such as pasox, one software timer is set on main process task node, its load information is sent to other processing nodes in distributed processing system(DPS) by the timing of main process task node, other above-mentioned processing nodes are received after the load information of main process task node, and the load information of itself is sent into main process task section
Point, the load information of main process task node other all processing nodes of processing node broadcasts into distributed processing system(DPS), or its load information is sent to other processing nodes in distributed processing system(DPS) by the timing of main process task node, other above-mentioned processing nodes are received after the load information of main process task node, and the load information of itself is sent to other processing nodes in distributed processing system(DPS).
The method for obtaining the load information of each processing node in distributed processing system(DPS) can also be：Master Control Center in distributed processing system(DPS) is by obtaining all load informations for handling nodes in distributed processing system(DPS), then other processing above-mentioned load informations of node broadcasts into distributed processing system(DPS).
The load information of above-mentioned processing node includes the dummy node being distributed on processing node and the load capacity on each dummy node.
Step 202, the load information according to each processing node, determine at least one processing node.Because each processing node can know the load information of other processing nodes in distributed processing system(DPS), if so average load of first load more than distributed processing system(DPS) of the first processing node, the processing node or Master Control Center then determine at least one the processing node of load less than the average load in distributed processing system(DPS) according to the load of the overload quantity of the above-mentioned first processing node and its dummy node being distributed on the first processing node.
Step 203, at least one processing node migration request message is sent, migration request message carries the information of at least one dummy node.
Determine after at least one above-mentioned processing node, the first processing node or Master Control Center send migration request message at least one processing node of determination, and migration request message carries virtual node information to be migrated.
Step 204, the migration request response message for receiving at least one the processing node determined.
At least one processing node of above-mentioned determination is received after above-mentioned migration request message, to above-mentioned first processing one migration request response message of node feeding back.
If step 205, migration request response message are success response message, it will be distributed at least one dummy node on the first processing node and move in distributed processing system(DPS) load less than at least one processing node of the average load.
For example, migration request response message disappears for failure response, it may is that above-mentioned determination at least one processing node underload total amount be less than above-mentioned at least one dummy node to be migrated load total amount.
The load balance process method that the present embodiment is provided, believes according to the load of each processing node first
Breath, determine at least one processing node, to this, at least one processing node sends migration request message again, if migration request response message is success response message, it then will be distributed at least one dummy node on the first processing node and move to load in distributed processing system(DPS) and be less than at least one processing node of the average load, the processing node in distributed processing system(DPS) can be made more balanced in the distribution of load, strengthen the parallel processing capability of distributed processing system(DPS), improve the efficiency of distributed processing system(DPS).
On the basis of the corresponding embodiments of Fig. 1 and Fig. 2, further, it will be distributed at least one dummy node on the first processing node and move in distributed processing system(DPS) load less than at least one processing node of average load, including：
Update the mapping table between the first processing node and at least one dummy node.
Other processing nodes into distributed processing system(DPS) send fresh information, so that other processing nodes in distributed processing system(DPS) update the mapping table between the processing node being locally stored and dummy node according to fresh information.
Specifically, above-mentioned at least one dummy node that will be distributed on the first processing node moves in distributed processing system(DPS) load and is less than a kind of implementation of the operation at least one processing node of average load：Above-mentioned first processing node or Master Control Center handle the mapping table between node and at least one above-mentioned dummy node according to the information updating first of at least one above-mentioned dummy node, and mapping table for example can be as shown in table 1：
Table 1
For example, if the overload quantity for handling node A is 50, the load for a dummy node A1 being distributed on processing node A is 50, the processing node of another in distributed processing system(DPS) B underload amount is 50, then dummy node A1 is moved on processing node B, then the mapping table after updating is as shown in table 2：
Table 2
Handle node A dummy node A2 (scopes）, handle node B dummy node A1 (scopes）, dummy node B1 (scopes）, dummy node
B2 (scopes）...
