CN102664812B - Two-stage service system load forecast and balancing method integrating service forecast and real-time load - Google Patents
Two-stage service system load forecast and balancing method integrating service forecast and real-time load Download PDFInfo
- Publication number
- CN102664812B CN102664812B CN201210148122.6A CN201210148122A CN102664812B CN 102664812 B CN102664812 B CN 102664812B CN 201210148122 A CN201210148122 A CN 201210148122A CN 102664812 B CN102664812 B CN 102664812B
- Authority
- CN
- China
- Prior art keywords
- service
- load
- migration
- business
- index
- 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.)
- Expired - Fee Related
Links
Abstract
The invention discloses a two-stage service system load forecast and balancing method integrating service forecast and real-time load, aiming to solve the limitation that the software service balance and service migration rely solely on the system load index. The traditional load balance depends on the short-term even or real-time system load index, which is substantially an internal reason for the representation of the software service load and the overloaded load; and an external factor is due to the increase of the traffic loaded by the software service, while the external factor affects the load of the service system by the internal reason; meanwhile, since the information reflected by the real-time load index cannot be ahead of time, and the frequent index changes easily brings about the problems of short balancing processing time and shaking service migration and the like. The method provided by the invention proceeds from the service forecast, and can be used for obtaining a possible variation trend of the service system load according to the rise and fall of the service bearing capacity, making an accurate forecast based on a probabilistic algorithm, combining with the real-time load index of the system and comprehensively making a balancing decision for optimized deployment and dynamic migration of the software service.
Description
Technical field
The present invention relates to a kind of software service system load predicting method, particularly relate to two benches load estimation and the equalization methods of the prediction of a kind of fused business and real time load.
Background technology
Service compute brings the major transformation of software architecture, namely (SaaS) is served for Typical Representative with Service-Oriented Architecture Based (SOA), software, the existing forms of software systems, published method, operation and maintenance mode have had the change of internal, facilitate the development of software architecture and theory and technology thereof.SaaS application system generally adopts many tenants (multi-tenant) pattern, meeting each tenant's performance index, while fluctuation etc. as continuous service, service throughput, adaptation service ability requires, a side is provided as service, by the Optimization deployment of service system, improve its overall resource utilization, reduce costs.This just requires that service system can adapt to dynamic change and dynamic expansion, takes into account efficiency, more efficiently utilize application resource under the prerequisite of meeting consumers' demand, and can maximize the resource-sharing between different user.Because service-user is numerous, the reliability in time of online service system requires also higher.
In the SaaS application of current main flow, end user use one only for used be not all virtual systems, this " system " is formed by some service component assemblies, therefore, the Optimization deployment ability of the corresponding assembly set of SaaS application becomes the key promoting SaaS application service performance and resource utilization.Different with guaranteed reliability from traditional optimizing application, by the dynamic migration of serviced component, can realize disposing more efficiently, to support operation demand that is real-time, online service system.
Services migrating is under software service trend, for one that proposes based on the system of service arrangement in Service-Oriented Architecture Based new problem, refers to and services migrating is deployed to new Home Location, and guarantee the minimum negative effect to operation system.Dynamically, extensive, need in the applied environment of automatic business processing, services migrating is one more preferably service arrangement method.Pass through services migrating, availability, the reliability of performance and the expansion demand of service ability of serving can be guaranteed with less migration cost, thus realize flexibility and the extensibility of whole service system, be a kind of available strategy guaranteeing that software service system persistent high efficiency is runed.
Services migrating can be divided into off-line to move and online migration type; Can be divided into again from service state maintenance has state transition and stateless to move two kinds; In the realization of migration, have Autonomic Migration Framework, semi automatic migration the multiple different mode such as to move with manual.
Off-line migration refers to position that service to be transferred to and reconfigure and restarts.Need suspend or stop current service in the process of migration, therefore the operation of normal business system is inevitably disturbed in off-line migration, thus causes the decline of service disruption and service quality; In addition, many tenants service system run time access and business load elasticity large, the temporary service system hydraulic performance decline caused due to the fluctuation of load or scarce capacity problem are very general, if the service that stops frequently is moved, certainly will cause the decline of system availability.
