CN108234356A - Optimization application resource Distribution Strategy based on application relational network - Google Patents
Optimization application resource Distribution Strategy based on application relational network Download PDFInfo
- Publication number
- CN108234356A CN108234356A CN201711231345.8A CN201711231345A CN108234356A CN 108234356 A CN108234356 A CN 108234356A CN 201711231345 A CN201711231345 A CN 201711231345A CN 108234356 A CN108234356 A CN 108234356A
- Authority
- CN
- China
- Prior art keywords
- application
- resource
- service
- node
- distribution strategy
- 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
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/70—Admission control; Resource allocation
- H04L47/76—Admission control; Resource allocation using dynamic resource allocation, e.g. in-call renegotiation requested by the user or requested by the network in response to changing network conditions
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/08—Configuration management of networks or network elements
- H04L41/0803—Configuration setting
- H04L41/0823—Configuration setting characterised by the purposes of a change of settings, e.g. optimising configuration for enhancing reliability
-
- 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
- H04L67/1034—Reaction to server failures by a load balancer
Abstract
The present invention provides a kind of optimization application resource Distribution Strategies based on application relational network, include the following steps:Based on application service dependence calculate using importance index;According to using importance index, the application resource Distribution Strategy under the influence of computational minimization resource O&M;According to the resource distribution situation of application resource Distribution Strategy dynamic adjustment application.The present invention is based on the importance of application, study in the case where resource is limited, to each most rational resource provisioning mode of application, so as to reach resource in O&M to the influence on system operation of entire operation system minimum.The present invention broken away from existing blindness according to time sequencing and manual operation come the mode for allocated resources, and the different importance of calculating of target automation optimized based on O&M change using required resource and dynamically according to the importance of application adjust the resource provisioning of application.
Description
Technical field
The present invention relates to a kind of resource allocation policies of field of cloud computer technology, and in particular, to one kind is based on application and closes
It is the optimization application resource Distribution Strategy of network.
Background technology
In current Web and mobile application development process, developer tends to go structure application program based on service, and
It is not from being made wheel.Under normal circumstances, these services are referred to as micro services --- single-use, and the addressable applications of API become
The foundation stone of the large-scale application of structure." micro services " framework is the very popular concept in recent software application field, can be significantly
Improve some typical problems that traditional application and development encounters.For example, use traditional monoblock type framework (Monolithic
Architecture) application development system, such as the large-scale application of CRM, ERP, with being continuously increased for new demand, enterprise's update and
Repairing large-scale monoblock type application becomes more and more difficult.With the development of mobile Internet, enterprise be forced by its application migrate to
Modernize UI interface architectures so as to can compliant mobile devices, this requires enterprise that can realize the Quick thread of application function.
Based on this demand, the forming types of the complication system of more and more enterprises and industry are gradually from traditional monomer
Start to change using to micro services framework.This direct result brought that changes is exactly that profession will be from by several relatively independent
Major comonomer be made of a large amount of mutual related micro- application using forming to be changed into.This pattern is by strong improvement row
The update development pattern of industry operation system pushes profession to increase using in a manner of internet with the explosion type that iteration rolls
It is long.This also will bring endless vigor for profession.A series of relevant skills of DevOps of hot spot are being gradually become now
Art and based on cloud platform using the abilities such as automatically dispose and publication significantly develop all by strong promotion this into
Journey.
But this profession system also brings new choose by the mode that a large amount of micro- applications are formed for the O&M of system
War.The increase of application brings the complexity of O&M.Especially in the cloud service epoch, these applications are usually all deployed in cloud platform
It is supplied with the optimization of the resource of realization.But how it is these applications rationally effective distribution resource that present main problem if being.
During O&M, if we need to carry out some resource attended operation or carry out troubleshooting to some resource, such as
What ensures that influence of this operation to entire operation system is minimum, so as to reduce O&M cost, realizes the application resource of optimization
Distribution, becomes this field urgent problem to be solved.
It is found by retrieval:
Application No. is:201410189197.8 Chinese patent application《The assessment of resource allocation policy in a kind of cloud computing
Method》, the appraisal procedure of resource allocation policy in cloud computing a kind of is disclosed, including:The item that will be deployed on cloud computing platform
Mesh is divided into multiple function modules, is modeled to generate Work flow model according to the relationship between function module;According to resource
Allocation strategy, which distributes cloud computing resources, to each function module, determines that the operating parameter of each function module realizes configuration work
Flow model;Work flow model is mapped as Timed Automata model;The run time error and run time of function module are missed
Difference cloth is mapped as the time error of Timed Automata model Neutron module;Specification test standard calculates Timed Automata model
Meet assessment result of the probability value as resource allocation policy of test stone.This method can automatically analyze current distribution
Whether the lower user demand of strategy can be satisfied, and quantitatively analyze the reliability of current allocation strategy, and clothes are violated so as to reduce
The probability of business layer protocol.