Handle node C dummy node C1 (scopes）, dummy node C2 (scopes）... above-mentioned first processing node or Master Control Center are handled according to the information updating first of at least one above-mentioned dummy node after the mapping table between node and at least one above-mentioned dummy node, other processing nodes that can be into distributed processing system(DPS) send fresh information, other processing nodes that can include in the fresh information in the mapping table after updating, distributed processing system(DPS) update the mapping table handled between node and dummy node being locally stored according to the fresh information.
The load balance process method that the present embodiment is provided, by updating the first mapping table for handling node and at least one dummy node, and other processing nodes into distributed processing system(DPS) send fresh information, realize that at least one dummy node on the processing node that will be loaded in distributed system more than average load moves to load in distributed processing system(DPS) and is less than at least one of average load in the way of making other processing nodes in distributed processing system(DPS) update the mapping table between the processing node and the dummy node that are locally stored according to fresh information to handle on node, realize that the processing node in distributed processing system(DPS) is more balanced in the distribution of load, strengthen the parallel processing capability of distributed processing system(DPS), improve the efficiency of distributed processing system(DPS).
On the basis of the corresponding embodiments of Fig. 1, further, it will be distributed at least one dummy node on the first processing node and move in distributed processing system(DPS) load less than after at least one processing node of average load, in addition to：
By on the corresponding data syn-chronization of at least one dummy node at least one processing node.
Specifically, if the overload quantity for handling node A is 50, the load for a dummy node A1 being distributed on processing node A is 50, the processing node of another in distributed processing system(DPS) B underload amount is 50, after dummy node A1 is moved on processing node B, processing node A corresponding data syn-chronizations of A1 be will be distributed over to processing node B, if the processing request that dummy node A1 is received is read operation, then by after the corresponding data syn-chronizations of A1 to processing node B, processing node B can also handle above-mentioned read operation processing request；If the processing request that dummy node A1 is received is write operation, then processing node B can also handle write operation progress, it is stored with hot resource for the dummy node A1 being distributed on processing node A, and cause the situation of processing node A load excessives, it then will be distributed over the dummy node A1 on processing node A to move on processing node B, and handle by processing node B the request.
The load balance process method that the present embodiment is provided, by the dummy node, corresponding data syn-chronization is handled on node at least one on the first processing node after by being migrated in dummy node, in the case where some processing nodes are stored with hot resource, still the processing node in distributed processing system(DPS) can be made more balanced in the distribution of load, strengthen the parallel processing capability of distributed processing system(DPS), improve the efficiency of distributed processing system(DPS).
The flow chart for the load balance process embodiment of the method three that Fig. 3 provides for the present invention, as shown in figure 3, load balance process method includes in the present embodiment：
Step 301, timer are then.
Above-mentioned timer can be arranged on main process task node or Master Control Center, and main process task node is elected by each processing node in distributed processing system(DPS) by consistency algorithms such as pasox, specifically described see embodiment one.
Its load information is sent the processing node of other in distributed processing system(DPS) by step 302, main process task node.
The load information of itself is sent to main process task node by the processing node of other in step 303, distributed processing system(DPS).
The load information of step 304, main process task node other all processing nodes of processing node broadcasts into distributed processing system(DPS).
Step 305, each processing node calculate the average load of distributed processing system(DPS) according to the load information received, and own load and average load are compared, and judge whether to need to migrate load.
In the present embodiment, processing node A load is more than average load, and processing node A load is referred to as the overload quantity for handling node A more than the load capacity of average load, pre-sets threshold value, when overload quantity is more than above-mentioned threshold value, just carries out load migration.
Step 306, processing node A determine that at least one dummy node and underload total amount are more than or equal at least one processing node that dummy node to be migrated loads total amount according to the load information of each processing node.