So off-line moves and is not suitable for the operation demand of current SaaS service system, need the method adopting online migration, guarantee the transparency of transition process to user, not the execution of interference user business; Wherein, the selection of migration opportunity and migratory route, is also move whether effective key factor, needs to formulate good migration strategy in conjunction with service load situation.Transition process be service host environment change, can have an impact to current service context, stateless services migrating due to do not need consider before service content, relatively simply.But in practical business application, use the services migrating of state in a large number, need the consistency of guaranteeing to move front and back service context, thus make the execution of service unaffected; But because the feature such as loose couplings, isomery, distribution of serving itself makes the expression of state, transfer and maintenance become very difficult.
Manual migration needs manually to judge, and under the intervention of keeper, starts migrator, completes the migration of service to new Home Location; Semi automatic migration then can rely on the evaluation of index, makes migration decision-making automatically, sends migration instruction, and completes the operation of migration by keeper; Autonomic Migration Framework is then migration strategy according to service system or load variations, real-time response migration demand, and completes migration work automatically.
Need during services migrating to consider all many-sided factors such as Service Instance, state representation, data, once there is migration operation repeatedly, namely move jitter problem, larger overhead can be brought.
Solve services migrating problem, first will find the reason causing migration demand.Although services migrating is the consideration for avoiding system crash or optimizing operation efficiency; its immediate cause is the change of system load; but cause the most basic reason of this change to be the change of the practical business bearing capacity of service system, this embodies system external cause (service bearer amount) by acting on the mechanism process of internal cause (service system) influential system performance.In practical application, the size of typical traffic is directly depended in the load of service system, and as purchase order amount in e-commerce order system increases, user concurrent access amount is large, significantly can increase, thus increase the load of service to the request of this service.Therefore from business dope send out make migration decision-making be a kind of reasonable contemplation, but, the operational indicator affecting service system load has a lot, they are different to the factor of influence of final load, mutual restriction and acting in conjunction are also different, and these problems are often difficult to determine due to the complexity of service system and diversity.Can consider to introduce importance degree theoretical, particularly based on multifactor coefficient altogether because importance degree is theoretical, calculate the influence degree of each factors vary to systematic function.Traffic forecast can adopt various mathematical tool, and the method based on probability is wherein comparatively ripe a kind of means.Commercial software service, as the operation system by market Natural regulation, the change procedure of the service traffics of its actual bearer can be regarded as a kind of random process meeting Markov chain feature, namely, the probability distribution of t+1 moment system mode is only relevant with the state of t, have nothing to do with the state before t, and have nothing to do to the state transitions in t+1 moment and the value of t from t, therefore can be used for Markov chain to predict the variation tendency of service traffics.
Summary of the invention
The present invention is for solving service system load estimation and equalization problem, for in software service equilibrium and services migrating, a kind of fused business that the simple limitation relying on real-time system loading index proposes is predicted and the two benches load estimation of real time load and equalization methods.Conventional load equilibrium depends on the even real-time system load index of short-term, essence is the presentation of software service load and the internal cause of load excess load, its external cause is then the growth due to software service institute bearer traffic, and external cause affects the load of service system by internal cause; Meanwhile, due to the reflection of real time load index information can not in advance, index changes frequently thus easily brings equilibrium treatment time is short and pressing and the problem such as services migrating shake.First the present invention dopes from business and sends out, draw the possible variation tendency of service system load according to the fluctuation of service bearer amount, and make based on probabilistic algorithm and predicting accurately, then coupling system real time load index, comprehensively make balanced decision-making, for Optimization deployment and the dynamic migration of software service.
To achieve these goals, the present invention adopts following technical scheme.
The two benches load estimation of fused business prediction and real time load and an equalization methods, comprise the steps:
1) business load index is altogether because of Significance Analysis;
The leading indicator of service load is affected based on importance degree theory analysis, determine the principal element affecting service load He cause service crashes, probability of use importance degree method, under the condition of known index delta data, evaluates the influence degree of each index to service system bearer traffic; Based on multi-service index altogether because importance degree evaluation model and algorithm weigh multi-service index co-variation to the influence degree of systematic function, thus determine the key business index needing prediction.