There are still following problems for the above method:
This method is the formation of project-based function module stream, this Object--oriented method there are function module not
Weight setting is carried out, is all the distribution resource of equality, and is commented according to the reasonability of the failure rate of each module progress resource allocation
The problem of estimating, can not verifying the optimum distributing scheme of function module and resource;
This method is the modeling carried out according to the mode of workflow, and this modeling method exists can not be according to the upgrading of application
The distribution of the dynamic adjustment application resource of change, reaches the problem of O&M resource minimizes cost.
Currently without the explanation or report for finding technology similar to the present invention, it is also not yet collected into money similar both at home and abroad
Material.
Invention content
For above-mentioned deficiency in the prior art, the object of the present invention is to provide a kind of based on using relational network
Optimize application resource Distribution Strategy.The strategy that the strategy is formed based on all micro services relationships of application, and be based on answering
One formed with micro services relational network and application resource relational network using importance index based on defining deployment strategy.
The present invention is achieved by the following technical solutions.
A kind of optimization application resource Distribution Strategy based on application relational network, includes the following steps:
Step S1 based on application service dependence calculate using importance index;
Step S2, according to using importance index, the application resource Distribution Strategy under the influence of computational minimization resource O&M;
Step S3, according to the resource distribution situation of application resource Distribution Strategy dynamic adjustment application.
Preferably, step S1, including following sub-step:
Step S1.1, using the analysis of weight:By application and the service call between application, the power of related application is obtained
Weight;
Step S1.2, using the evolution of importance:After the deployment of each new opplication or after old application is offline periodically
Recalculate the importance index of all applications in system.
Preferably, step S1.1 uses the application network node weights calculation of multiple links, including following process:
1 is defined, application network digraph is G, is shown below:
G(E,V)
In formula, E represents node relationships set, and V represents node set;
2, effectively service reference set Ef (u) are defined, are shown below:
Ef (u)=v | v ∈ Follower (u) ∩ Response (u) > ε }
In formula, ε is non-negative constant threshold, represents the degree thresholding that the reference service node V of node u feeds back node u, is surpassed
The application node crossed the threshold value and belong to node u is effectively application;
3 are defined, the node weights IRL (U as caused by linking relationshipi), computational methods are shown below:
IRL(Ui)=δ N+ (1- δ) ∑ Uj∈Follower(ui)IRL(ui)L(ui)
In formula, IRL (Ui) represent node UiThe node weights that linking relationship generates, Follower (ui) it is node UiIt is all
Association service set, L (ui) it is node UiAssociation service number, δ are the damped coefficients between 0 and 1, and N is total section in network
Points.
Preferably, step S2, specially:
For specified resource R, the application that R will be directly affected in O&M is analyzed, if these applications are DE (R);Root
According to application service dependence graph, all applications being affected are found, the range of final all applications being affected is entirely should
With a subgraph in service dependence graph, the weights of all applications are resources defined as the impact factor F of R in this subgraph
(R);
Assuming that can be R1 to Rn by the total resources of O&M in system, then O&M on entire profession it is total influence because
Son is exactlyThe application resource Distribution Strategy of optimization should causeReach minimum;Thus, it would be desirable to
A kind of computational methods A of innovation is worked out in this subject to causeReach the application resource Distribution Strategy of minimum;
The computational methods A of the innovation is specially:
Step SA.1 by the good and bad situation of resource, is classified such as situations such as memory, CPU, hard disk and network interface card into row label;
Step SA.2 is in the form of a label allocated resource by the importance of application and the good and bad situation of resource;
Step SA.3, by elastic telescopic and extending transversely and carry out the total shadow of modes such as resource allocation by labelling
Ring the factorReach minimum.
Preferably, step S3, specially:
After the variation using importance occurs every time, system needs to calculate the optimization distributed model of application resource, so
Afterwards according to the conclusion for optimizing distributed model, adjust automatically corresponds to the resource provisioning situation of application;Wherein, distributed model is optimized
It is obtained, included the following steps using computational methods B:
Step SB.1 by application service dependence graph, calculate again using weights of importance;
Step SB.2, according to computational methods A so that the total resources O&M service impact factor minimizes;
Step SB.3, the optimization distributed model that application resource is carried out by elastic telescopic and/or trunking mode calculate.
Preferably, in step S1 to S3, according to cloud product provide infrastructure, using automatically dispose and service from
Dynamicization delivers structure system operation and maintenance system, and the system operation and maintenance system includes application service dependence management module, using money
Source control module, application resource Distribution Strategy management module and application resource dynamic dispatching module;Wherein:
The application service dependence management module for generate application service rely on network and application importance and
The calculating of weight;
The application resource management module relies on network for generating application resource;
The application resource Distribution Strategy management module is used to calculate and generate the Optimal Distribution strategy of application resource;
The application resource dynamic dispatching module is used to be rescheduled simultaneously according to the Optimal Distribution strategy of application resource
Carry out the distribution of application resource.