Specifically, if at least one above-mentioned dummy node is a dummy node, then above-mentioned load total amount is the load capacity of a dummy node, if at least one above-mentioned dummy node is multiple dummy nodes, then above-mentioned load total amount is exactly the load sum of multiple dummy nodes, likewise, if at least one above-mentioned processing node is a processing node, above-mentioned underload total amount is this processing node
Load less than average load load capacity, if it is above-mentioned at least one processing node be multiple processing nodes, above-mentioned underload total amount at least one processing node in each processing node load less than average load load capacity sum.
In the present embodiment, processing node B load is less than average load, and its underload amount is more than or equal to the load total amount of dummy node to be migrated, and the load capacity for the dummy node A1 being distributed on processing node A is less than processing node A overload quantity.
Step 307, processing node A send migration request message to processing node B.
Above-mentioned migration message carries dummy node A1 load information.
Step 308, processing node B send migration request response message to processing node A.
If the underload amount for handling node B is more than or equal to dummy node A1 load capacity, processing node B sends migration request response message to processing node A.
Step 309, processing node A update dummy node A1 and processing node A mapping table.
A1-A before migration is updated to the Al-B after migration by processing node A.
Step 310, processing node B update dummy node A1 and processing node B mapping table.
Other processing nodes of step 311, processing node A into the distributed processing system(DPS) send fresh information.
Other processing nodes in step 312, distributed processing system(DPS) update the mapping table between the processing node being locally stored and dummy node according to fresh information.
The load balance process method that the present embodiment is provided, update first handle node and the mapping table of dummy node and the mapping table of at least one processing node of the determination and the dummy node being distributed at least one processing node of the determination that are distributed on the first processing node by way of realize that at least one dummy node that will be distributed in distributed system on processing node of the load more than average load moves in distributed processing system(DPS) load and handled less than at least one of average load on node, the processing node in distributed processing system(DPS) can be made more balanced in the distribution of load, strengthen the parallel processing capability of distributed processing system(DPS), improve the efficiency of distributed processing system(DPS).
The structural representation for the load balance process device embodiment one that Fig. 4 provides for the present invention, as shown in Fig. 4, the load balance process device that the present embodiment is provided includes acquisition module 41 and processing module
42, wherein, the first load and the average load of distributed processing system(DPS) for the first processing node that acquisition module 41 is used to obtain in distributed processing system(DPS), if processing module 42 is used for the first load and is more than average load, at least one corresponding dummy node of the first processing node is moved to and loaded in distributed processing system(DPS) at least one processing node less than average load.
The device of the present embodiment, can be used for the technical scheme for performing embodiment of the method shown in Fig. 1, and its implementing principle and technical effect is similar, and here is omitted.
Alternatively, processing module 42 loads total amount and is less than or equal at least one dummy node of overload quantity and moves to underload total amount in distributed processing system(DPS) to be more than or equal at least one processing node of load total amount specifically for will be distributed on the first processing node, overload quantity is load capacity of first load more than average load, and underload total amount handles load capacity sum of the load less than average load of each processing node in node at least one.
Specifically, if at least one above-mentioned dummy node is a dummy node, then above-mentioned load total amount is the load capacity of a dummy node, if at least one above-mentioned dummy node is multiple dummy nodes, then above-mentioned load total amount is exactly the load sum of multiple dummy nodes, same, if at least one above-mentioned processing node is a processing node, then above-mentioned underload total amount is load capacity of the load less than average load of this processing node, if at least one above-mentioned processing node is multiple processing nodes, then above-mentioned underload total amount is less than the load capacity sum of average load for the load of each processing node at least one processing node.