2) based on the traffic forecast of Markov chain model;
In the Service Operation of reality, the change of service bearer amount has certain rule, is the result under the combined influence of the factors such as the cyclic variation of the market demand, the transition of enterprise operation strategy; Therefore, with the change based on bayes method and markovian probabilistic method prediction service bearer amount, from macroscopic view and trend, accurately grasp the load variations trend of service, and in conjunction with the monitoring of real-time service load, overall merit provides migration decision-making.
3) system real time load is weighed;
The information when load of node runs with some usually reflects, such as cpu busy percentage, memory usage, swapace utilance etc., but these indexs can only reflect the load state of whole node, can not reflect the pressure that the process of process Web request is born.The present invention proposes a kind of load balancing method and the loading index utility function that can reflect online service request change actual conditions, in conjunction with actual operating data measuring and calculating and checking, determines rational parameter.
4) serviced component migration decision-making is determined;
An important function of services migrating is the business load of balancing service, to reduce service response time, improve quality and the availability of service, therefore need to formulate rational migration strategy to ensure to carry out services migrating in time when the business load of service carrier is too high.When selecting the target Home Location needing example and the migration of moving, according to the migration decision-making drawn, complex transfer efficiency and moving costs calculate and provide migratory route.
The invention has the beneficial effects as follows, change the simple limitation relying on real-time system loading index in software service equilibrium and services migrating; It is ageing that tendency and the real time load of integrated service prediction are weighed, and can draw more accurate and comprehensive prediction, thus win the processing time of preciousness for load balancing; More rational migratory route can also be made based on advanced prediction to select, thus avoid migration jitter problem for occurring.Therefore, the present invention can be widely used in Optimization deployment and the dynamic migration of software service.
Accompanying drawing explanation
Fig. 1 is the inventive method schematic diagram.
Fig. 2 is online service migratory system prototype frame schematic diagram.
Embodiment
Below in conjunction with accompanying drawing and embodiment, the invention will be further described.
Preferred forms of the present invention is combined in commercial service system, and the model provided by system and method call interface use present system.
The fused business that the present embodiment proposes is predicted and the two benches load estimation of real time load and equalization methods, as shown in Figure 1.The method that services migrating process relates to is as follows:
1) business load index is altogether because of Significance Analysis;
The leading indicator of service load is affected based on importance degree theory analysis, determine the principal element affecting service load He cause service crashes, probability of use importance degree method, under the condition of known index delta data, evaluates the influence degree of each index to service system bearer traffic; Based on multi-service index altogether because importance degree evaluation model and algorithm weigh multi-service index co-variation to the influence degree of systematic function, thus determine the key business index needing prediction.
2) based on the traffic forecast of Markov chain model;
In the Service Operation of reality, the change of service bearer amount has certain rule, is the result under the combined influence of the factors such as the cyclic variation of the market demand, the transition of enterprise operation strategy; Therefore, with the change based on bayes method and markovian probabilistic method prediction service bearer amount, from macroscopic view and trend, accurately grasp the load variations trend of service, and in conjunction with the monitoring of real-time service load, overall merit provides migration decision-making.
In the service system of many tenants, the excess load of tenant's business is the one of the main reasons that services migrating demand occurs, the extreme peak value access of the online business system that the such as busy season in red-letter day brings.And in the Service Operation of reality, the change of tenant's business has certain rule, it is the result under the combined influence of the factors such as the cyclic variation of the market demand, the transition of enterprise operation strategy, therefore, service tenant applied business load predicting method, accurately can grasp the load variations trend of service from macroscopic view and trend.The basic step of prediction is as follows:
The first step: determine the major parameter weighing business load, and this business is carried out to the state demarcation of science, at least mark off two states, these states are exactly the content that will predict;
Second step: statistical survey is carried out to the current state probability of the various states of service parameter change, the current residing state of the business that namely determines;
3rd step: the transition probability of each state future development of service parameter is measured, if be steady development transfer within the following long period, then each transfer of system mode can keep identical transition probability; If be the concussion that rises and falls within the following long period, then state often shifts once just needs to measure once transition probability, and the time interval that state shifts at every turn then needs to determine according to concrete business;
4th step: each state probability current according to the service parameter selected and state transition probability use the method for matrix, deduces out following after several times transfer, remains on the probable value of a certain state, thus judge.