Preferably, following steps are further included:
Step S4, including following any one or any number of processes:
Application is upgraded;
Application is migrated and disposed, is calculated by the weight to application with dependence, important application is disperseed
Into different physical nodes, the high availability of application is realized;
It is automated using O&M, including following any one or arbitrary multinomial:
Failure self-isolation when application in operation is broken down, is isolated, and notify phase in time the failure application
It closes service stopping or carries out failure transfer;
High Availabitity self-balancing, when important application is deployed in different physical nodes, when event occurs for physical node
The High Availabitity backup of application is switched on available physical node by barrier, applicative notifications cloud platform, realizes High Availabitity self-balancing.
Preferably, it carries out upgrading to application to carry out using alternative, i.e., disposes different Service Instances in advance, then pass through
Upgraded using the mode that relational dependence is replaced.
Preferably, before upgrading to application, after the first functionality of progress as the case may be and non-functional test
It replaces again, retracts after upgrading failure and carry out rapid deterioration.
Preferably, the mode using relational dependence replacement is specially:Using the access side between associated service
The replacement of formula, i.e., by way of service registration and service discovery, the access address of old service that will be accessed is switched to newly
Service Instance on.
Preferably, the weight of described pair of application is calculated with dependence, is specifically comprised the following steps:
Step a, according to the application service incidence relation that cloud product provides, construction profession system application service relies on net
Network;
Step b relies on network for simple profession system application service, can by the call number of application into
Row weight calculation relies on network for complicated profession system application service, can add in and apply in operation system
Importance carries out weight analysis;
Step c carries out application resource by the distributed model that optimizes that the computational methods for optimizing distributed model obtain
Predistribution calculates.
Preferably, during application is migrated and disposed, setting is using weight self-balancing mechanism, mechanism tool
Body is:Whenever the application of operation system has situation that is newly-increased, deleting modification, system should be able to be according to application service dependence net
Network carries out the Regulation mechanism of weight.
Compared with prior art, the present invention has following advantageous effect:
1st, the present invention is based on the importance of application, study in the case where resource is limited, the most rational money to each application
Source method of supplying, so as to reach resource in O&M to the influence on system operation of entire operation system minimum.
2nd, the present invention broken away from existing blindness according to time sequencing and manual operation come the side for allocated resources
Formula, and the required resource of application for the different importance of calculating that the target optimized based on O&M is automated and dynamic root
Change the resource provisioning to adjust application according to the importance of application.
Description of the drawings
Upon reading the detailed description of non-limiting embodiments with reference to the following drawings, other feature of the invention,
Objects and advantages will become more apparent upon:
Fig. 1 relies on network diagram for application service;
Fig. 2 relies on network diagram for application resource;
Fig. 3 is system operation and maintenance system Organization Chart;
Fig. 4 is using weight distribution physical node figure;
Fig. 5 is Operation and Maintenance Center system architecture diagram.
Specific embodiment
It elaborates below to the embodiment of the present invention:The present embodiment is carried out lower based on the technical solution of the present invention
Implement, give detailed embodiment and specific operating process.It should be pointed out that those of ordinary skill in the art
For, without departing from the inventive concept of the premise, various modifications and improvements can be made, these belong to the guarantor of the present invention
Protect range.
Embodiment
A kind of optimization application resource Distribution Strategy based on application relational network is present embodiments provided, including walking as follows
Suddenly:
Step S1, based on application service dependence calculate using importance index
According to existing application service dependence graph, each application can schemed by it in a profession system
In the number that is called simply indicate using importance index, it is secondary similar to being cited in research work by paper
Count the action value to indicate paper.
Further, it may be considered that the calling of the importance for having different levels in business or service is applied to exist not
With the importance of degree, for this purpose, can be added in when calculating and applying the importance index in figure using weight analysis.
1) using weight analysis
Using the service call between application, there are different calling degree, therefore applications different in systems
Significance level is not consistent, according to application and the call relation between application, it can be deduced that the weight of related application.
Reference count mode may be used using the calculation of weight or more independent path modes are calculated.This implementation
In example, the application network node weights calculation that calculation can refer to multiple links carries out.
The basic principle of this method is as follows.
1 application network digraph G is defined, is shown below:
G(E,V)
In formula, E represents node relationships set, and V represents node set.
2 effectively service reference set Ef (u) are defined, are shown below:
Ef (u)=v | v ∈ Follower (u) ∩ Response (u) > ε }
In formula, ε is non-negative constant threshold, represents the degree thresholding that the reference service node v of node u feeds back node u, is surpassed
The application node crossed the threshold value and belong to node u could be applied effectively at last.
The node weights IRL (U as caused by linking relationship of definition 3i), computational methods are shown below:
IRL(Ui)=δ N+ (1- δ) ∑ Uj∈Follower(ui)IRL(ui)L(ui)
In formula, IRL (Ui) represent node UiThe node weights that linking relationship generates, Follower (ui) it is node uiIt is all
Association service set, L (ui) it is node uiAssociation service number, δ are the damped coefficients between 0 and 1, and N is total section in network
Points.
