CN113010270A - Kubernetes platform-based dynamic resource load balancing scheduling method and system - Google Patents
Kubernetes platform-based dynamic resource load balancing scheduling method and system Download PDFInfo
- Publication number
- CN113010270A CN113010270A CN202110379007.9A CN202110379007A CN113010270A CN 113010270 A CN113010270 A CN 113010270A CN 202110379007 A CN202110379007 A CN 202110379007A CN 113010270 A CN113010270 A CN 113010270A
- Authority
- CN
- China
- Prior art keywords
- resource
- node
- utilization rate
- weight
- resource 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/44—Arrangements for executing specific programs
- G06F9/455—Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
- G06F9/45533—Hypervisors; Virtual machine monitors
- G06F9/45558—Hypervisor-specific management and integration aspects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/44—Arrangements for executing specific programs
- G06F9/455—Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
- G06F9/45533—Hypervisors; Virtual machine monitors
- G06F9/45558—Hypervisor-specific management and integration aspects
- G06F2009/4557—Distribution of virtual machine instances; Migration and load balancing
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a Kubernetes platform-based resource dynamic load balancing scheduling method and a Kubernetes platform-based resource dynamic load balancing scheduling system. And when the resource utilization rate of any one resource index of a certain node exceeds a set load threshold, triggering the rescheduling of the node. According to the invention, on the basis of considering various resource indexes and weights, cluster load is balanced through dynamic scheduling, the pressure of each node is effectively dispersed, and cluster resources are fully utilized.
Description
Technical Field
The invention relates to the technical field of container cloud, in particular to a Kubernetes platform-based resource dynamic load balancing scheduling method and system.
Background
With the development of cloud computing technology, container technology represented by Docker is widely applied in the industry due to its characteristics of lightweight, migratability, and rapid deployment. Kubernetes is the first choice for industry container organization systems because of its superior container organization capabilities. However, Kubernetes only realizes static scheduling of Pod applications, only considers two resource indexes of a CPU and a memory, and cannot meet different resource requirements of the Pod, and meanwhile, as the cluster runs for a long time, the Pod applications are deployed more and more, and the condition of unbalanced node load is easily caused, thereby causing many problems.
Many documents research on optimizing node load and improving cluster resource utilization, but have the following disadvantages: resource indexes referred when the Kubernetes schedules Pod application are not comprehensive, the weight of each resource index is not considered, and the Kubernetes schedule Pod application cannot adapt to refined resource scheduling requirements; in the cluster operation process, the dynamic load balance of a Kubernetes platform is not considered, and the situation of extreme unbalanced load is easy to occur under the condition that the cluster operates for a long time. In recent years, the number of small and medium-sized data centers is increasing, and a scheduling method which considers various indexes and weights and can dynamically balance loads is urgently needed to disperse the pressure of each node and fully utilize cluster resources.
Disclosure of Invention
The invention aims to solve the problem that the existing resource scheduling method based on Kubernetes platform resources cannot adapt to the requirement of refined resource scheduling, and provides a dynamic load balancing scheduling method and system based on Kubernetes platform resources.