3) system real time load is weighed;
Server is mainly used for receiving the request of user, and therefore, the quantity of user's request has reacted the performance of this server to a certain extent.The node of different configuration is different to the disposal ability of request, but the performance of different servers to request process can be obtained by test experiments.
Process the performance of user's request when carrying out response service device peak load with rpmC value, rpm (Requests Per Minutes), refers to the new number of request of interior system process per minute; C refers to the atomic operation in rpm, and namely dissimilar request can represent with the form of kilounit (KC).The computational process of rpmC value is as follows:
Suppose server i there be j service, following symbol is expressed as:
R
ij: the jth on server i serves largest request number during peak value;
T
ij: the response time threshold value of a jth service on server i;
Regulation, during server peak load, cpu busy percentage is 75%, then:
Based on Markov Chain method, system synthesis predicts that the tenant that serves drawn is applied in the business fluctuation situation of future time instance and judges whether this node transships by the rpmC value of monitoring each node, determine whether that Trigger services moves.
4) serviced component migration decision-making is determined
An important function of services migrating is the business load of balancing service, to reduce service response time, improve quality and the availability of service, therefore need to formulate rational migration strategy to ensure to carry out services migrating in time when the business load of service carrier is too high.When selecting the target Home Location needing example and the migration of moving, according to the migration decision-making drawn, complex transfer efficiency and moving costs calculate and provide migratory route.
Migration strategy mainly comprises the content of two aspects: be first weigh based on traffic forecast and real time load, determine to move decision-making; Then be select migration example and target Home Location based on transport efficiency and moving costs.When selecting the target Home Location needing example and the migration of moving, according to the migration demand drawn, complex transfer efficiency and moving costs calculate and provide migratory route.
1. migration service example is selected;
In order to avoid selecting the improper migration jitter phenomenon caused due to Service Instance, need select the example that finally will move in conjunction with traffic forecast, concrete steps are:
The first step: server i has n Service Instance, the mark value of each Service Instance of initialization is 0, wherein, mark value be 1 expression not transportable, mark value is the expression easily extensible of 2.When this server load, find the Service Instance that number of request is this moment maximum, be designated as: S
ik;
Second step: except S
ikoutside Service Instance, an entirety is regarded in other service as, is designated as: W, S
ij∈ W, j ≠ k;
3rd step: the value calculating following parameters;
R
ik: the number of request of Service Instance k;
predict the largest request number in one period in the future;
R
i: the number of request of the service of corresponding W;
the prediction largest request number of corresponding W.
4th step: calculate
Situation 1. Δ K > 0 & & Δ W > 0;
If R
ik> Δ W, eligiblely can move this example, if the migration of this example consumes too large, be 1 by this service mark, the example selecting number of request maximum from residue service, and the mark value of this example is 0, turn back to second step, otherwise move this example; Return this Service Instance, prepare migration;
If R
ik< Δ W, moves this Service Instance and can not solve load, is 1 by this service mark, the example selecting number of request maximum from residue service, and the mark value of this example is 0, turns back to second step;
Situation 2. Δ K > 0 & & Δ W < 0;
If Δ K is < | Δ W|, then do not need migration;
If Δ K is > | Δ W|, then need to move this example, if the migration of this example consumes too large, then needs to expand this example and change the mark value of this example into 2.
Situation 3. Δ K < 0 & & Δ W > 0;
If | Δ K| < Δ W & & R
ik> Δ W, needs migration example, if the migration of this example consumes too large, is 1 by this service mark, the example selecting number of request maximum from residue service, and the mark value of this example is 0, turns back to second step, otherwise moves this example.Return this Service Instance, prepare migration.
If | Δ K| < Δ W & & R
ik< Δ W, moves this Service Instance and can not solve load, is 1 by this service mark, the example selecting number of request maximum from residue service, and the mark value of this example is 0, turns back to second step;
If | Δ K| > Δ W, then do not need migration;
Situation 4. Δ K < 0 & & Δ W < 0, does not need migration example.
5th step: if last neither one example is transportable, then extending marking value is a Service Instance of 2.
2. destination node (weighing formula according to load to obtain) is selected;
Determine the Service Instance S that will move
kafter, predict the largest request number of this example in following a period of time, and find suitable server according to the response time threshold value of this implementation requirements, selection step is:
The first step: suppose S
kmove on server i, then remember R
ikfor S
kthe largest request number of prediction, T
ikfor S
kresponse time threshold value;
Second step: calculate, S
kwhen being deployed on i, the average response time threshold value of all services during i peak load:
3rd step: calculate S
kwhen joining on i, the largest request number that can bear;
If
new service can move on server i, otherwise can not move.
Based on the online service migratory system principle of the inventive method, as shown in Figure 2:
Online service migratory system supports subsystem as the backstage O&M of business software service system, is made up of request forward device and services migrating controller.
Request forward device operates on monitoring server, by message components, and request locator, services migrating engine, migration strategy module, load evaluate module and monitoring module composition.Main function be according to serve tenant apply business load and the serviced component faced real time access load start migration, and control services migrating whole flow process and according to migration after state perform request forward.
Services migrating controller operates on each service multihome node.By message components, services migrating engine and monitor client composition.During the operation of primary responsibility collector node, the migration of bag and state information is issued in information and service.
By reference to the accompanying drawings the specific embodiment of the present invention is described although above-mentioned; but not limiting the scope of the invention; one of ordinary skill in the art should be understood that; on the basis of technical scheme of the present invention, those skilled in the art do not need to pay various amendment or distortion that creative work can make still within protection scope of the present invention.
Claims (2)
1. fused business is predicted and the two benches load estimation of real time load and an equalization methods, comprises the steps:
1) business load index is altogether because of Significance Analysis;
The leading indicator of service load is affected based on importance degree theory analysis, determine the principal element affecting service load He cause service crashes, probability of use importance degree method, under the condition of known index delta data, evaluates the influence degree of each index to service system bearer traffic; Based on multi-service index altogether because importance degree evaluation model and algorithm weigh multi-service index co-variation to the influence degree of systematic function, thus determine the key business index needing prediction;
2) based on the traffic forecast of Markov chain model;
In the Service Operation of reality, the change of service bearer amount has certain rule, be the cyclic variation of the market demand, enterprise operation strategy transition factor combined influence under result; Therefore, with the change based on bayes method and markovian probabilistic method prediction service bearer amount, from macroscopic view and trend, accurately grasp the load variations trend of service, and in conjunction with the monitoring of real-time service load, overall merit provides migration decision-making;
3) system real time load is weighed;
Load balancing method and the loading index utility function of online service request change actual conditions can be reflected, in conjunction with actual operating data measuring and calculating and checking, determine rational parameter;
Server is mainly used for receiving the request of user, and therefore, the quantity of user's request has reacted the performance of this server to a certain extent.The node of different configuration is different to the disposal ability of request, but the performance of different servers to request process can be obtained by test experiments;
Process the performance of user's request when carrying out response service device peak load with rpmC value, rpm (Requests Per Minutes), refers to the new number of request of interior system process per minute; C refers to the atomic operation in rpm, and namely dissimilar request can represent with the form of kilounit (kC).The computational process of rpmC value is as follows:
Suppose server i there be j service, following symbol is expressed as:
R
ij: the jth on server i serves largest request number during peak value;
T
ij: the response time threshold value of a jth service on server i;
Regulation, during server peak load, cpu busy percentage is 75%, then:
Based on Markov Chain method, system synthesis predicts that the tenant that serves drawn is applied in the business fluctuation situation of future time instance and judges whether this node transships by the rpmC value of monitoring each node, determine whether that Trigger services moves;
4) serviced component migration decision-making is determined;
An important function of services migrating is the business load of balancing service, to reduce service response time, improve quality and the availability of service, therefore need to formulate rational migration strategy to ensure to carry out services migrating in time when the business load of service carrier is too high; When selecting the target Home Location needing example and the migration of moving, according to the migration decision-making drawn, complex transfer efficiency and moving costs calculate and provide migratory route;
Migration strategy mainly comprises the content of two aspects: be first weigh based on traffic forecast and real time load, determine to move decision-making; Then be select migration example and target Home Location based on transport efficiency and moving costs; When selecting the target Home Location needing example and the migration of moving, according to the migration demand drawn, complex transfer efficiency and moving costs calculate and provide migratory route;
The first step, selects migration service example;
Second step, selects destination node, weighs formula obtain according to load.