2) using the evolution of importance
The importance of application is not unalterable in industry operation system.With the development of profession, increasingly
More applications are developed.In this process, each application may there are one from be widely used gradually being eliminated
Or newer life cycle.Therefore the citation times of each existing application or even weights are likely to be disposed in new opplication
It is changed after a period of use.For simplified model, it can be initially believed that and each apply at the very start can in deployment
Specify its business importance.It so just needs to recalculate in system after the deployment of each new opplication or after old application is offline
The importance index of all applications.
Further, the business importance of application is to be determined in use for some time according to the access situation of user.
In this case, the importance index needs of all applications are periodically recalculated.
Step S2, according to using importance index, the application resource Distribution Strategy under the influence of computational minimization resource O&M
For O&M angle, first simply O&M can be interpreted as to the upgrade maintenance of resource or troubleshooting work
Make.In this case, a resource will be stopped, it is possible to cause with the work of the relevant application of this resource by
To influence.Network is relied on according to the application resource that aforementioned operation is constructed, for specified resource R, R can be analyzed in O&M
The application that will be directly affected, if these applications are DE (R).Again from application service dependence graph, can find all these
The application being affected, final all application ranges for being affected are a subgraphs in entire application service dependence graph.This
The impact factor F (R) that the weights of all applications will be resources defined as R in a subgraph.
Assuming that can be R1 to Rn by the total resources of O&M in system, then O&M on entire profession it is total influence because
Son is exactly.The application resource Distribution Strategy of optimization should causeReach minimum.Thus, it would be desirable to
A kind of computational methods A is worked out in this subject to causeReach the application resource Distribution Strategy of minimum.
The computational methods A of the innovation, includes the following steps:
1) it by the good and bad situation of resource, is classified such as situations such as memory, CPU, hard disk and network interface card into row label;
2) it is allocated in the form of a label by the good and bad situation of the importance of application and resource;
3) by elastic telescopic and extending transversely and carry out the total impact factor of modes such as resource allocation by labellingReach minimum.
Step S3, according to the resource distribution situation of application resource Distribution Strategy dynamic adjustment application
On the basis of step S1 and step S2 is calculated, current embodiment require that realizing the dynamic of application resource in cloud platform
Adjustment.After the variation using importance occurs every time, system needs to calculate the optimization distributed model of application resource, then
According to the resource provisioning situation of the conclusion of model, the automatically corresponding application of adjustment, so that real system reaches optimum efficiency.
The optimization distributed model of application resource is calculated, specific method includes the following steps:
1) by application service dependence graph, calculate again using weights of importance;
2) according to computational methods A the total resources O&M service impact factor is minimized;
3) the optimization distributed model that application resource is carried out by modes such as elastic telescopic, clusters calculates.
For this purpose, system needs the framework in realization system O&M Organization Chart, the basic thought of the framework is produced in existing cloud
The infrastructure of product offer builds system operation and maintenance system on the basis of being delivered using automatically dispose and automatization of service.System is transported
Dimension system is by the management of application service dependence, application resource management, the management of application resource Distribution Strategy and application resource dynamic
Four module compositions of scheduling.
In this step, system operation and maintenance system is act as:Application service dependence network is generated by calculating and is answered
With Resource Dependence network, and pass through the Optimal Distribution strategy for calculating generation application resource, and then when system O&M upgrades,
Carry out the self-balancing of application resource, readjustment degree.
Further, the application service dependence management module relies on network, Yi Jiying for generating application service
With the calculating of importance and weight;
The application resource management module relies on network for generating application resource;
The application resource Distribution Strategy management module is used to calculate and generate the Optimal Distribution strategy of application resource;
The application resource dynamic dispatching module is used to be rescheduled simultaneously according to the Optimal Distribution strategy of application resource
Carry out the distribution of application resource.
Step S4, O&M extension variation
With the research of the present embodiment core innovative point, can build using the Proper Match mode with resource.In this base
On plinth, it is further contemplated that more complicated O&M behavior, such as application upgrading update, using High Availabitity, using O&M from
Dynamicization etc..
1) application upgrade is with replacing
Application upgrade flow and corresponding updating operation and traditional main body of O&M flow change under micro services framework
Present the following aspects:
Influence of the upgrading of i applications for total system can be controlled, be isolated in a relatively controllable section, single
The upgrading of application does not influence the use of whole system;
Ii application upgrades are more upgraded using " replacement " mode, can first dispose different Service Instances, then into
Row is upgraded using the mode that relational dependence is replaced;It is described using relational dependence replace mode be specially:Using associated
Service between access mode replace, the access of old service that will be accessed by way of service registration and service discovery
Address is switched on new Service Instance.
It can carry out replacing again after functional and non-functional test as the case may be before iii application upgrades, rise
Retracting after grade failure can be with rapid deterioration, and the influence after failing for upgrading is controllable.