In order to solve the problems, the invention is realized by the following technical scheme:
a resource dynamic load balancing scheduling method based on a Kubernetes platform comprises the following steps:
step 1, when a Pod application deployment request is received, a Kubernetes cluster collects the resource utilization rate of all nodes of the cluster;
step 2, for each node of the Kubernetes cluster, calculating the hierarchical analysis weight of each resource index by using a hierarchical analysis method, calculating the entropy weight of each resource index by using an entropy weight method based on the resource utilization rate of the node, and calculating the combination weight of each resource index based on the hierarchical analysis weight and the entropy weight; wherein:
step 3, calculating the relative attaching degree of each node of the Kubernetes cluster by utilizing a multi-attribute decision algorithm according to the resource utilization rate of the node and the combined weight of the resource indexes; wherein:
step 4, deploying the Pod application which is requested to be deployed at present to a node with the highest relative fitting degree in the Kubernetes cluster;
step 5, the Kubernetes cluster monitors the resource utilization rate of all the nodes, when the resource utilization rate of any resource index of a certain node exceeds a set load threshold value, the rescheduling of the node is triggered, at the moment, a newly deployed Pod application is selected from the node for redeployment, and the step 1 is returned;
in the above formulae, wjIs the combined weight of the jth resource indicator, wAjFor the hierarchal analysis weight of the jth resource index, wEjEntropy weight of j-th resource index, SiIs the relative degree of fitness of the ith node, zijWeighting the normalized value of the j resource index at the ith node in the weighted decision matrix,xijresource utilization, max (x), for the jth resource indicator on the ith nodej) The maximum value of the resource utilization rate of the jth resource index on all nodes, min (x)j) Is the minimum value, Z, of the resource utilization rate of the jth resource index on all nodesj +Positive ideal solution, Z, of the jth resource index representing a weighted decision matrixj +=max(z1j,z2j,...,znj),Zj -Negative ideal solution, Z, of the jth resource index representing a weighted decision matrixj -=min(z1j,z2j,...,znj) N, n is the number of nodes, and j is 1, 2.
In the step 1, the resource utilization rate of the node includes resource utilization rates of five resource indexes, namely, CPU, memory, bandwidth, disk capacity, and IO rate.
In the step 5, the Pod application selected for redeployment further needs to satisfy the restart policy of the emulated and the resource service quality of the Burst at the same time.
The resource dynamic load balancing scheduling system based on the Kubernetes platform comprises a data acquisition module, a weight calculation module, a scheduling module, a monitoring module and a rescheduling module;
the data acquisition module is used for deploying Proxy monitoring agents on each node of the cluster, acquiring resource utilization rate information of the nodes and storing the resource utilization rate information in the database;
the weight calculation module is used for calculating the hierarchical analysis weight of each resource index according to the hierarchical analysis method; calculating the entropy weight of each resource index according to the entropy weight method and the resource utilization rate of each resource index, and combining the two weights to obtain the combined weight of each resource index;
the scheduling module is used for applying the combined weight to a TOPSIS multi-attribute decision algorithm, calculating the distance between a Pod application scheduling scheme solution and an ideal optimal solution and a worst solution, and selecting a node with the largest relative fitting degree to deploy Pod application through the sorting of the relative fitting degree;
the monitoring module is used for monitoring the resource utilization rate information and the Pod application information of the node in real time, and if the resource utilization rate of any resource index of the node exceeds a set load threshold value, the node is added into a rescheduling queue;
and the rescheduling module is used for selecting a proper Pod application from the nodes of the rescheduling queue to reschedule.
Compared with the prior art, the method and the device have the advantages that the cluster load is balanced through dynamic scheduling under the basis of considering various resource indexes and weights, the pressure of each node is effectively dispersed, and the cluster resources are fully utilized.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to specific examples.
A resource dynamic load balancing scheduling method based on a Kubernetes platform comprises the following steps:
step 1: whenever a Kubernetes cluster receives a Pod application deployment request, the Kubernetes cluster collects resource utilization rates of m resource indicators on all n nodes thereof.
In this embodiment, the m resource indexes are CPU, memory, bandwidth, disk capacity and IO rate, where m is 5.
Step 2: for each node of the Kubernetes cluster, calculating the hierarchical analysis weight w of each resource index by using an analytic hierarchy processAjWherein j is 1,2, and m is the number of resource indexes.
Under the hierarchical analysis method, if m resource indexes exist, the importance degree of the indexes is determined according to pairwise comparison of the resource indexes, and then a judgment matrix A is obtained (a)ij)m×m,(i=1,2,...,m;j=1,2,...,m),aijIndicating the importance of the ith resource indicator relative to the jth resource indicator. Significance definition as shown in table 1, to verify consistency, CI and RI were calculated:
wherein λ ismaxTo determine the maximum characteristic root of the matrix, m is the order of the matrix, CI is the consistency check, CR is the consistency ratio, and RI is the average consistency index, and the values thereof are shown in table 2. If CR is<0.1, passing consistency check, otherwise, judging that the matrix needs to be corrected.