The structural representation for the load balance process device embodiment two that Fig. 5 provides for the present invention, as shown in Fig. 5, the load balance process device that the present embodiment is provided, on the basis of Fig. 4 illustrated embodiments, further, also include determining module 43, sending module 44, receiving module 45, wherein, acquisition module 41 is additionally operable to obtain the load information of each processing node in the distributed processing system(DPS), the load information includes the load information for each dummy node being distributed on each processing node, determining module 43 is used for the load information according to each processing node, determine at least one processing node, sending module 44 is used to send migration request message at least one described processing node, the migration request message carries the information of at least one dummy node, receiving module 45 is used to receive the migration request response message that at least one processing node feeds back according to migration request message, if processing module 42 is success response message specifically for the migration request response message, at least one dummy node on the first processing node is then will be distributed over to move at least one described processing node.
The device of the present embodiment, can be used for the technical scheme for performing embodiment of the method shown in Fig. 2, its
Implementing principle and technical effect are similar, and here is omitted.
The structural representation for the load balance process device embodiment three that Fig. 6 provides for the present invention, as shown in Fig. 6, the load balance process device that the present embodiment is provided, on the basis of Fig. 5 illustrated embodiments, further, in addition to：The update module 46 of update module 46 is used to update the mapping table between the first processing node and at least one described dummy node, other processing nodes that above-mentioned sending module 44 is additionally operable into the distributed processing system(DPS) send fresh information, so that other processing nodes in the distributed processing system(DPS) update the mapping table between the processing node being locally stored and dummy node according to the fresh information.
The device of the present embodiment, can be used for the technical scheme for performing embodiment of the method shown in Fig. 3, and its implementing principle and technical effect is similar, and here is omitted.
The structural representation for the load balance process device embodiment four that Fig. 7 provides for the present invention, as shown in Fig. 7, the load balance process device that the present embodiment is provided, on the basis of Fig. 6 illustrated embodiments, further, also include synchronization module 47, synchronization unit 47 is used for the corresponding data syn-chronization of at least one dummy node at least one described processing node.
Alternatively, load balance process device provided in an embodiment of the present invention can be processing node or Master Control Center.
Specifically, load balance process device can include for the situation of processing node：Any processing node can carry out load balance process to distributed processing system(DPS) or carry out load balance process to distributed processing system(DPS) by a main process task node in distributed processing system(DPS) in distributed processing system(DPS).
For the situation that load balance process device provided in an embodiment of the present invention is Master Control Center, a Master Control Center can be set to be used for carrying out load balance process to distributed processing system(DPS) in distributed processing system(DPS).
The structural representation for the processing node embodiment one that Fig. 8 provides for the present invention, as shown in figure 8, the processing node that the present embodiment is provided includes：At least one processor 801, at least one network interface 804 or other users interface 803, memory 805, and an at least communication bus 802.The load balance process device optionally includes user interface 803, including display, keyboard or pointing device.Memory 805 may include high-speed RAM memory, it is also possible to also including non-labile memory（Non- volatile memory), for example, at least one magnetic disk storage.Memory 805 can optionally be located remotely from the storage device of foregoing load balance process device comprising at least one.Deposit
Reservoir 805 stores execute instruction, when load balance process plant running, is communicated between processor 801 and memory 805, and processor 801 performs execute instruction and allows load balance process device to perform above-mentioned embodiment of the method.Operating system 806, comprising various programs, for realizing various basic businesses and handling hardware based task.
Processing node provided in an embodiment of the present invention, can perform the technical scheme of the embodiment of load balance process method, and its implementing principle and technical effect is similar, and here is omitted.The structural representation for the Master Control Center embodiment one that Fig. 9 provides for the present invention, as shown in figure 9, the Master Control Center that the present embodiment is provided includes：At least one processor 901, at least one network interface 904 or other users interface 903, memory 905, and an at least communication bus 902.The load balance process device optionally includes user interface 903, including display, keyboard or pointing device.Memory 905 may include high-speed RAM memory, it is also possible to also including non-labile memory（Non- volatile memory), for example, at least one magnetic disk storage.Memory 905 can optionally be located remotely from the storage device of foregoing load balance process device comprising at least one.Memory 905 stores execute instruction, when load balance process plant running, is communicated between processor 901 and memory 905, and processor 901 performs execute instruction and allows load balance process device to perform above-mentioned embodiment of the method.Operating system 906, comprising various programs, for realizing various basic businesses and handling hardware based task.
Master Control Center provided in an embodiment of the present invention, can perform the technical scheme of the embodiment of load balance process method, and its implementing principle and technical effect is similar, and here is omitted.
The structural representation for the distributed processing system(DPS) embodiment one that Figure 10 provides for the present invention, such as schemes
Shown in 10, the distributed processing system(DPS) of the present embodiment, including at least two processing nodes 1000, the structure of Fig. 4 Fig. 8 any device embodiments can be used by handling node 1000, it is accordingly, any one processing node 1000 can perform the technical scheme of the embodiment of load balance process method, and its implementing principle and technical effect is similar, and here is omitted.
The structural representation for the distributed processing system(DPS) embodiment two that Figure 11 provides for the present invention, such as schemes
Shown in 11, the distributed processing system(DPS) of the present embodiment, including at least two processing nodes 1010 and Master Control Center 1020, handling node 1010 can be using the structure of the prior art for handling node, Master Control Center 1020 can use the structure of Fig. 4 Fig. 8 any device embodiments, and it accordingly can perform the technical scheme of the embodiment of load balance process method, its implementing principle and technical effect is similar, and here is omitted.
The embodiment of the present invention also provides the processing node in a kind of distributed processing system(DPS), including at least two embodiments as shown in Figure 8.
The embodiment of the present invention provides the Master Control Center in a kind of distributed processing system(DPS), including two processing nodes and embodiment as shown in Figure 9 again.
The embodiment of the present invention also provides a kind of computer-readable medium, comprising computer executed instructions, and the computer executed instructions may be used to the processing node of any one in above-described embodiment and perform above-mentioned load balance process method.
The embodiment of the present invention provides a kind of computer-readable medium again, comprising computer executed instructions, and the computer executed instructions may be used to the Master Control Center of any one in above-described embodiment and perform above-mentioned load balance process method.
One of ordinary skill in the art will appreciate that：Realizing all or part of step of above method embodiment can be completed by the related hardware of programmed instruction, and foregoing program can be stored in a computer-readable medium, and the program upon execution, performs the step of including above method embodiment；And foregoing computer-readable medium includes：ROM, RAM, magnetic disc or CD etc. are various can be with the medium of store program codes.
Finally it should be noted that：The above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although the present invention is described in detail with reference to the foregoing embodiments, it will be understood by those within the art that：It can still modify to the technical scheme described in foregoing embodiments, or carry out equivalent substitution to which part technical characteristic；And these modifications or replacement, the essence of appropriate technical solution is departed from the spirit and scope of various embodiments of the present invention technical scheme.
Claims (1)
- Claims1st, a kind of load balance process method, it is characterised in that including：Obtain the average loads of all processing nodes in the in distributed processing system(DPS) first the first load for handling node and the distributed processing system(DPS)；If first load is more than the average load, it will be distributed at least one dummy node on the first processing node and move in the distributed processing system(DPS) load less than at least one processing node of the average load.2nd, according to the method described in claim 1, it is characterized in that, described at least one dummy node that will be distributed on the first processing node moves to load in the distributed processing system(DPS) and is less than at least one processing node of the average load, including：Load total amount on the first processing node is will be distributed over to be less than or equal at least one dummy node of overload quantity and move to underload total amount in the distributed processing system(DPS) to be more than or equal at least one processing node of the load total amount, the overload quantity is the load capacity that the described first load exceedes the average load, and the underload total amount is less than the load capacity sum of the average load for the load of each processing node at least one described processing node.3rd, the method according to any one of claim 1 or 2, it is characterized in that, before described at least one dummy node that will be distributed on the first processing node is moved at least one the processing node for being loaded in the distributed processing system(DPS) and be less than the average load, in addition to：The load information of each processing node in the distributed processing system(DPS) is obtained, the load information includes the load information that each dummy node on node is throughout managed in the distribution；According to the load information of each processing node, it is determined that at least one described processing node；Migration request message is sent at least one described processing node, the migration request message carries the information of at least one dummy node；Receive the migration request response message of at least one processing node；Described at least one dummy node that will be distributed on the first processing node moves to load in the distributed processing system(DPS)：If the migration request response message is success response message, it will be distributed at least one described dummy node on the first processing node and move at least one described processing node.4th, the method according to any one of claim 1 or 2, it is characterised in that described to incite somebody to action At least one dummy node being distributed on the first processing node moves to load in the distributed processing system(DPS) and is less than at least one processing node of the average load, including：Update the mapping table between the first processing node and at least one described dummy node；Other processing nodes into the distributed processing system(DPS) send fresh information, so that other processing nodes in the distributed processing system(DPS) update the mapping table between the processing node being locally stored and dummy node according to the fresh information.5th, the method according to any one of claim 1 or 2, it is characterized in that, after described at least one dummy node that will be distributed on the first processing node is moved at least one the processing node for being loaded in the distributed processing system(DPS) and be less than the average load, in addition to：By on the corresponding data syn-chronization of at least one dummy node at least one described processing node.6th, a kind of load balance process device, it is characterised in that including：Acquisition module, the average loads for obtaining all processing nodes in the in distributed processing system(DPS) first the first load for handling node and the distributed processing system(DPS)；Processing module, if first load is more than the average load, will be distributed at least one dummy node on the first processing node and moves in the distributed processing system(DPS) load less than at least one processing node of the average load.7, device according to claim 6, it is characterized in that, the processing module, it is less than or equal at least one dummy node of overload quantity and moves to underload total amount in the distributed processing system(DPS) to be more than or equal at least one processing node of the load total amount specifically for will be distributed over load total amount on the first processing node, the overload quantity is load capacity of described first load more than the average load, the underload total amount is less than the load capacity sum of the average load for the load of each processing node at least one described processing node.8th, the device according to any one of claim 6 or 7, it is characterised in that also including determining module, sending module and receiving module；The acquisition module, is additionally operable to obtain the load information of each processing node in the distributed processing system(DPS), and the load information includes the load information for each dummy node being distributed on each processing node；The determining module, for according to it is described it is each processing node load information, it is determined that it is described extremely Few processing node；The sending module, for sending migration request message at least one described processing node, the migration request message carries the information of at least one dummy node；The receiving module, for receiving the migration request response message that at least one described processing node feeds back according to the migration request message；If the processing module is success response message specifically for the migration request response message, it will be distributed at least one dummy node on the first processing node and move at least one described processing node.9th, the device according to any one of claim 6 or 7, it is characterised in that also include：Update module, for updating the mapping table between the first processing node and at least one described dummy node；The sending module, other processing nodes being additionally operable into the distributed processing system(DPS) send fresh information, so that other processing nodes in the distributed processing system(DPS) update the mapping table between the processing node being locally stored and dummy node according to the fresh information.10th, the device according to any one of claim 6 or 7, it is characterised in that also include：Synchronization module, at least one described dummy node corresponding data to be synchronized at least one described processing node.11, the device according to any one of claim 6 or 7, it is characterised in that described device is processing node or Master Control Center.12nd, a kind of processing node, it is characterized in that, including processor and memory, the memory storage execute instruction, when described device is run, communicated between the processor and the memory, execute instruction described in the computing device causes described device to perform the method as described in claim 1 to 5.13rd, a kind of Master Control Center, it is characterized in that, including processor and memory, the memory storage execute instruction, when described device is run, communicated between the processor and the memory, execute instruction described in the computing device causes described device to perform the method as described in claim 1 to 5.14th, a kind of distributed processing system(DPS), it is characterised in that including：At least two such as rights will Seek the processing node described in 11.15th, a kind of distributed processing system(DPS), it is characterised in that including：At least two processing nodes and Master Control Center as claimed in claim 11.16th, a kind of distributed processing system(DPS), it is characterised in that including at least two processing nodes as claimed in claim 12.17th, a kind of distributed processing system(DPS), it is characterised in that including：At least two processing nodes and Master Control Center as claimed in claim 13.18th, a kind of computer-readable medium, it is characterised in that comprising computer executed instructions, the computer executed instructions are used to make processing node perform claim require the method described in 1 to 5 any one.19th, a kind of computer-readable medium, it is characterised in that comprising computer executed instructions, the computer executed instructions are used to make Master Control Center perform claim require the method described in 1 to 5 any one.
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/CN2012/082382 WO2014047902A1 (en) | 2012-09-28 | 2012-09-28 | Load balancing method, device, system and computer readable medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103918239A true CN103918239A (en) | 2014-07-09 |
Family
ID=50386874
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201280001746.3A Pending CN103918239A (en) | 2012-09-28 | 2012-09-28 | Load balancing method, device, system and computer readable medium |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN103918239A (en) |
| WO (1) | WO2014047902A1 (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104301243A (en) * | 2014-09-22 | 2015-01-21 | 华为技术有限公司 | Load control method and device |
| CN104702691A (en) * | 2015-03-13 | 2015-06-10 | 华为技术有限公司 | Distributed load balancing method and device |
| CN107656813A (en) * | 2017-09-29 | 2018-02-02 | 上海联影医疗科技有限公司 | The method, apparatus and terminal of a kind of load dispatch |
| CN110633151A (en) * | 2019-09-20 | 2019-12-31 | 北京小米移动软件有限公司 | Method, device and storage medium for balancing distributed message issuing cluster partitions |
| CN110753372A (en) * | 2018-07-24 | 2020-02-04 | 中兴通讯股份有限公司 | Information processing method and device in baseband processing separation architecture and storage medium |
| CN111580968A (en) * | 2020-05-07 | 2020-08-25 | 广西大学 | Medical cloud platform load automatic balancing method, system and medium based on fog computing |
| CN114666335A (en) * | 2022-03-21 | 2022-06-24 | 北京计算机技术及应用研究所 | DDS-based distributed system load balancing device |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106897116A (en) * | 2017-02-27 | 2017-06-27 | 郑州云海信息技术有限公司 | A kind of virtual machine migration method and device |
| CN109617989B (en) * | 2018-12-28 | 2021-11-26 | 浙江省公众信息产业有限公司 | Method, apparatus, system, and computer readable medium for load distribution |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101883113A (en) * | 2010-06-25 | 2010-11-10 | 中兴通讯股份有限公司 | Method and physical nodes for realizing overlay network load balance |
| CN102480502A (en) * | 2010-11-26 | 2012-05-30 | 联想(北京)有限公司 | I/O load equilibrium method and I/O server |
| WO2012121736A1 (en) * | 2011-03-09 | 2012-09-13 | Unisys Corporation | Runtime virtual process creation for load sharing |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101504620A (en) * | 2009-03-03 | 2009-08-12 | 华为技术有限公司 | Load balancing method, apparatus and system of virtual cluster system |
-
2012
- 2012-09-28 WO PCT/CN2012/082382 patent/WO2014047902A1/en active Application Filing
- 2012-09-28 CN CN201280001746.3A patent/CN103918239A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101883113A (en) * | 2010-06-25 | 2010-11-10 | 中兴通讯股份有限公司 | Method and physical nodes for realizing overlay network load balance |
| CN102480502A (en) * | 2010-11-26 | 2012-05-30 | 联想(北京)有限公司 | I/O load equilibrium method and I/O server |
| WO2012121736A1 (en) * | 2011-03-09 | 2012-09-13 | Unisys Corporation | Runtime virtual process creation for load sharing |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104301243A (en) * | 2014-09-22 | 2015-01-21 | 华为技术有限公司 | Load control method and device |
| CN104301243B (en) * | 2014-09-22 | 2018-06-15 | 华为技术有限公司 | A kind of load control method and device |
| CN104702691A (en) * | 2015-03-13 | 2015-06-10 | 华为技术有限公司 | Distributed load balancing method and device |
| CN104702691B (en) * | 2015-03-13 | 2017-12-01 | 华为技术有限公司 | Distributed load equalizing method and device |
| CN107656813A (en) * | 2017-09-29 | 2018-02-02 | 上海联影医疗科技有限公司 | The method, apparatus and terminal of a kind of load dispatch |
| CN110753372A (en) * | 2018-07-24 | 2020-02-04 | 中兴通讯股份有限公司 | Information processing method and device in baseband processing separation architecture and storage medium |
| CN110753372B (en) * | 2018-07-24 | 2023-05-30 | 中兴通讯股份有限公司 | Information processing method, device and storage medium in baseband processing separation architecture |
| CN110633151A (en) * | 2019-09-20 | 2019-12-31 | 北京小米移动软件有限公司 | Method, device and storage medium for balancing distributed message issuing cluster partitions |
| CN111580968A (en) * | 2020-05-07 | 2020-08-25 | 广西大学 | Medical cloud platform load automatic balancing method, system and medium based on fog computing |
| CN111580968B (en) * | 2020-05-07 | 2023-04-18 | 广西大学 | Medical cloud platform load automatic balancing method, system and medium based on fog computing |
| CN114666335A (en) * | 2022-03-21 | 2022-06-24 | 北京计算机技术及应用研究所 | DDS-based distributed system load balancing device |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2014047902A1 (en) | 2014-04-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103918239A (en) | Load balancing method, device, system and computer readable medium | |
| US10156986B2 (en) | Gang migration of virtual machines using cluster-wide deduplication | |
| US10705965B2 (en) | Metadata loading in storage systems | |
| US10901796B2 (en) | Hash-based partitioning system | |
| US9372726B2 (en) | Gang migration of virtual machines using cluster-wide deduplication | |
| Deshpande et al. | Gang migration of virtual machines using cluster-wide deduplication | |
| US10356150B1 (en) | Automated repartitioning of streaming data | |
| CN105320773A (en) | Distributed duplicated data deleting system and method based on Hadoop platform | |
| CN103942087A (en) | Virtual machine thermal migration method, related device and cluster computing system | |
| CN107423301B (en) | Data processing method, related equipment and storage system | |
| EP4170491A1 (en) | Resource scheduling method and apparatus, electronic device, and computer-readable storage medium | |
| Babu et al. | Virtual machine placement for improved quality in IaaS cloud | |
| CN111913670A (en) | Load balancing processing method and device, electronic equipment and storage medium | |
| Zacheilas et al. | Dynamic load balancing techniques for distributed complex event processing systems | |
| Tchana et al. | Software consolidation as an efficient energy and cost saving solution for a SaaS/PaaS cloud model | |
| CN109960579B (en) | Method and device for adjusting service container | |
| Fan et al. | A live migration algorithm for containers based on resource locality | |
| CN107391039B (en) | Data object storage method and device | |
| US10642520B1 (en) | Memory optimized data shuffle | |
| CN107920129A (en) | A kind of method, apparatus, equipment and the cloud storage system of data storage | |
| Breitgand et al. | Network aware virtual machine and image placement in a cloud | |
| CN101783814A (en) | Metadata storing method for mass storage system | |
| US20160117107A1 (en) | High Performance Hadoop with New Generation Instances | |
| Zhang et al. | Speeding up vm startup by cooperative vm image caching | |
| US20230012021A1 (en) | Feature Resource Self-Tuning and Rebalancing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20140709 |
|
| RJ01 | Rejection of invention patent application after publication |