2) using High Availabitity
In conventional architectures, using being deployed in virtual machine or physical machine, it is difficult to realize the scheduling of underlying resource.Each
It is limited using occupied underlying resource, realizes the High Availabitity of application, must be requested that High Availabitity is realized between resource.Newly
Framework under, resource used in different application is adjustable, therefore can realize the migration of application and corresponding deployment.
It is calculated, important application can be distributed in different physical nodes with relying on by the weight of application, realized
The reliable High Availabitity of application, with reference to figure 4.The weight of the application is calculated with relying on, and computational methods are specially:
1) the application service incidence relation provided according to product, the application service of construction profession system rely on network;
2) simple profession system application service, which relies on network, to carry out weight meter by the call number of application
It calculates,
Complicated profession system application service dependence network, which can add in, applies the importance in operation system
Carry out weight analysis;
3) dependence of application resource calculates the optimization for being to obtain by optimizing the computational methods of distributed model and is distributed
The predistribution that model carries out application resource calculates.
As shown in Fig. 5 Organization Charts:Different application is all deployed in cloud platform, and external physical resource carries out pipe by cloud platform
Reason, the resource including calculating, network, storage are allocated application by cloud platform.
In large complicated environment, often there is a large amount of application to increase/delete operation newly, different applications, which enters or leaves, is
System, it will the relationship of each application and service in systems is caused to change, therefore system is needed to have using weight certainly
Balancing gradually adjusts the weight and topological relation of each application and service.The application weight self-balancing mechanism, specially
Whenever the application of operation system have it is newly-increased, delete modification situation, system should be able to according to application service dependence network into
The Regulation mechanism of row weight.
3) it is automated using O&M
Number of applications under micro services framework increases, and upgrading and the maintenance of application are required to rely on the ability that cloud platform provides
Carry out automation O&M.This automation O&M can be presented as following two abilities:
First, failure self-isolation:
Due to there is a large amount of application in environment, once certain running applications are broken down, need to carry out application
Isolation notifies related service to stop or carry out failure transfer in time.
2nd, High Availabitity self-balancing
The High Availabitity of application needs to be deployed in important application in different physical nodes, once event occurs for physical node
Barrier, using needing notice cloud platform that the High Availabitity backup of application is switched on enabled node, to reach High Availabitity self-balancing
Ability.
The present embodiment proposes the application resource allocation strategy scheme of the optimization of an adaptation resource O&M.Its principle
For:
If the resource of whole system is enough, then each application should be deployed in incoherent multiple resources, so that
When any one resource is in O&M state (upgrading update or Breakdown Maintenance), the operation of entire profession system is completely not
It is impacted.Such case is known as ideal application resource allocation strategy.But the system in reality is often reached in inadequate resource
Situation.In this case, the application resource allocation strategy that can only set an optimization is important using should portion as possible
Administration is in incoherent multiple resources, so that when any cost is in O&M state, entire profession system is by as possible
Small influence.
The importance applied from the above is the key that deduce this optimisation strategy.Body is applied in the micro- of scale of industry
In system, can set using major part is mutually related by service.Profession system is one by application service tune
The application service network constructed with relationship.Without loss of generality, in this network, the service of an application should by other
It is more with calling by reference, it is believed that this application is more important.Setting based on front, it is believed that such application should be by
Distribution is deployed in multiple incoherent resources to ensure the reliability of system and O&M.
According to this setting, the present embodiment is broadly divided into the content of three links:
1st, the importance of application how is identified;
2nd, how according to the importance of application resource distribution strategy is defined with being optimal
3rd, the landing of this optimization strategy how is realized by the automatically dispose of application.
Meanwhile with the development of profession system, there may be continually changing mistakes for the importance of each micro- application
Journey.Therefore, the work of three above link may need re-executing so that entire profession system is in always repeatedly
The state of optimization.
The innovative point of the present embodiment mainly includes the content of following three aspects:
1st, it is calculated using importance index;
2nd, according to using importance index, the application resource Distribution Strategy under the influence of computational minimization resource O&M;
3rd, according to the resource distribution situation of application resource Distribution Strategy dynamic adjustment application.
Specific embodiments of the present invention are described above.It is to be appreciated that the invention is not limited in above-mentioned
Particular implementation, those skilled in the art can make various deformations or amendments within the scope of the claims, this not shadow
Ring the substantive content of the present invention.
Claims (10)
1. a kind of optimization application resource Distribution Strategy based on application relational network, which is characterized in that include the following steps:
Step S1 based on application service dependence calculate using importance index;
Step S2, according to using importance index, the application resource Distribution Strategy under the influence of computational minimization resource O&M;
Step S3, according to the resource distribution situation of application resource Distribution Strategy dynamic adjustment application.