TABLE 1 degree of relative importance table
TABLE 2 average random consistency index
m | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
RI | 0 | 0 | 0.52 | 0.89 | 1.12 | 1.24 | 1.36 |
Calculating coincidence A.wAj=λmax·wAjIs then calculated, and thenThe weight w of each resource index under the hierarchical analysis method can be obtained by normalizingAj。
And step 3: based on the resource utilization rate of m resource indexes on n nodes of the Kubernetes cluster, the entropy weight w of each resource index is calculated by using an entropy weight methodEjWhere j is 1, 2., m, m is the number of resource indicators, and n is the number of nodes.
Under the entropy weight methodThe resource utilization rates of the five collected indexes form a decision matrix, and X is (X)ij)n×m,(i=1,2,...,n;j=1,2,...,m),xijAnd the resource utilization rate of the jth resource index on the ith node is shown. Normalizing the decision matrix:
wherein, max (x)j) And min (x)j) Respectively representing the maximum value and the minimum value of the resource utilization rate of the jth resource index.
Calculating the proportion P of the value of the ith node in the jth resource index to the resourceij:
Calculating an entropy value e of a resource indicatorj:
Calculating a difference coefficient d of resource indicatorsj:
dj=1-ej
Calculating the weight w of each indexEj:
And 4, step 4: for each node of the Kubernetes cluster, the hierarchical analysis weight w of each resource index obtained in the step 2 is usedAjAnd the entropy weight w of each resource index obtained in the step 3EjCalculating a combining weight w for each resource indicatorjWherein j is 1,2, and m is the number of resource indexes.
Calculating a combined weight w of the two weightsj:
Wherein, wAjAnd wEjThe weights obtained by the analytic hierarchy process and the entropy weight process are respectively.
The analytic hierarchy process weight and the entropy weight are calculated according to the real-time utilization rate of the five resources including the CPU, the memory, the bandwidth, the disk capacity and the IO rate in the cluster, and the subjective and objective factor influence is avoided by combining the two weights.
And 5: based on the combining weight w obtained in step 4jAnd the resource utilization rate of m resource indexes on n nodes of the Kubernetes cluster, and calculating the relative fitting degree S of each node of the Kubernetes cluster by using a TOPSIS multi-attribute decision algorithmiWherein i is 1,2, and n, j is 1,2, and m, n is the number of nodes, and m is the number of resource indexes.
Under the TOPSIS multi-attribute decision algorithm, the resource utilization rates of the five collected indexes form a decision matrix, and X is (X)ij)n×m,(i=1,2,...,n;j=1,2,...,m),xijAnd the resource utilization rate of the jth resource index on the ith node is shown. Normalizing the decision matrix by adopting a range normalization method:
wherein, max (x)j) And min (x)j) Respectively representing the maximum value and the minimum value of the resource utilization rate of the jth resource index.
Combining the weight w obtained in step 4jAfter the weighted decision matrix Z is applied to the decision matrix of the TOPSIS multi-attribute decision algorithm, the weighted decision matrix Z is obtained (Z isij)n×m,i=1,2,...,n;j=1,2,...,m:
Wherein z isijIs the jth resource index weighted normalized value, w, of the ith nodejIs the combining weight of the jth resource indicator.
Calculating a positive ideal solution and a negative ideal solution:
Zj +=max(z1j,z2j,...,znj),j=1,2,...,m
Zj -=min(z1j,z2j,...,znj),j=1,2,...,m
wherein Z isj +The positive ideal solution of the jth resource index representing the weighting decision matrix is formed by the maximum value of the jth resource index weighting normalization values on all the nodes; zj -And the negative ideal solution of the jth resource index representing the weighting decision matrix is formed by the minimum value of the jth resource index weighting normalized values on all the nodes.