2. fused business as claimed in claim 1 is predicted and the two benches load estimation of real time load and equalization methods, it is characterized in that, described step 2) in,
In the service system of many tenants, the excess load of tenant's business is the one of the main reasons that services migrating demand occurs, and in the Service Operation of reality, the change of tenant's business has certain rule, it be the cyclic variation of the market demand, enterprise operation strategy transition factor combined influence under result, therefore, serve tenant's applied business load predicting method, from macroscopic view and trend, accurately can grasp the load variations trend of service, the basic step of prediction is as follows:
The first step, determine the major parameter weighing business load, and this business is carried out to the state demarcation of science, at least mark off two states, these states are exactly the content that will predict;
Second step, carries out statistical survey to the current state probability of the various states of service parameter change, the current residing state of the business that namely determines;
3rd step, measures the transition probability of each state future development of service parameter, if be steady development transfer within the following long period, then each transfer of system mode can keep identical transition probability; If be the concussion that rises and falls within the following long period, then state often shifts once just needs to measure once transition probability, and the time interval that state shifts at every turn then needs to determine according to concrete business;
4th step, each state probability current according to the service parameter selected and state transition probability use the method for matrix, deduce out following after several times transfer, remain on the probable value of a certain state, thus judge.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210148122.6A CN102664812B (en) | 2012-05-14 | 2012-05-14 | Two-stage service system load forecast and balancing method integrating service forecast and real-time load |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210148122.6A CN102664812B (en) | 2012-05-14 | 2012-05-14 | Two-stage service system load forecast and balancing method integrating service forecast and real-time load |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102664812A CN102664812A (en) | 2012-09-12 |
| CN102664812B true CN102664812B (en) | 2015-05-20 |
Family
ID=46774230
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210148122.6A Expired - Fee Related CN102664812B (en) | 2012-05-14 | 2012-05-14 | Two-stage service system load forecast and balancing method integrating service forecast and real-time load |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102664812B (en) |
Families Citing this family (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2512847A (en) * | 2013-04-09 | 2014-10-15 | Ibm | IT infrastructure prediction based on epidemiologic algorithm |
| CN103889001B (en) * | 2014-03-13 | 2018-04-20 | 南京邮电大学 | A kind of adaptive load balancing method based on future load prediction |
| GB2544049A (en) * | 2015-11-03 | 2017-05-10 | Barco Nv | Method and system for optimized routing of data streams in telecommunication networks |
| EP3560146B1 (en) | 2016-12-26 | 2022-03-02 | Morgan Stanley Services Group Inc. | Predictive asset optimization for computer resources |
| CN106603695B (en) * | 2016-12-28 | 2020-10-02 | 北京奇艺世纪科技有限公司 | Method and device for adjusting query rate per second |
| CN107798104A (en) * | 2017-10-31 | 2018-03-13 | 郑州云海信息技术有限公司 | A kind of catalog management method, device, equipment and computer-readable recording medium |
| CN108134821B (en) * | 2017-12-14 | 2020-09-08 | 南京邮电大学 | Multi-domain resource perception migration method based on cooperation of pre-calculation and real-time calculation |
| CN108566424B (en) * | 2018-04-11 | 2021-04-20 | 深圳市腾讯网络信息技术有限公司 | Scheduling method, device and system based on server resource consumption prediction |
| CN108710553B (en) * | 2018-05-08 | 2021-02-26 | 国家计算机网络与信息安全管理中心 | System and method for detecting reliability of application server |
| CN110213351A (en) * | 2019-05-17 | 2019-09-06 | 北京航空航天大学 | A kind of dynamic self-adapting I/O load equalization methods towards wide area high-performance computing environment |
| CN110311987A (en) * | 2019-07-24 | 2019-10-08 | 中南民族大学 | Node scheduling method, apparatus, equipment and the storage medium of microserver |
| CN111930526B (en) * | 2020-10-19 | 2021-01-22 | 腾讯科技（深圳）有限公司 | Load prediction method, load prediction device, computer equipment and storage medium |
| CN112866131B (en) * | 2020-12-30 | 2023-04-28 | 神州绿盟成都科技有限公司 | Traffic load balancing method, device, equipment and medium |
| CN113535530B (en) * | 2021-08-13 | 2024-03-01 | 广州虎牙科技有限公司 | Method and device for predicting server load, electronic equipment and storage medium |
| CN113904974B (en) * | 2021-10-09 | 2023-08-15 | 咪咕文化科技有限公司 | Intelligent routing method, device and equipment |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101593133A (en) * | 2009-06-29 | 2009-12-02 | 北京航空航天大学 | Load balancing of resources of virtual machine method and device |
-
2012
- 2012-05-14 CN CN201210148122.6A patent/CN102664812B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101593133A (en) * | 2009-06-29 | 2009-12-02 | 北京航空航天大学 | Load balancing of resources of virtual machine method and device |
Non-Patent Citations (2)
| Title |
|---|
| "A Dependency-aware Hierarchical Service Model for SaaS and Cloud Services";Rui Wang等;《2011 IEEE International Conference on Services Computing》;20110709;480-487 * |
| "LBVS:A Load Balancing Strategy for Virtual Storage";Hao Liu等;《2010 International Conference on Service Sciences》;20100514;257-262 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102664812A (en) | 2012-09-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102664812B (en) | Two-stage service system load forecast and balancing method integrating service forecast and real-time load | |
| US8756441B1 (en) | Data center energy manager for monitoring power usage in a data storage environment having a power monitor and a monitor module for correlating associative information associated with power consumption | |
| CN102111337B (en) | Method and system for task scheduling | |
| US9207993B2 (en) | Dynamic application placement based on cost and availability of energy in datacenters | |
| CN103152393B (en) | A kind of charging method of cloud computing and charge system | |
| CN103067297B (en) | A kind of dynamic load balancing method based on resource consumption prediction and device | |
| CN103294546B (en) | The online moving method of virtual machine of multi-dimensional resource performance interference aware and system | |
| CN106156115B (en) | A kind of resource regulating method and device | |
| CN114930293A (en) | Predictive auto-expansion and resource optimization | |
| EP2913756B1 (en) | Operation management apparatus and operation management method | |
| CN1936849A (en) | Resource dynamic regulation method and apparatus | |
| CN103995735A (en) | Device and method for scheduling working flow | |
| CN102012891B (en) | Computer cluster management method, device and system | |
| CN103605567A (en) | Cloud computing task scheduling method facing real-time demand change | |
| AU2017237704A1 (en) | Control device for estimation of power consumption and energy efficiency of application containers | |
| CN104536804A (en) | Virtual resource dispatching system for related task requests and dispatching and distributing method for related task requests | |
| CN107707612B (en) | Method and device for evaluating resource utilization rate of load balancing cluster | |
| CN115269108A (en) | Data processing method, device and equipment | |
| CN104346214A (en) | Device and method for managing asynchronous tasks in distributed environments | |
| Rawas et al. | Power and Cost-aware Virtual Machine Placement in Geo-distributed Data Centers. | |
| CN103634167A (en) | Security configuration check method and system for target hosts in cloud environment | |
| CN116862549A (en) | Service cost determining method and device, electronic equipment and nonvolatile storage medium | |
| US8352092B2 (en) | Method and system for workload balancing to assist in power grid load management | |
| CN106997310A (en) | The apparatus and method of load balancing | |
| US9027035B2 (en) | Non real-time metrology data management |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20150520 |