2. the optimization application resource Distribution Strategy according to claim 1 based on application relational network, which is characterized in that
Step S1, including following sub-step:
Step S1.1, using the analysis of weight:By application and the service call between application, the weight of related application is obtained;
Step S1.2, using the evolution of importance:After the deployment of each new opplication or after old application is offline periodically again
The importance index of all applications in computing system.
3. the optimization application resource Distribution Strategy according to claim 2 based on application relational network, which is characterized in that
Step S1.1 uses the application network node weights calculation of multiple links, including following process:
1 is defined, application network digraph is G, is shown below:
G=(E, V)
In formula, E represents node relationships set, and V represents node set;
2, effectively service reference set Ef (u) are defined, are shown below:
Ef (u)=v | v ∈ Follower (u) ∩ Response (u) > ε }
In formula, ε is non-negative constant threshold, the degree thresholding that the reference service node V of node u feeds back node u is represented, more than this
Threshold value and belong to the application node of node u for effectively application;
3 are defined, the node weights IRL (U as caused by linking relationshipi), computational methods are shown below:
IRL(Ui)=δ N+ (1- δ) ∑ Uj∈Follower(ui)IRL(ui)L(ui)
In formula, IRL (Ui) represent node UiThe node weights that linking relationship generates, Follower (ui) it is node UiThe relevant clothes of institute
Business set, L (ui) it is node UiAssociation service number, δ are the damped coefficients between 0 and 1, and N is the total node number in network.
4. the optimization application resource Distribution Strategy according to claim 1 based on application relational network, which is characterized in that
Step S2, specially:
For specified resource R, the application that R will be directly affected in O&M is analyzed, if these applications are DE (R);According to should
With service dependence graph, all applications being affected are found, the range of final all applications being affected is entire application clothes
A subgraph being engaged in dependence graph, the weights of all applications are resources defined as the impact factor F (R) of R in this subgraph;
Assuming that can be R1 to Rn by the total resources of O&M in system, then O&M to total impact factor of entire profession just
It isThe application resource Distribution Strategy of optimization should causeReach minimum;For this purpose, using the side of being calculated as below
Method A causesReach the application resource Distribution Strategy of minimum:
Step SA.1 is classified by the good and bad situation of resource into row label;
Step SA.2 is in the form of a label allocated resource by the importance of application and the good and bad situation of resource;
Step SA.3, by elastic telescopic, it is extending transversely and/or in a manner that label carries out resource allocation it is total influence because
SonReach minimum.
5. the optimization application resource Distribution Strategy according to claim 4 based on application relational network, which is characterized in that
Step S3, specially:
After the variation using importance occurs every time, system needs to calculate the optimization distributed model of application resource, Ran Hougen
According to the conclusion for optimizing distributed model, adjust automatically corresponds to the resource provisioning situation of application;Wherein, distributed model is optimized to use
Computational methods B is obtained, and is included the following steps:
Step SB.1 by application service dependence graph, calculate again using weights of importance;
Step SB.2, according to computational methods A so that the total resources O&M service impact factor minimizes;
Step SB.3, the optimization distributed model that application resource is carried out by way of elastic telescopic and/or cluster calculate.
6. the optimization application resource Distribution Strategy according to claim 5 based on application relational network, which is characterized in that
In step S1 to S3, structure system is delivered according to the infrastructure of cloud product offer, using automatically dispose and automatization of service
System operation and maintenance system, the system operation and maintenance system include application service dependence management module, application resource management module, application
Resource distribution policy management module and application resource dynamic dispatching module;Wherein:
The application service dependence management module relies on network and using importance and weight for generating application service
Calculating;
The application resource management module relies on network for generating application resource;
The application resource Distribution Strategy management module is used to calculate and generate the Optimal Distribution strategy of application resource;
The application resource dynamic dispatching module is used to be rescheduled and be carried out according to the Optimal Distribution strategy of application resource
The distribution of application resource.
7. the optimization application resource Distribution Strategy according to any one of claim 1 to 6 based on application relational network,
It is characterized in that, further include following steps:
Step S4, including following any one or any number of processes:
Application is upgraded;
Application is migrated and disposed, is calculated by the weight to application with dependence, important application is distributed to not
In same physical node, the high availability of application is realized;
It is automated using O&M, including following any one or arbitrary multinomial:
Failure self-isolation when application in operation is broken down, is isolated the failure application, and the related clothes of notice in time
Business stops or carries out failure transfer;
High Availabitity self-balancing,, should when physical node breaks down when important application is deployed in different physical nodes
The High Availabitity backup of application is switched on available physical node with notice cloud platform, realizes High Availabitity self-balancing.
8. the optimization application resource Distribution Strategy according to claim 7 based on application relational network, which is characterized in that
Application is upgraded, further includes following any one or arbitrary multinomial feature:
Carry out upgrading to application to carry out using alternative, i.e., dispose different Service Instances in advance, then by application relationship according to
The mode replaced is relied to be upgraded;
Before upgrading to application, first carry out replacing again after functional and non-functional test as the case may be, upgrade
It retracts after failure and carries out rapid deterioration;
The mode replaced using relational dependence is specially using the replacement of the access mode between associated service, that is, is led to
The mode of service registration and service discovery is crossed, the access address of old service that will be accessed is switched on new Service Instance.
9. the optimization application resource Distribution Strategy according to claim 7 based on application relational network, which is characterized in that
The weight of described pair of application is calculated with dependence, is specifically comprised the following steps:
Step a, according to the application service incidence relation that cloud product provides, construction profession system application service relies on network;
Step b, by the call number of application carry out weight calculation or by add in apply the importance in operation system come
Carry out weight analysis;
Step c, the predistribution calculating for optimizing distributed model and carrying out application resource obtained by computational methods B.
10. the optimization application resource Distribution Strategy according to claim 7 based on application relational network, feature exist
During application is migrated and disposed, setting is specially using weight self-balancing mechanism, the mechanism:Whenever industry
Business systematic difference has situation that is newly-increased or deleting modification, and system should can be weighed according to application service dependence network
The Regulation mechanism of weight.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711231345.8A CN108234356B (en) | 2017-11-29 | 2017-11-29 | Optimized application resource distribution strategy based on application relation network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711231345.8A CN108234356B (en) | 2017-11-29 | 2017-11-29 | Optimized application resource distribution strategy based on application relation network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108234356A true CN108234356A (en) | 2018-06-29 |
CN108234356B CN108234356B (en) | 2021-07-06 |
Family
ID=62652968
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711231345.8A Active CN108234356B (en) | 2017-11-29 | 2017-11-29 | Optimized application resource distribution strategy based on application relation network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108234356B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109586952A (en) * | 2018-11-07 | 2019-04-05 | 广州虎牙信息科技有限公司 | Method of server expansion, device |
CN110738431A (en) * | 2019-10-28 | 2020-01-31 | 北京明略软件系统有限公司 | Method and device for allocating monitoring resources |
CN110943867A (en) * | 2019-12-05 | 2020-03-31 | 上交所技术有限责任公司 | System and method for deducing application architecture information through network relationship |
CN111324471A (en) * | 2020-01-22 | 2020-06-23 | 远景智能国际私人投资有限公司 | Service adjusting method, device, equipment and storage medium |
CN111817870A (en) * | 2019-04-10 | 2020-10-23 | 瞻博网络公司 | Intent-based network-aware network device software upgrade scheduling |
CN115484159A (en) * | 2022-09-16 | 2022-12-16 | 建信金融科技有限责任公司 | Network demand and resource supply determining system and method |
WO2024032239A1 (en) * | 2022-08-12 | 2024-02-15 | 华为云计算技术有限公司 | Application scheduling method, cloud service platform, and related device |
CN117687891A (en) * | 2024-02-02 | 2024-03-12 | 深圳迅策科技股份有限公司 | Index calculation optimization system based on AI |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102214212A (en) * | 2011-05-20 | 2011-10-12 | 西北工业大学 | Method for ordering microblog network node weights based on multi-link |
CN102413186A (en) * | 2011-12-02 | 2012-04-11 | 北京星网锐捷网络技术有限公司 | Resource scheduling method and device based on private cloud computing, and cloud management server |
US20140201371A1 (en) * | 2013-01-15 | 2014-07-17 | Hewlett-Packard Development Company, L.P. | Balancing the allocation of virtual machines in cloud systems |
US20150055483A1 (en) * | 2013-08-21 | 2015-02-26 | Electronics And Telecommunications Research Institute | Method and apparatus for controlling resource of cloud virtual base station |
CN105183610A (en) * | 2015-09-22 | 2015-12-23 | 浪潮集团有限公司 | Cloud data center service monitoring system and method based on resource dependency relationship |
CN106020927A (en) * | 2016-05-05 | 2016-10-12 | 中国人民解放军国防科学技术大学 | Universal method for task scheduling and resource configuration in cloud computing system |
-
2017
- 2017-11-29 CN CN201711231345.8A patent/CN108234356B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102214212A (en) * | 2011-05-20 | 2011-10-12 | 西北工业大学 | Method for ordering microblog network node weights based on multi-link |
CN102413186A (en) * | 2011-12-02 | 2012-04-11 | 北京星网锐捷网络技术有限公司 | Resource scheduling method and device based on private cloud computing, and cloud management server |
US20140201371A1 (en) * | 2013-01-15 | 2014-07-17 | Hewlett-Packard Development Company, L.P. | Balancing the allocation of virtual machines in cloud systems |
US20150055483A1 (en) * | 2013-08-21 | 2015-02-26 | Electronics And Telecommunications Research Institute | Method and apparatus for controlling resource of cloud virtual base station |
CN105183610A (en) * | 2015-09-22 | 2015-12-23 | 浪潮集团有限公司 | Cloud data center service monitoring system and method based on resource dependency relationship |
CN106020927A (en) * | 2016-05-05 | 2016-10-12 | 中国人民解放军国防科学技术大学 | Universal method for task scheduling and resource configuration in cloud computing system |
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109586952A (en) * | 2018-11-07 | 2019-04-05 | 广州虎牙信息科技有限公司 | Method of server expansion, device |
CN109586952B (en) * | 2018-11-07 | 2022-04-12 | 广州虎牙信息科技有限公司 | Server capacity expansion method and device |
US11922162B2 (en) | 2019-04-10 | 2024-03-05 | Juniper Networks, Inc. | Intent-based, network-aware network device software-upgrade scheduling |
US11640291B2 (en) | 2019-04-10 | 2023-05-02 | Juniper Networks, Inc. | Intent-based, network-aware network device software-upgrade scheduling |
CN111817870A (en) * | 2019-04-10 | 2020-10-23 | 瞻博网络公司 | Intent-based network-aware network device software upgrade scheduling |
CN111817870B (en) * | 2019-04-10 | 2021-12-21 | 瞻博网络公司 | Method for managing a plurality of network devices, controller device and storage medium |
CN110738431A (en) * | 2019-10-28 | 2020-01-31 | 北京明略软件系统有限公司 | Method and device for allocating monitoring resources |
CN110738431B (en) * | 2019-10-28 | 2022-06-17 | 北京明略软件系统有限公司 | Method and device for allocating monitoring resources |
CN110943867B (en) * | 2019-12-05 | 2022-08-16 | 上交所技术有限责任公司 | System and method for deducing application architecture information through network relationship |
CN110943867A (en) * | 2019-12-05 | 2020-03-31 | 上交所技术有限责任公司 | System and method for deducing application architecture information through network relationship |
CN111324471A (en) * | 2020-01-22 | 2020-06-23 | 远景智能国际私人投资有限公司 | Service adjusting method, device, equipment and storage medium |
WO2024032239A1 (en) * | 2022-08-12 | 2024-02-15 | 华为云计算技术有限公司 | Application scheduling method, cloud service platform, and related device |
CN115484159A (en) * | 2022-09-16 | 2022-12-16 | 建信金融科技有限责任公司 | Network demand and resource supply determining system and method |
CN115484159B (en) * | 2022-09-16 | 2023-06-20 | 建信金融科技有限责任公司 | Network demand and resource supply determining system and method |
CN117687891A (en) * | 2024-02-02 | 2024-03-12 | 深圳迅策科技股份有限公司 | Index calculation optimization system based on AI |
CN117687891B (en) * | 2024-02-02 | 2024-04-30 | 深圳迅策科技股份有限公司 | Index calculation optimization system based on AI |
Also Published As
Publication number | Publication date |
---|---|
CN108234356B (en) | 2021-07-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108234356A (en) | Optimization application resource Distribution Strategy based on application relational network | |
Jiang et al. | Heterogeneity-aware distributed parameter servers | |
Wu et al. | Energy and migration cost-aware dynamic virtual machine consolidation in heterogeneous cloud datacenters | |
Taherizadeh et al. | Key influencing factors of the Kubernetes auto-scaler for computing-intensive microservice-native cloud-based applications | |
CN110249311A (en) | The resource management of virtual machine is directed in cloud computing system | |
CN110249310A (en) | The resource management for virtual machine in cloud computing system | |
Lin et al. | On scientific workflow scheduling in clouds under budget constraint | |
CN117290085A (en) | Method, system and apparatus for dynamically facilitating borderless, high availability system management | |
CN104484222B (en) | A kind of dispatching method of virtual machine based on genetic algorithm | |
Wang et al. | eSplash: Efficient speculation in large scale heterogeneous computing systems | |
CN107291536B (en) | Application task flow scheduling method in cloud computing environment | |
Cao | Topology-aware multi-objective virtual machine dynamic consolidation for cloud datacenter | |
CN106681839B (en) | Elastic calculation dynamic allocation method | |
CN105577475A (en) | Automatic performance test system and method | |
CN106708625A (en) | Minimum-cost maximum-flow based large-scale resource scheduling system and minimum-cost maximum-flow based large-scale resource scheduling method | |
CN103336722A (en) | Virtual machine CPU source monitoring and dynamic distributing method | |
CN104935455A (en) | Cluster construction method and device therefor | |
CN105704054A (en) | Data center network flow migration method and system thereof | |
Ferdaus et al. | Multi-objective, decentralized dynamic virtual machine consolidation using aco metaheuristic in computing clouds | |
Bharathi et al. | Virtual machine placement strategies in cloud computing | |
Sáez et al. | Utility-based decision making for migrating cloud-based applications | |
Cheng et al. | A novel task provisioning approach fusing reinforcement learning for big data | |
Sujaudeen et al. | TARNN: Task‐aware autonomic resource management using neural networks in cloud environment | |
CN109688068A (en) | Network load balancing method and device based on big data analysis | |
WO2023154051A1 (en) | Determining root causes of anomalies in services |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |