CN1512380A - Load weighing method based on systematic grade diagnosis information - Google Patents
Load weighing method based on systematic grade diagnosis information Download PDFInfo
- Publication number
- CN1512380A CN1512380A CNA021594805A CN02159480A CN1512380A CN 1512380 A CN1512380 A CN 1512380A CN A021594805 A CNA021594805 A CN A021594805A CN 02159480 A CN02159480 A CN 02159480A CN 1512380 A CN1512380 A CN 1512380A
- Authority
- CN
- China
- Prior art keywords
- node
- resource
- application
- system level
- factor
- 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.)
- Granted
Links
Images
Abstract
The load weighing method based on system level diagnosis information includes for the computer group system comprising several host computer nodes to allocate the same application on several host computer nodes, of which at least one has the application in running state at any time; and for the control node to select some spare node to re-start the application or transfer the application from current node to some spare node. The said technological scheme can remove fault, take over faulty node, allocate task ever reasonably based on the difference in performance and load among the nodes and the condition of resource utilization. The method calculates the resource residue rate, and this provides firm data information guarantee in system level for equalizing load inside the system and raising overall performance.
Description
Technical field
The present invention relates to a kind of load balancing method, be meant a kind of take all factors into consideration node isomery and application differences especially, based on the load balancing method of system level diagnostic information based on system level diagnostic information.Belong to computer networking technology.
Background technology
Commercial Network of Workstation self has very strong dirigibility; Identical application can select different nodes to dispose, and same task also can be distributed to different nodes and finish.Along with the continuous expansion that commerce is used scale, the load that application place node is born is more and more heavier; Simultaneously, also might occur load between the node and distribute unbalanced phenomenon.The business application task has and the tangible different characteristics of science calculation task: task granularity is less, working time is relative shorter, does not need that generally task is carried out parallelization and handles; But the quantity of business application task is bigger, and the load of each node changes obviously in time.
In sum, can not only carry out static configuration according to subjective desire as global decisions activities such as " application deployment ", " Task Distribution " by the system manager.Actual loading according to each node, reasonably will be applied in the group of planes dynamically dispose, balanced with task dynamic assignment between a plurality of nodes, to be to realize the balanced and final key that improves the entire system performance of Network of Workstation internal burden, and the basis of everything be exactly the operating load of weighing in real time and accurately on each node.
At present, at problem of load balancing, people have done a large amount of work, really realize the load balancing of a system, need take all factors into consideration many-sided factors such as load information collection, load measurement, load distribution.As shown in Figure 1, prior art mainly concentrates on the improvement aspect of loads-scheduling algorithm to this way to solve the problem; But load how to weigh system all sidedly but is an ignored problems always.
Traditional method is often too simple: some system is fixing load factor of each program setting, weighs system load according to the number of times of program execution and the product of load factor.This method is not considered real-time system-level load diagnostic message (as CPU usage, memory usage etc.), thereby lacks dynamic and science.Though some system has considered system-level load information, and compare the relative size of load between each node on this basis, but lack analysis system's isomerism and application of difference.System's isomerism here mainly is meant the difference of resource of the same race on performance on the different nodes; For example: the difference of the arithmetic speed of two processors that processor had.Also may there are differences between the different application; For example: a large amount of internal memory of the application need consumption that has is not and high to network bandwidth requirement, and some is just in time opposite.
A kind of real reasonably cluster nodes load is weighed technology and should be helped to realize:
1. when joint behavior was identical, the operating load that is assigned on each node should be roughly the same;
2. not simultaneously, at joint behavior more load should be assigned to that performance is higher, on the more node of resource;
3. be that dissimilar application distributes the most suitable their running environment.
Summary of the invention
Fundamental purpose of the present invention is to propose a kind of load balancing method based on system level diagnostic information, be suitable for Network of Workstation, take all factors into consideration its node isomery and use between difference, make when joint behavior is identical, make the operating load that is assigned on each node should be roughly the same; At joint behavior not simultaneously, do not make that more load should be assigned to that performance is higher, on the more node of resource.