Calculating the distance from each node to the positive and negative ideal solutions:
wherein D isi +And Di -Respectively representing Euclidean distances, Z, from each candidate node to the positive and negative ideal solutionsj +And Zj -Positive and negative ideal solutions, z, respectively representing the jth resource index of the weighted decision matrixijAnd weighting the normalized value for the jth resource index of the ith node.
Calculating the relative fitting degree S of each node and the optimal nodei:
Wherein the relative degree of sticking SiThe larger the candidate node, the moreIs suitable for the Pod application which needs to be deployed currently.
Step 6: relative fitting degree S of each node obtained in step 5iAnd sequencing, wherein the dispatcher binds the Pod application which is requested to be deployed at present to a node with the highest relative fitting degree in the Kubernetes cluster, the Kubelet on the node monitors the binding event through the APIServer, and the corresponding mirror image is pulled from the mirror image warehouse in the node according to the specific information of the Pod application and the container is started.
And 7: the Kubernetes cluster monitors the resource utilization of m resource indicators on all n nodes: and when the resource utilization rate of any resource index of a certain node exceeds a load threshold, triggering the rescheduling of the node, selecting a proper and newly deployed Pod application from the node for redeployment, and returning to the step 1.
In this embodiment, when the resource utilization of a certain resource index of a node reaches 85%, rescheduling is triggered.
Selecting the Pod application needing to be redeployed meets the following conditions: and simultaneously, the Pod application with the restart strategy of Onfailed and the resource service quality of Burst is met.
In summary, the method in the embodiment of the present invention can effectively disperse the load pressure of each node, so that the cluster resources are fully utilized.
A resource dynamic load balancing scheduling system based on a Kubernetes platform for realizing the method comprises a data acquisition module, a weight calculation module, a scheduling module, a monitoring module and a rescheduling module.
The data acquisition module is used for deploying Proxy monitoring agents on each node of the cluster, acquiring resource utilization rate information on the nodes and storing the resource utilization rate information in the database;
the weight calculation module is used for calculating the hierarchical analysis weight of each resource index according to the hierarchical analysis method; calculating the entropy weight of each resource index according to the entropy weight method and the resource utilization rate of each index, and combining the two weights to obtain the combined weight of each resource index;
the scheduling module is used for applying the combined weight to a TOPSIS multi-attribute decision algorithm, calculating the distance between a Pod application scheduling scheme solution and an ideal optimal solution and a worst solution, and selecting a node with the largest relative fitting degree to deploy Pod application through the sorting of the relative fitting degree;
the monitoring module is used for monitoring the resource utilization rate information and the Pod application information of the node in real time, and if the resource utilization rate of any resource index of the node is greater than a set load threshold value, the node is added into a high-load (i.e. rescheduling) queue;
and the rescheduling module is used for selecting a proper Pod application from the nodes of the high-load queue to reschedule.
The effectiveness of the invention is illustrated below by using a self-programmed Pod application scheduling simulation platform and setting two control group experiments in total, the set control groups are as follows:
(1) the LeastRequestedpriority algorithm (hereinafter LRP) native to the Kubernetes platform. The algorithm is used for selecting the node with the minimum resource consumption from the candidate nodes, namely the more CPU and memory free resources are, the higher the score is, and the calculation formula is as follows:
(2) the BalancedResourceAllocation algorithm (hereinafter BRA) native to the Kubernetes platform. The algorithm is used for selecting the node with the most balanced CPU and memory utilization rate from the candidate nodes, namely the closer the CPU and the memory utilization rate are, the higher the score is, and the calculation formula is as follows:
wherein, Scpu and Smem respectively represent the total CPU and memory capacity on the node, and Ncpu and Nmem respectively represent the sum of the CPU and memory capacity used on the node plus the CPU and memory capacity of the Pod application to be deployed.
The experiment is operated on 60 simulated nodes, the nodes are divided into 3 types, 20 nodes are provided in each type, as shown in table 3, in view of the requirement of diversified Pod application resources, the Pod resource requirements are simulated according to 6 Pod applications of CPU sensitive type, memory sensitive type, bandwidth sensitive type, disk capacity sensitive type, IO rate sensitive type and trend-free type, and part of the requests are shown in table 4.
TABLE 3 node resource information
TABLE 4Pod application resource Requirements Table
The experimental results are shown in table 5, which proves that the method of the present embodiment can disperse the pressure of each node, so that the cluster resources are fully utilized.
TABLE 5 results of the experiment
Resource balance degree: and the average value of the standard deviation of the utilization rate of each resource on all the nodes is represented, and the smaller the value of the average value is, the more balanced the utilization rate of each resource in the cluster is represented, so that the resource inclination is not easy to occur.
Node resource saturation: the larger the value of the ratio of the number of nodes representing that the utilization rate of any resource in the cluster is greater than 100% to the total number of the nodes, the more uneven the pressure of each node in the cluster is represented.
It should be noted that, although the above-mentioned embodiments of the present invention are illustrative, the present invention is not limited thereto, and thus the present invention is not limited to the above-mentioned embodiments. Other embodiments, which can be made by those skilled in the art in light of the teachings of the present invention, are considered to be within the scope of the present invention without departing from its principles.
Claims (4)
1. A resource dynamic load balancing scheduling method based on a Kubernetes platform is characterized by comprising the following steps:
step 1, when a Pod application deployment request is received, a Kubernetes cluster collects the resource utilization rate of all nodes of the cluster;
step 2, for each node of the Kubernetes cluster, calculating the hierarchical analysis weight of each resource index by using a hierarchical analysis method, calculating the entropy weight of each resource index by using an entropy weight method based on the resource utilization rate of the node, and calculating the combination weight of each resource index based on the hierarchical analysis weight and the entropy weight; wherein:
step 3, calculating the relative attaching degree of each node of the Kubernetes cluster by utilizing a multi-attribute decision algorithm according to the resource utilization rate of the node and the combined weight of the resource indexes; wherein:
step 4, deploying the Pod application which is requested to be deployed at present to a node with the highest relative fitting degree in the Kubernetes cluster;
step 5, the Kubernetes cluster monitors the resource utilization rate of all the nodes, when the resource utilization rate of any resource index of a certain node exceeds a set load threshold value, the rescheduling of the node is triggered, at the moment, a newly deployed Pod application is selected from the node for redeployment, and the step 1 is returned;
in the above formulae, wjIs the combined weight of the jth resource indicator, wAjFor the hierarchal analysis weight of the jth resource index, wEjEntropy weight of j-th resource index, SiIs the relative degree of fitness of the ith node, zijWeighting the normalized value of the j resource index at the ith node in the weighted decision matrix,xijresource utilization, max (x), for the jth resource indicator on the ith nodej) The maximum value of the resource utilization rate of the jth resource index on all nodes, min (x)j) Is the minimum value, Z, of the resource utilization rate of the jth resource index on all nodesj +Positive ideal solution, Z, of the jth resource index representing a weighted decision matrixj +=max(z1j,z2j,...,znj),Zj -Negative ideal solution, Z, of the jth resource index representing a weighted decision matrixj -=min(z1j,z2j,...,znj) N, n is the number of nodes, and j is 1, 2.
2. The Kubernetes platform resource dynamic load balancing scheduling method according to claim 1, wherein in step 1, the resource utilization rate of the node includes resource utilization rates of five resource indexes, namely CPU, memory, bandwidth, disk capacity and IO rate.
3. The Kubernetes platform-based dynamic resource load balancing scheduling method of claim 1, wherein in step 5, the Pod application selected for redeployment further needs to satisfy the restart policy of impacted and the resource quality of service of Burst simultaneously.
4. The Kubernetes platform resource dynamic load balancing scheduling system for realizing the method of claim 1 is characterized by comprising a data acquisition module, a weight calculation module, a scheduling module, a monitoring module and a rescheduling module;
the data acquisition module is used for deploying Proxy monitoring agents on each node of the cluster, acquiring resource utilization rate information of the nodes and storing the resource utilization rate information in the database;
the weight calculation module is used for calculating the hierarchical analysis weight of each resource index according to the hierarchical analysis method; calculating the entropy weight of each resource index according to the entropy weight method and the resource utilization rate of each resource index, and combining the two weights to obtain the combined weight of each resource index;
the scheduling module is used for applying the combined weight to a TOPSIS multi-attribute decision algorithm, calculating the distance between a Pod application scheduling scheme solution and an ideal optimal solution and a worst solution, and selecting a node with the largest relative fitting degree to deploy Pod application through the sorting of the relative fitting degree;
the monitoring module is used for monitoring the resource utilization rate information and the Pod application information of the node in real time, and if the resource utilization rate of any resource index of the node exceeds a set load threshold value, the node is added into a rescheduling queue;
and the rescheduling module is used for selecting a proper Pod application from the nodes of the rescheduling queue to reschedule.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110379007.9A CN113010270A (en) | 2021-04-08 | 2021-04-08 | Kubernetes platform-based dynamic resource load balancing scheduling method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110379007.9A CN113010270A (en) | 2021-04-08 | 2021-04-08 | Kubernetes platform-based dynamic resource load balancing scheduling method and system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113010270A true CN113010270A (en) | 2021-06-22 |
Family
ID=76388105
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110379007.9A Pending CN113010270A (en) | 2021-04-08 | 2021-04-08 | Kubernetes platform-based dynamic resource load balancing scheduling method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113010270A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113886081A (en) * | 2021-09-29 | 2022-01-04 | 南京地铁建设有限责任公司 | Station multi-face-brushing array face library segmentation method based on load balancing |
CN114064296A (en) * | 2022-01-18 | 2022-02-18 | 北京建筑大学 | Kubernetes scheduling method, Kubernetes scheduling device and storage medium |
CN115297112A (en) * | 2022-07-31 | 2022-11-04 | 南京匡吉信息科技有限公司 | Dynamic resource quota and scheduling component based on Kubernetes |
CN115665158A (en) * | 2022-10-31 | 2023-01-31 | 浪潮云信息技术股份公司 | Dynamic management method and system for container cluster service |
CN117971505A (en) * | 2024-03-29 | 2024-05-03 | 苏州元脑智能科技有限公司 | Method and device for deploying container application |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103577888A (en) * | 2013-09-05 | 2014-02-12 | 西安电子科技大学 | Improved entropy weight AHP and application thereof |
CN107734512A (en) * | 2017-09-30 | 2018-02-23 | 南京南瑞集团公司 | A kind of network selecting method based on the analysis of gray scale relevance presenting levelses |
CN109120715A (en) * | 2018-09-21 | 2019-01-01 | 华南理工大学 | Dynamic load balancing method under a kind of cloud environment |
CN109547230A (en) * | 2017-09-22 | 2019-03-29 | 中国移动通信集团浙江有限公司 | A kind of internet cache resources QoS evaluating method and system based on weight |
US20190253490A1 (en) * | 2016-10-31 | 2019-08-15 | Huawei Technologies Co., Ltd. | Resource load balancing control method and cluster scheduler |
CN110780998A (en) * | 2019-09-29 | 2020-02-11 | 武汉大学 | Kubernetes-based dynamic load balancing resource scheduling method |
-
2021
- 2021-04-08 CN CN202110379007.9A patent/CN113010270A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103577888A (en) * | 2013-09-05 | 2014-02-12 | 西安电子科技大学 | Improved entropy weight AHP and application thereof |
US20190253490A1 (en) * | 2016-10-31 | 2019-08-15 | Huawei Technologies Co., Ltd. | Resource load balancing control method and cluster scheduler |
CN109547230A (en) * | 2017-09-22 | 2019-03-29 | 中国移动通信集团浙江有限公司 | A kind of internet cache resources QoS evaluating method and system based on weight |
CN107734512A (en) * | 2017-09-30 | 2018-02-23 | 南京南瑞集团公司 | A kind of network selecting method based on the analysis of gray scale relevance presenting levelses |
CN109120715A (en) * | 2018-09-21 | 2019-01-01 | 华南理工大学 | Dynamic load balancing method under a kind of cloud environment |
CN110780998A (en) * | 2019-09-29 | 2020-02-11 | 武汉大学 | Kubernetes-based dynamic load balancing resource scheduling method |
Non-Patent Citations (1)
Title |
---|
钟立俊等: "基于EW 和TOPSIS 的分布式模型调度技术", 《电子信息对抗技术》 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113886081A (en) * | 2021-09-29 | 2022-01-04 | 南京地铁建设有限责任公司 | Station multi-face-brushing array face library segmentation method based on load balancing |
CN114064296A (en) * | 2022-01-18 | 2022-02-18 | 北京建筑大学 | Kubernetes scheduling method, Kubernetes scheduling device and storage medium |
CN114064296B (en) * | 2022-01-18 | 2022-04-26 | 北京建筑大学 | Kubernetes scheduling method, Kubernetes scheduling device and storage medium |
CN115297112A (en) * | 2022-07-31 | 2022-11-04 | 南京匡吉信息科技有限公司 | Dynamic resource quota and scheduling component based on Kubernetes |
CN115665158A (en) * | 2022-10-31 | 2023-01-31 | 浪潮云信息技术股份公司 | Dynamic management method and system for container cluster service |
CN117971505A (en) * | 2024-03-29 | 2024-05-03 | 苏州元脑智能科技有限公司 | Method and device for deploying container application |
CN117971505B (en) * | 2024-03-29 | 2024-06-07 | 苏州元脑智能科技有限公司 | Method and device for deploying container application |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113010270A (en) | Kubernetes platform-based dynamic resource load balancing scheduling method and system | |
CN107038069A (en) | Dynamic labels match DLMS dispatching methods under Hadoop platform | |
CN103338228B (en) | Cloud computing load balancing dispatching algorithms based on double weighting Smallest connection algorithms | |
CN113347267B (en) | MEC server deployment method in mobile edge cloud computing network | |
CN109711526B (en) | Server cluster scheduling method based on SVM (support vector machine) and ant colony algorithm | |
CN111522667A (en) | Resource scheduling method based on mirror image existence mechanism scoring strategy in container cloud environment | |
CN112866132B (en) | Dynamic load balancer and method for massive identification | |
CN112527448A (en) | Openstack-based dynamic load adjustment method and system | |
CN111796933B (en) | Resource scheduling method, device, storage medium and electronic equipment | |
CN115995838A (en) | Distributed energy storage and electric automobile oriented weighted clustering method, storage medium and electronic equipment | |
CN102331948A (en) | Resource state-based virtual machine structure adjustment method and adjustment system | |
CN112348369A (en) | Multi-target multi-resource dynamic scheduling method for major activity security | |
CN111131447A (en) | Load balancing method based on intermediate node task allocation | |
CN112817757B (en) | Pre-adjusted resource elastic allocation method and device, electronic equipment and storage medium | |
CN111144701B (en) | ETL job scheduling resource classification evaluation method under distributed environment | |
CN114356531A (en) | Edge calculation task classification scheduling method based on K-means clustering and queuing theory | |
CN112882805A (en) | Profit optimization scheduling method based on task resource constraint | |
CN115314500B (en) | Dynamic load balancing method based on improved TOPSIS model | |
CN116360922A (en) | Cluster resource scheduling method, device, computer equipment and storage medium | |
CN115665157A (en) | Balanced scheduling method and system based on application resource types | |
CN106960213A (en) | A kind of Key Unit of Fire Safety grade sequence system analyzed based on big data | |
CN113159540A (en) | Demand side resource cascade calling method and device considering load value | |
CN113138838A (en) | Virtual machine placement method based on artificial bee colony algorithm | |
CN117971505B (en) | Method and device for deploying container application | |
CN114466020B (en) | Service request processing method, device, equipment, storage medium and program product |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210622 |