Another purpose of the present invention is to propose a kind of load balancing method based on system level diagnostic information, for dissimilar application distributes the most suitable their running environment.
The object of the present invention is achieved like this:
A kind of load balancing method based on system level diagnostic information, on the multiple host node, arbitrary moment has at least this application on the node to be in running status to the Network of Workstation of being made up of a plurality of host nodes with same application deployment; Control Node is selected backup node to restart application maybe will to use from present node and move to backup node.
Described Control Node is selected the maximum backup node of resources left to restart application maybe will to use from present node and move, and comprise at least:
Step 1: the resource allocation factor.
Step 2: resource allocation relies on the factor;
Step 3: acquisition system level diagnostic message;
Step 4: calculate the node comprehensive resources surplus ratio relevant with application;
Step 5: upgrade the preposition node of load balancing;
Step 6: judge that it still is that load balancing is used that application is deployed as high useful application, if high useful application then selects the most suitable backup node that this uses operation as new operation node; Finish;
Step 7: if load balancing is used, then dispose this application according to the weight of secondary node, the weight of node is high more, and then the task of Fen Peiing is just many more; Finish.
Described step 1 specifically comprises:
Step 11: be the generic resource of Network of Workstation definition application, and be each resource appointing system level diagnostic message;
Step 12: be Network of Workstation static configuration nodal information, and when adding new node, indicate all the resource kinds and the resource factor that this node comprises.
Described system level diagnostic information comprises at least: processor resource, memory source and Internet resources; Wherein, processor utilization is the system level diagnostic information of processor resource, and memory usage is the system level diagnostic information of memory source, and network interface card actual transfer rate/network interface card peak transfer rate is the system level diagnostic information of Internet resources.
When setting the resource factor of node, need to carry out following steps at least:
Step 121: the parameter index of selecting to represent this node resource performance level;
Step 122: calculate the average behavior parameter of this node resource in system according to following formula;
P＝(∑P
i/N)
Wherein,
P is the average behavior parameter of this resource,
P
iBe the actual performance parameter of this resource on the i node,
N is the interstitial content in the group of planes;
Step 123: the ratio that calculates the actual performance parameter and the average behavior parameter of this node resource according to following formula;
P’＝(P/ P)
Wherein,
P ' is the ratio of the actual performance parameter and the average behavior parameter of this node resource;
P is the average behavior parameter of this node resource in system;
P is the actual performance parameter of this node resource;
Step 124: set the numerical value of the resource factor, and the value of this resource factor 1 and P ' between.
Described step 4 is specially according to the comprehensive resources surplus ratio of following formula computing node at application:
Wherein,
The surplus ratio of the respective resources that remain obtains from node;
Res_factor is by the resource factor of computational resource on the node;
Dep_factor relies on the factor for the resource of using respective resources;
All_resource is all resources that application is concerned about.
The concrete operations of step 5 are: Control Node is periodically calculated the resources left rate of all nodes, then they directly or are after treatment used the preposition node that weight sends to the load balancing application as node, be used to upgrade out-of-date node and use weight.
Described preposition node is logically organized the node of a plurality of operation same application as a whole, and the distribution of responsible task.Preserve the weight that each uses the operation node in the preposition node, select the highest node of weight when this preposition node carries out the task distribution.Use the weight of operation node and determine that according to this node resource surplus surplus resources is many more, corresponding weighted value should be big more.
By above technical scheme, the present invention is more realistically according to the performance difference between each node, load difference and utilization of resources situation, make that the distribution of troubleshooting of faults and adapter, task is more reasonable, solved in the classic method system's isomerism and application of difference are considered not enough problem; Utilize resources left rate that this method calculates to provide solid system level data Information Assurance for the equilibrium that realizes system's internal burden and the raising of overall performance.
Description of drawings
Fig. 1 is a load balancing process synoptic diagram;
Fig. 2 is a load object model synoptic diagram of the present invention;
Fig. 3 uses high available model synoptic diagram for a group of planes;
Fig. 4 is a group of planes application load balancing model synoptic diagram.
Embodiment
The present invention is described in further detail below in conjunction with accompanying drawing and specific embodiment:
Network of Workstation is made up of a plurality of host nodes at hardware aspect, is made up of a plurality of application aspect software; Node is the aggregate of variety classes resource, can dispose a plurality of different application on same node; And resource is the basis of system's operation, and each resource all has corresponding system level diagnostic information, and for example, cpu busy percentage is the system level diagnostic information of processor resource.
Operating system can provide method or the instrument that obtains various resource diagnostic messages.If when running into certain resource and having a plurality of system level diagnostic information, for example, the state parameter of weighing CPU has: cpu busy percentage, CPU context switching frequency, waiting process queue length etc., can set weights to the influence degree of using operation for it by the diagnosis project, the weighted mean value of getting each checkup item then is as last system level diagnostic information, perhaps, if checkup item is less to using influence on system operation, can ignore.System only considers to influence big checkup item.
Different nodes can have the resource of identical type, but isomerism allows to there are differences in many aspects between the similar resource.Want to carry out load accurately and weigh, must take into full account the performance difference between the similar resource.Therefore, as the isomery parameter, when describing the resource of each node, will be that benchmark is set the resource factor with the resource factor with the average behavior level in the group of planes scope of this kind resource.The resource factor is high more, shows that this kind resource performance is excellent more on certain node.But attention: generally should not be simply with the absolute ratio of resource performance index between node numerical value as the resource factor.
Application is the program of disposing in a group of planes, can be externally or certain service internally is provided.According to self task characteristic, application can be divided into polytypes such as computation-intensive, I/O intensity and communications-intensive.Because dissimilar application have nothing in common with each other to the degree of attentiveness of various system resources, therefore concerning different application: the node load value that is drawn based on identical system level diagnostic information also should be different, therefore, load is relevant with service, that is: application is to the degree of dependence information of each resource, rely on the factor with resource and represent, when using for system configuration, resource allocation relies on the factor simultaneously.
Referring to Fig. 2: the object model that can get group of planes load thus.In Network of Workstation, hardware aspect is made up of many nodes, and the software aspect is made up of many application; Node has various resources, and different nodes can have same asset, also can have different resources; The resource information of this node is by the resource factor representation; Can dispose a plurality of different application on same node, this application relies on factor representation to the degree of dependence of various resources by resource.Node be applied as the relation of disposing, resource be applied as the relation of using.
The present invention is example explanation the technical program with the centralized control:
As shown in Figure 1: in Network of Workstation, can with same application deployment on the multiple host node so that high available service to be provided.Referring to Fig. 3, arbitrary moment always has this application on the node node1 to be in running status (active).When application in service is broken down (the own fault of node failure or application), Control Node (control node) selects a backup node (standby) to restart application, i.e. " application and trouble processing ".Fatal problem do not occur even operation is used, system also may need the backup node of moving to other from present node with using, i.e. " application migration ".No matter being any situation, is one on the question essence that is faced---how reasonably to select backup node.Principle of the present invention is: select the maximum backup node of resources left.
In Network of Workstation, also can on a plurality of nodes, move identical application simultaneously, whole application exists a preposition node (dispatcher) to be responsible for the distribution of task, it logically organizes the node of a plurality of operation same application as a whole, referring to as the application app1 among Fig. 4, use app2.In preposition node, preserve each and used the weight of operation node, should select the highest node of weight during the task distribution.Problem is: how the application weight of node is set, and principle of the present invention is that surplus resources is many more according to the remaining size how many node application weights is set of node resource, and corresponding weighted value should be big more.
At any one time, the load of node is not a monodrome, but a many-valued set all should have its application corresponding load at each application that is deployed on the node, based on identical system level diagnostic information, the application load of same node may be different.
The application load method of concrete measurement node is:
Step 1: the resource allocation factor; Specifically comprise:
Step 11: be the generic resource of Network of Workstation definition application, and specify the system level diagnostic information that to represent its load for each resource; For example: processor resource (CPU), memory source (MEMORY), Internet resources (NETWORK) etc., processor utilization is as the system level diagnostic information of processor resource, memory usage is as the system level diagnostic information of memory source, and network interface card actual transfer rate/network interface card peak transfer rate is as the system level diagnostic information of Internet resources;
Step 12: be Network of Workstation static configuration nodal information, when adding new node, indicate all resource kinds that node comprises, and the performance level of every kind of resource on this node---the resource factor;
When setting the resource factor, need further carry out:
Step 121: select a kind of parameter index that can represent the resource performance level,, and determine the average behavior parameter (P) of this resource in system as the CPU frequency speed of processor resource;
Step 122: to be benchmark compare (P/P) with the actual performance parameter (P) of certain node resource with the average behavior parameter draws ratio P ' with the average behavior parameter;
Step 12 3: rule of thumb value and experimental data are set the numerical values recited of the resource factor, the general resource factor answer value 1 and P ' between.
Table 1:
In table 1, the resource factor of two node node1, node2 is disposed.System definition three kinds of generic resources: processor resource, memory source and Internet resources, the performance index of weighing them are respectively CPU frequency speed, memory size and network interface card peak transfer rate;
When the resource component to certain node adds, deletes or changes, should in time upgrade resource factor configuration information.
Step 2: resource allocation relies on the factor; The keeper should dispose the degree of dependence of application to different resource when disposing application program, that is: resource relies on the factor.For certain application, it is high more that the pairing resource of certain resource relies on the factor, illustrates that this applications thirsts for obtaining this kind resource more, so the effect of this resource when the node load of weighing this application is just big more.
Keeper's work for convenience, predefine one cover resource of the present invention relies on masterplate, has set the dependence factor of each resource that is applicable to that particular type is used in each masterplate.The summation that all resources in each masterplate rely on the factor should equal 1.For example:
Table 2:
In table 2, define two resources and relied on masterplate (computation-intensive template and communications-intensive template).In compute-intensive applications, task is the highest to the desirability of processor resource; In communications-intensive was used, task was the highest to the network resources demand degree; In two types application, task all has higher requirement to internal memory.
Step 3: acquisition system level diagnostic message;
System level diagnostic information has been represented the current load state of node resource, and all nodes all should the system level diagnostic information with this locality regularly, the cycle send to other node in the group of planes.In a distributed control group of planes, can adopt the mode of broadcasting to send.In a centralized control group of planes, can adopt the Control Node of mode in a group of planes of point-to-point communication to report.No matter adopt which kind of mode, finally all will make Control Node see complete (comprising the survival node in all current group of planes), consistent (it is identical that the redundant information on the different nodes should keep) system level diagnostic information.
How to gather: each system level diagnostic information all should have special Agent execution to obtain.System can provide some basic Agents when initial, the user also can write corresponding Agent according to the form oneself of system's regulation, then it is configured in the system, and system has reserved calling interface for them.
In actual computation, the original diagnostic message about utilization factor need be converted to diagnostic message about surplus ratio, such conclusion that draws will can be used for the remaining general status of each node resource of comparison.As shown in table 3:
Table 3:
Step 4: calculate the node resource surplus ratio relevant with application;
Finish above-mentioned a series of configuration effort, and after periodically receiving the system level diagnostic information of each node, just can carry out the measurement of node surplus resources at each application.Here need to prove: the present invention does not directly weigh node load, but weighs the node surplus resources, because final objective is Task Distribution will be carried out to the maximum node of resources left;
At application, the resources left rate computing formula of node is as follows:
Wherein,
Remain is the surplus ratio of certain resource on the node;
Res_factor goes up the resource factor of certain resource for node N;
Dep_factor relies on the factor for the resource of using certain resource;
All_resource is all resources that application is concerned about.
Suppose to dispose simultaneously two and use app1, app2 on node node1 and node node2, this application app1 is a compute-intensive applications, and using app2 is that communications-intensive is used.Now according to above-mentioned formula, and configuration information that sets above and the diagnostic message that collects, use at each and to weigh all nodes in the resources left situation that engraves for the moment, referring to table 4:
Table 4:
Step 5: upgrade the preposition node of load balancing; Control Node is periodically calculated the resources left rate of all nodes, then their direct (or through suitably handling the back) is sent to the preposition node that load balancing is used as node application weight, is used to upgrade out-of-date node and uses weight.
Step 6: all be deployed as high useful application if use app1 with using app2, except the operation node, node node1 and node node2 are current available backup nodes (standby), and serve as the adapter that foundation is used with the data in the table 4.When the operation node breaks down, use app1 and will take over by node node1, use app2 and will take over by node node2.
Step 7: all be deployed as load balancing and use with using app2 if use app1, and serve as according to carrying out task scheduling with the data in the table 4, the new task of using app1 will be scheduled for node 1node1 and carry out; The new task of using 2app2 will be scheduled for node 2 (node2) and carry out.
It should be noted that at last: above embodiment only in order to the explanation the present invention and and unrestricted technical scheme described in the invention; Therefore, although this instructions has been described in detail the present invention with reference to each above-mentioned embodiment,, those of ordinary skill in the art should be appreciated that still and can make amendment or be equal to replacement the present invention; And all do not break away from the technical scheme and the improvement thereof of the spirit and scope of the present invention, and it all should be encompassed in the middle of the claim scope of the present invention.
Claims (10)
1, a kind of load balancing method based on system level diagnostic information is characterized in that: on the multiple host node, arbitrary moment has at least this application on the node to be in running status to the Network of Workstation of being made up of a plurality of host nodes with same application deployment; Control Node is selected backup node to restart application maybe will to use from present node and move to backup node.
2, the load balancing method based on system level diagnostic information according to claim 1 is characterized in that: described Control Node is selected the maximum backup node of resources left to restart application maybe will to use from present node and move, and comprise at least:
Step 1: the resource allocation factor.
Step 2: resource allocation relies on the factor;
Step 3: acquisition system level diagnostic message;
Step 4: calculate the node comprehensive resources surplus ratio relevant with application;
Step 5: upgrade the preposition node of load balancing;
Step 6: judge that it still is that load balancing is used that application is deployed as high useful application, if high useful application then selects the most suitable backup node that this uses operation as new operation node; Finish;
Step 7: if load balancing is used, then dispose this application according to the weight of secondary node, the weight of node is high more, and then the task of Fen Peiing is just many more; Finish.
3, the load balancing method based on system level diagnostic information according to claim 2, it is characterized in that: described step 1 specifically comprises:
Step 11: be the generic resource of Network of Workstation definition application, and be each resource appointing system level diagnostic message;
Step 12: be Network of Workstation static configuration nodal information, and when adding new node, indicate all the resource kinds and the resource factor that this node comprises.
4, the load balancing method based on system level diagnostic information according to claim 3, it is characterized in that: described system level diagnostic information comprises at least: processor resource, memory source and Internet resources; Wherein, processor utilization is the system level diagnostic information of processor resource, and memory usage is the system level diagnostic information of memory source, and network interface card actual transfer rate/network interface card peak transfer rate is the system level diagnostic information of Internet resources.
5, the load balancing method based on system level diagnostic information according to claim 3 is characterized in that: when setting the resource factor of node, need to carry out following steps at least:
Step 121: the parameter index of selecting to represent this node resource performance level;
Step 122: calculate the average behavior parameter of this node resource in system according to following formula;
P＝(∑P
i/N)
Wherein,
P is the average behavior parameter of this resource,
P
iBe the actual performance parameter of this resource on the i node,
N is the interstitial content in the group of planes;
Step 123: the ratio that calculates the actual performance parameter and the average behavior parameter of this node resource according to following formula;
P’＝(P/ P)
Wherein,
P ' is the ratio of the actual performance parameter and the average behavior parameter of this node resource;
P is the average behavior parameter of this node resource in system;
P is the actual performance parameter of this node resource;
Step 124: set the numerical value of the resource factor, and the value of this resource factor 1 and P ' between.
6, the load balancing method based on system level diagnostic information according to claim 2 is characterized in that: described step 4 is specially according to the comprehensive resources surplus ratio of following formula computing node at application:
Wherein,
The surplus ratio of the respective resources that remain obtains from node;
Res_factor is by the resource factor of computational resource on the node;
Dep_factor relies on the factor for the resource of using respective resources;
All_resource is all resources that application is concerned about.
7, the load balancing method based on system level diagnostic information according to claim 2, it is characterized in that: the concrete operations of step 5 are: Control Node is periodically calculated the resources left rate of all nodes, then they directly or are after treatment used the preposition node that weight sends to the load balancing application as node, be used to upgrade out-of-date node and use weight.
8, the load balancing method based on system level diagnostic information according to claim 2 is characterized in that: described preposition node is logically organized the node of a plurality of operation same application as a whole, and the distribution of responsible task.
9, according to claim 2 or 8 described load balancing methods, it is characterized in that: preserve the weight that each uses the operation node in the described preposition node, select the highest node of weight when this preposition node carries out the task distribution based on system level diagnostic information.
10, the load balancing method based on system level diagnostic information according to claim 9 is characterized in that: the weight of described application operation node determines that according to this node resource surplus surplus resources is many more, and corresponding weighted value should be big more.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 02159480 CN1251111C (en) | 2002-12-31 | 2002-12-31 | Load weighing method based on systematic grade diagnosis information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 02159480 CN1251111C (en) | 2002-12-31 | 2002-12-31 | Load weighing method based on systematic grade diagnosis information |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1512380A true CN1512380A (en) | 2004-07-14 |
| CN1251111C CN1251111C (en) | 2006-04-12 |
Family
ID=34237492
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 02159480 Expired - Fee Related CN1251111C (en) | 2002-12-31 | 2002-12-31 | Load weighing method based on systematic grade diagnosis information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1251111C (en) |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100451970C (en) * | 2005-09-27 | 2009-01-14 | 浪潮电子信息产业股份有限公司 | Method for realizing dynamic layout of high-performance server based on group structure |
| CN101374143B (en) * | 2007-08-21 | 2011-11-02 | 华为技术有限公司 | Method, system and apparatus for issuing object status |
| CN102521691A (en) * | 2011-12-05 | 2012-06-27 | 济南钢铁股份有限公司 | Intelligent method for unified schedule and automatic allocation of measurement tasks |
| CN102591921A (en) * | 2010-12-20 | 2012-07-18 | 微软公司 | Scheduling and management in a personal datacenter |
| CN103324554A (en) * | 2012-03-21 | 2013-09-25 | 日本电气株式会社 | Standby system device, a control method, and a program thereof |
| CN104182296A (en) * | 2013-05-23 | 2014-12-03 | 索尼公司 | Data processing system, data processing apparatus, and storage medium |
| CN104426795A (en) * | 2013-09-09 | 2015-03-18 | 中国电信股份有限公司 | Method, node scheduling server and system for load balancing |
| CN104580338A (en) * | 2013-10-29 | 2015-04-29 | 华为技术有限公司 | Service processing method, system and equipment |
| CN107204878A (en) * | 2017-05-27 | 2017-09-26 | 国网山东省电力公司 | A kind of annular escape system of certificate server and method |
| CN107645454A (en) * | 2016-07-22 | 2018-01-30 | 中国电信股份有限公司 | Virtualization and mixed networking method and system, the load balancing device of legacy equipment |
| CN107707612A (en) * | 2017-08-10 | 2018-02-16 | 北京奇艺世纪科技有限公司 | A kind of appraisal procedure and device of the resource utilization of load balancing cluster |
| CN108280007A (en) * | 2017-01-05 | 2018-07-13 | 中国移动通信集团福建有限公司 | A kind of method and apparatus for assessment equipment resource utilization |
| CN108414226A (en) * | 2017-12-25 | 2018-08-17 | 哈尔滨理工大学 | Fault Diagnosis of Roller Bearings under the variable working condition of feature based transfer learning |
| CN109643090A (en) * | 2016-06-24 | 2019-04-16 | 施耐德电子系统美国股份有限公司 | Dynamically promote non-boundary, method, system and the equipment of high availability M:N active configuration system administration |
| WO2020259326A1 (en) * | 2019-06-28 | 2020-12-30 | 深圳前海微众银行股份有限公司 | Signal transmission method and apparatus |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102467451B (en) * | 2010-11-09 | 2015-02-04 | 金蝶软件（中国）有限公司 | System and method for processing node memory resources in dual-server hot-backup mode |
-
2002
- 2002-12-31 CN CN 02159480 patent/CN1251111C/en not_active Expired - Fee Related
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100451970C (en) * | 2005-09-27 | 2009-01-14 | 浪潮电子信息产业股份有限公司 | Method for realizing dynamic layout of high-performance server based on group structure |
| CN101374143B (en) * | 2007-08-21 | 2011-11-02 | 华为技术有限公司 | Method, system and apparatus for issuing object status |
| US9268613B2 (en) | 2010-12-20 | 2016-02-23 | Microsoft Technology Licensing, Llc | Scheduling and management in a personal datacenter |
| CN102591921A (en) * | 2010-12-20 | 2012-07-18 | 微软公司 | Scheduling and management in a personal datacenter |
| CN102521691A (en) * | 2011-12-05 | 2012-06-27 | 济南钢铁股份有限公司 | Intelligent method for unified schedule and automatic allocation of measurement tasks |
| CN103324554A (en) * | 2012-03-21 | 2013-09-25 | 日本电气株式会社 | Standby system device, a control method, and a program thereof |
| CN104182296A (en) * | 2013-05-23 | 2014-12-03 | 索尼公司 | Data processing system, data processing apparatus, and storage medium |
| CN104426795A (en) * | 2013-09-09 | 2015-03-18 | 中国电信股份有限公司 | Method, node scheduling server and system for load balancing |
| CN104426795B (en) * | 2013-09-09 | 2017-09-12 | 中国电信股份有限公司 | Method, node scheduling server and system for load balancing |
| CN104580338A (en) * | 2013-10-29 | 2015-04-29 | 华为技术有限公司 | Service processing method, system and equipment |
| US11362961B2 (en) | 2013-10-29 | 2022-06-14 | Huawei Technologies Co., Ltd. | Service processing method and system and device |
| US10805231B2 (en) | 2013-10-29 | 2020-10-13 | Huawei Technologies Co., Ltd. | Service processing method and system and device |
| CN109643090A (en) * | 2016-06-24 | 2019-04-16 | 施耐德电子系统美国股份有限公司 | Dynamically promote non-boundary, method, system and the equipment of high availability M:N active configuration system administration |
| CN107645454B (en) * | 2016-07-22 | 2020-03-20 | 中国电信股份有限公司 | Hybrid networking method and system of virtualization and traditional equipment and load sharing device |
| CN107645454A (en) * | 2016-07-22 | 2018-01-30 | 中国电信股份有限公司 | Virtualization and mixed networking method and system, the load balancing device of legacy equipment |
| CN108280007B (en) * | 2017-01-05 | 2021-08-13 | 中国移动通信集团福建有限公司 | Method and device for evaluating equipment resource utilization rate |
| CN108280007A (en) * | 2017-01-05 | 2018-07-13 | 中国移动通信集团福建有限公司 | A kind of method and apparatus for assessment equipment resource utilization |
| CN107204878A (en) * | 2017-05-27 | 2017-09-26 | 国网山东省电力公司 | A kind of annular escape system of certificate server and method |
| CN107204878B (en) * | 2017-05-27 | 2018-01-02 | 国网山东省电力公司 | A kind of certificate server annular escape system and method |
| CN107707612A (en) * | 2017-08-10 | 2018-02-16 | 北京奇艺世纪科技有限公司 | A kind of appraisal procedure and device of the resource utilization of load balancing cluster |
| CN108414226B (en) * | 2017-12-25 | 2019-07-19 | 哈尔滨理工大学 | Fault Diagnosis of Roller Bearings under variable working condition based on feature transfer learning |
| CN108414226A (en) * | 2017-12-25 | 2018-08-17 | 哈尔滨理工大学 | Fault Diagnosis of Roller Bearings under the variable working condition of feature based transfer learning |
| WO2020259326A1 (en) * | 2019-06-28 | 2020-12-30 | 深圳前海微众银行股份有限公司 | Signal transmission method and apparatus |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1251111C (en) | 2006-04-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1251111C (en) | Load weighing method based on systematic grade diagnosis information | |
| CN1292346C (en) | System and method for performing task in distributing calculating system structure | |
| CN1306754C (en) | Method and system for balancing working load in network computing environment | |
| CN1956457A (en) | Method and apparatus for arranging mesh work in mesh computing system | |
| Zeng et al. | An integrated task computation and data management scheduling strategy for workflow applications in cloud environments | |
| Das et al. | Parallel processing of adaptive meshes with load balancing | |
| CN1956456A (en) | Method and apparatus for presenting resource demand in mesh computing system | |
| CN104050042B (en) | The resource allocation methods and device of ETL operations | |
| CN111444019B (en) | Cloud collaborative deep learning model distributed training method and system | |
| Kord et al. | An energy-efficient approach for virtual machine placement in cloud based data centers | |
| CN107861796B (en) | Virtual machine scheduling method supporting energy consumption optimization of cloud data center | |
| CN1523498A (en) | Dynamic processor redistribution between partitions in a computing system | |
| CN110347515B (en) | Resource optimization allocation method suitable for edge computing environment | |
| Keat et al. | Scheduling framework for bandwidth-aware job grouping-based scheduling in grid computing | |
| CN111090456A (en) | Construction method, device, equipment and medium for deep learning development environment | |
| CN110084507B (en) | Scientific workflow scheduling optimization method based on hierarchical perception in cloud computing environment | |
| He et al. | Energy-efficient framework for virtual machine consolidation in cloud data centers | |
| Keerthika et al. | A multiconstrained grid scheduling algorithm with load balancing and fault tolerance | |
| Zhang et al. | An energy-aware host resource management framework for two-tier virtualized cloud data centers | |
| Yusoh et al. | Clustering composite SaaS components in cloud computing using a grouping genetic algorithm | |
| JP2011141703A (en) | System, method and program for arranging resource | |
| CN111209098A (en) | Intelligent rendering scheduling method, server, management node and storage medium | |
| Mandal et al. | An empirical study and analysis of the dynamic load balancing techniques used in parallel computing systems | |
| Joshi et al. | Enhancement of cloud performance metrics using dynamic degree memory balanced allocation algorithm | |
| Balla et al. | Towards a predictable open source FaaS |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20060412 Termination date: 20201231 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |