CN117271143B - Data center optimization energy-saving scheduling method and system - Google Patents
Data center optimization energy-saving scheduling method and system Download PDFInfo
- Publication number
- CN117271143B CN117271143B CN202311558255.5A CN202311558255A CN117271143B CN 117271143 B CN117271143 B CN 117271143B CN 202311558255 A CN202311558255 A CN 202311558255A CN 117271143 B CN117271143 B CN 117271143B
- Authority
- CN
- China
- Prior art keywords
- virtual machine
- correlation
- parameters
- data center
- virtual machines
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000005457 optimization Methods 0.000 title abstract description 13
- 230000008859 change Effects 0.000 claims abstract description 74
- 230000001629 suppression Effects 0.000 claims abstract description 36
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 29
- 238000005265 energy consumption Methods 0.000 claims abstract description 14
- 239000002245 particle Substances 0.000 claims abstract description 13
- 238000013508 migration Methods 0.000 claims description 38
- 230000005012 migration Effects 0.000 claims description 38
- 238000012937 correction Methods 0.000 claims description 34
- 238000004364 calculation method Methods 0.000 claims description 31
- 230000006870 function Effects 0.000 claims description 24
- 230000003044 adaptive effect Effects 0.000 claims description 13
- 239000011159 matrix material Substances 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 6
- 238000013507 mapping Methods 0.000 claims description 4
- 230000000694 effects Effects 0.000 description 6
- 230000003631 expected effect Effects 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 230000007613 environmental effect Effects 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000004134 energy conservation Methods 0.000 description 3
- 230000009467 reduction Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 239000005431 greenhouse gas Substances 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
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
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5061—Partitioning or combining of resources
- G06F9/5077—Logical partitioning of resources; Management or configuration of virtualized resources
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/30—Monitoring
- G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
- G06F11/3452—Performance evaluation by statistical analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- 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
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
- G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
-
- 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)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Data Mining & Analysis (AREA)
- Probability & Statistics with Applications (AREA)
- Computer Hardware Design (AREA)
- Quality & Reliability (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the technical field of data center optimization energy-saving scheduling, in particular to a data center optimization energy-saving scheduling method and system. The method comprises the steps that initial similarity between virtual machines is obtained through the change similarity of CPU utilization rate; further correcting the initial similarity parameters by using the correlation suppression factors obtained according to the mutual influence of the virtual machines under the same physical server, and readjusting by means of the similarity of other allocated resources to obtain target correlation parameters; adjusting the inertia weight according to the obtained correlation parameters, and constructing a scheduling objective function; and finally, an optimal scheduling scheme is obtained, and scheduling of the virtual machine of the data center is completed. According to the invention, the data center is optimally scheduled through the self-adaptive particle swarm algorithm, so that the energy consumption of the data center can be reduced, and the use efficiency of each server of the data center is improved.
Description
Technical Field
The invention relates to the technical field of data center optimization energy-saving scheduling, in particular to a data center optimization energy-saving scheduling method and system.
Background
With the rapid development of cloud computing, artificial intelligence and other technologies in recent years, the scale and energy consumption of data centers are in an increasing trend. The high energy consumption of the data center not only places a burden on the environment, but also places a certain pressure on the economy. The data center optimizing energy-saving scheduling method can help the data center to realize energy conservation and environmental protection, reduce operation cost, improve energy utilization efficiency and stability, promote the data center to advance towards sustainable development, and simultaneously have important significance for reducing greenhouse gas emission, coping with climate change and the like.
When the particle swarm optimization is used for carrying out optimal scheduling on the data center, the optimization effect of the data center is affected by the size of the inertia weight parameter, the scheduling result is not ideal due to improper inertia weight setting, the data center cannot be helped to achieve the purposes of energy conservation, environmental protection, operation cost reduction and the like, and even side effects are generated.
Disclosure of Invention
In order to solve the technical problems that the inertia weight of the traditional particle swarm algorithm is not properly set and the data center cannot be accurately scheduled, the invention aims to provide an optimized energy-saving scheduling method and system for the data center, and the adopted technical scheme is as follows:
A data center optimized energy-efficient scheduling method, the method comprising:
acquiring state parameters corresponding to each physical server in a data center, CPU utilization rate of a virtual machine, the position of the physical server where the virtual machine is located and migration loss parameters of the virtual machine according to preset acquisition frequency;
analyzing the variation similarity of the CPU utilization rate between the virtual machines to obtain initial correlation parameters between each virtual machine and other virtual machines; screening out a high-load time period of the virtual machine according to the CPU utilization rate of the virtual machine; screening out a plurality of servers according to the position of the physical server where each virtual machine is located; in each multi-machine server, obtaining a correlation suppression factor of each multi-machine server according to the fluctuation influence of CPU utilization rate on other virtual machines in the high-load time period of all the virtual machines; correcting the initial correlation parameters between each virtual machine and other virtual machines in each multi-machine server by using the corresponding correlation suppression factors to obtain corrected correlation parameters between each virtual machine and other virtual machines; in each multi-machine server, analyzing the similarity of the allocated resources between each virtual machine and other virtual machines, and combining the corrected correlation parameters to obtain target correlation parameters; obtaining position related features according to all target related parameters of each virtual machine;
The preset inertia weight is adjusted according to the initial correlation parameters of each virtual machine and the virtual machines in the physical servers at different positions and the position correlation characteristics of each virtual machine, so that the self-adaptive inertia weight of each virtual machine is obtained; predicting the state parameters of each physical server after dispatching according to the state parameters of each physical server, a preset state parameter change curve and the CPU utilization rate of the virtual machine, and combining the corresponding migration loss parameters to obtain a dispatching objective function;
obtaining an optimal scheduling scheme by utilizing a particle swarm algorithm and the scheduling objective function according to the self-adaptive inertia weight; and controlling the migration of the virtual machine according to the optimal scheduling scheme to complete energy-saving optimal scheduling of the data center.
Further, the method for acquiring the initial correlation parameter comprises the following steps:
acquiring a CPU utilization rate change curve between the current moment of each virtual machine and the starting moment of the last scheduling; segmenting the CPU utilization rate change curve to obtain a segmented change curve;
selecting any two different virtual machines as a first virtual machine and a second virtual machine, acquiring the curve similarity of each segment change curve of the first virtual machine and the second virtual machine by using a dynamic time warping algorithm, constructing all the curve similarity into a similarity matrix, wherein the transverse length of the similarity matrix is the segment change curve number of the first virtual machine, the longitudinal length is the segment change curve number of the second virtual machine, and the elements in the matrix are the corresponding curve similarity of the segment change curve of the first virtual machine and the segment change curve of the second virtual machine;
Processing the similarity matrix by using a Hungary algorithm to obtain a matching result; clustering the segmented change curves by using a clustering algorithm according to the matched curve similarity, and taking the element in the cluster with the maximum average curve similarity as a curve similarity group to be analyzed;
obtaining a matching correction factor according to the duration time of the segmented change curve of the matching result corresponding to the similarity group of the curve to be analyzed; multiplying the matching correction factor by the average curve similarity of the curve similarity group to be analyzed to obtain initial correlation parameters of the first virtual machine and the second virtual machine; and changing the first virtual machine or the second virtual machine to obtain initial correlation parameters between each virtual machine and other virtual machines.
Further, the method for acquiring the piecewise variation curve comprises the following steps:
clustering by using a clustering algorithm according to the slope change of the CPU utilization rate change curve of each virtual machine, and connecting adjacent data points in the same clustering cluster and the time domain to obtain a segmented change curve.
Further, the method for obtaining the matching correction factor comprises the following steps:
obtaining a matching correction factor by using a matching correction factor calculation formula; the calculation formula of the matching correction factor includes:
The method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>；/>Representing a matching correction factor; />Representing the number of elements in the similarity group of the curve to be analyzed; />A sequence number representing a matching piecewise variation curve; />Indicate->Time difference values of the segment change curves are matched; />、/>Respectively represent +.>Starting time and ending time of each matched sectional change curve; />、/>Respectively represent +.>The start and end times of the segment change curves are matched.
Further, the method for obtaining the correlation suppression factor comprises the following steps:
obtaining a correlation suppression factor according to a calculation formula of the correlation suppression factor; the calculation formula of the correlation suppression factor includes:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>，/>，The expression number is->A correlation suppression factor corresponding to the virtual machine; />Representation +.>The number of virtual machines actually operated at the same position; />Representing virtual machine +.>Is>Virtual machine under high load time period +.>A CPU utilization rate change parameter of the CPU utilization rate; />Indicate->The number of acquisitions during a high load period; />Indicate->Virtual machine under high load time period +.>Is>CPU utilization rate corresponding to each moment; />Indicate->Virtual machine under high load time period +. >CPU utilization corresponding to time 1; />Representing virtual machine +.>Is set, the number of high load time periods of (a); />Representing virtual machine +.>Influence parameters on CPU utilization rate of other virtual machines.
Further, the method for acquiring the correction correlation parameter includes:
and taking the quotient of the initial correlation parameter and the normalized correlation suppression factor as a corrected correlation parameter.
Further, the method for acquiring the target correlation parameter comprises the following steps:
and obtaining the memory difference and the bandwidth difference distributed between the two virtual machines, mapping the standard deviation negative correlation of the memory difference and the bandwidth difference, multiplying the standard deviation negative correlation with the corrected correlation parameter, and normalizing the product to obtain the target correlation parameter.
Further, the method for acquiring the adaptive inertial weight comprises the following steps:
acquiring an adaptive inertia weight according to an adaptive inertia weight calculation formula; the adaptive inertial weight calculation formula comprises:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein->；；/>Representing the current virtual machine +.>Average target correlation parameters of other virtual machines on the same physical server where the average target correlation parameters and the other virtual machines are located; />Representing the current virtual machine +.>The number of other virtual machines on the same physical server as the virtual machine; />Representing the current virtual machine +. >The serial numbers of other virtual machines on the same physical server as the virtual machine; />Representing current virtualMachine->Virtual machine on the same physical server as it is +.>Target correlation parameters of (2); />Representing the current virtual machine +.>Average initial correlation parameters of other virtual machines not on the same physical server; />Representing the current virtual machine +.>The number of other virtual machines that are not on the same physical server as it; />Representing the current virtual machine +.>Serial numbers of other virtual machines which are not on the same physical server;current virtual machine->Virtual machine not on the same physical server as it +.>Is used for the initial correlation parameters of the (a); />、/>Respectively represent the maximum value and the minimum value of preset inertia weightsA value; />Representing the current iteration times; />Represents the maximum number of iterations,/->Representing virtual machine +.>Is provided.
Further, the method for acquiring the scheduling objective function comprises the following steps:
according to the scheme of migrating the virtual machine in each iteration, taking the sum of the migration loss parameters of the migrating virtual machine as a migration cost parameter; predicting the state parameters of each physical server after migration according to the state parameters of each physical server before migration, a preset state parameter change curve and the CPU utilization rate of the virtual machine, and obtaining predicted state parameters of each physical server; multiplying the power in the predicted state parameters by the utilization efficiency of the physical servers to obtain predicted energy consumption parameters of each physical server; and summing the predicted energy consumption parameters corresponding to all the physical servers and then summing the predicted energy consumption parameters with the migration cost parameters to obtain a scheduling objective function.
The invention also provides a data center optimizing energy-saving dispatching system, which comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the processor realizes the steps of any one data center optimizing energy-saving dispatching method when executing the computer program.
The invention has the following beneficial effects:
firstly, acquiring various data required by follow-up so as to facilitate the follow-up analysis; further utilizing the variation similarity of CPU utilization rate to carry out preliminary evaluation on the similarity between virtual machines; further considering that the virtual machine is likely to influence other virtual machines under the high load condition, obtaining a correlation suppression factor by utilizing the fluctuation characteristic of CPU utilization rate generated by the influence of the high load time period of the virtual machine on other virtual machines in the multi-machine server, and correcting the initial correlation parameter to obtain a more accurate corrected correlation parameter; further analyzing the similarity of the allocated resources among the virtual machines in each multi-machine server, and further adjusting the corrected correlation parameters, so that the finally obtained target correlation parameters are more accurate, the condition that resources are wasted due to mobilization of the virtual machines with stronger correlation is reduced, and preparation is made for obtaining an optimal scheduling scheme subsequently; further carrying out self-adaptive adjustment on the inertia weights so that the self-adaptive inertia weights accord with the actual conditions of the respective virtual machines to obtain an optimal scheduling scheme; the state parameters of each physical server, the preset state parameter change curve, the CPU utilization rate of the virtual machine and the corresponding migration loss parameters are further utilized to provide a scheduling objective function for each iteration scheduling, so that an optimal scheduling scheme is conveniently searched; after the self-adaptive inertia weight and the scheduling objective function are further obtained, a particle swarm algorithm is utilized to obtain a scheduling scheme, the migration of the virtual machine is controlled according to the optimal scheduling scheme, the energy-saving optimal scheduling of the data center is completed, the data center is helped to achieve the expected effects of energy saving, environmental protection and operation cost reduction, and the energy utilization efficiency is improved.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a data center optimization energy-saving scheduling method according to an embodiment of the present invention.
Detailed Description
In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following is a detailed description of a specific implementation, structure, characteristics and effects of the data center optimization energy-saving scheduling method and system according to the invention in combination with the accompanying drawings and the preferred embodiment. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following specifically describes a specific scheme of the data center optimization energy-saving scheduling method and system provided by the invention with reference to the accompanying drawings.
Referring to fig. 1, a flowchart of a data center optimization energy-saving scheduling method according to an embodiment of the present invention specifically includes:
step S1: and acquiring state parameters corresponding to each physical server in the data center, CPU utilization rate of the virtual machine, the position of the physical server and migration loss parameters according to the preset acquisition frequency.
In order to analyze the migration effect of the scheduling and determine a migration scheduling scheme for the follow-up, firstly, acquiring various data required by the follow-up so as to facilitate the follow-up analysis; in one embodiment of the invention, the preset acquisition frequency is 1 second for 1 time, and all acquired data are pure values without units, so that the subsequent calculation is convenient.
Step S2: analyzing the variation similarity of CPU utilization rates among the virtual machines, and obtaining initial correlation parameters between each virtual machine and other virtual machines; screening out a high-load time period of the virtual machine according to the CPU utilization rate of the virtual machine; screening out a plurality of servers according to the position of the physical server where each virtual machine is located; in each multi-machine server, obtaining a correlation suppression factor of each multi-machine server according to the fluctuation influence of CPU utilization rate of other virtual machines in a high-load time period of all the virtual machines; correcting the initial correlation parameters between each virtual machine and other virtual machines in each multi-machine server by using corresponding correlation suppression factors to obtain corrected correlation parameters between each virtual machine and other virtual machines; in each multi-machine server, analyzing the similarity of the allocated resources between each virtual machine and other virtual machines, and obtaining target correlation parameters by combining correction correlation parameters; and obtaining the position related characteristics according to all the target related parameters of each virtual machine.
The greater the CPU utilization rate and the similarity of consumed resources among the virtual machines, the stronger the correlation among the virtual machines, and when the scheduling process is carried out, the related virtual machines in the same position are not scheduled as much as possible, and the virtual machines with strong correlations in different positions are scheduled to the physical server in the same position as much as possible; the virtual machines with stronger correlation are placed in the same physical server, so that the resource requirements of the virtual machines can be met more easily, the performance is improved, the virtual machines can share similar resource utilization characteristics, unnecessary server opening can be reduced, the energy consumption is reduced, and the similarity between the virtual machines is analyzed.
Preferably, in one embodiment of the present invention, since the running state of the virtual machine processes multiple tasks, and when the related virtual machine processes the tasks, there may be a part of processing task related, and when the related virtual machine processes the related degree, there is a time delay, if the whole data is directly analyzed, an accurate virtual machine correlation may not be obtained, so the present invention segments the CPU usage rate change curve between the current time and the starting time of the last scheduling of the virtual machine, and matches the segments and then compares the similarities; the segments with higher similarity are selected for analysis, so that the similarity between virtual machines can be highlighted, and a certain calculated amount can be saved, so that a similarity matrix formed by the curve similarity is processed through a Hungary algorithm, and then a set of curve segments to be analyzed is screened out through clustering; meanwhile, the situation that the two virtual machines are highly similar in a short period of time but are dissimilar in other long periods of time is considered, the matching correction factors are obtained by utilizing the duration time of the piecewise change curve of the matching result corresponding to the similarity group of the curve to be analyzed, the accuracy of initial correlation parameters is further improved, and preparation is made for finally obtaining the optimal scheduling scheme.
Based on this, the method for acquiring the initial correlation parameter includes: acquiring a CPU utilization rate change curve between the current moment of each virtual machine and the starting moment of the last scheduling; segmenting a CPU utilization rate change curve to obtain a segmented change curve; selecting any two different virtual machines as a first virtual machine and a second virtual machine, acquiring the curve similarity of each sectional change curve by using a dynamic time warping algorithm, constructing all the curve similarities into a similarity matrix, wherein the transverse length of the similarity matrix is the number of sectional change curve segments of the first virtual machine, the longitudinal length is the number of sectional change curve segments of the second virtual machine, and the elements in the matrix are the corresponding curve similarities of the sectional change curve of the first virtual machine and the sectional change curve of the second virtual machine; processing the similarity matrix by using a Hungary algorithm to obtain a matching result; clustering the segmented change curves by using a clustering algorithm according to the matched curve similarity, and taking the element in the cluster with the maximum average curve similarity as a curve similarity group to be analyzed; obtaining a matching correction factor according to the duration time of the segmented change curve of the matching result corresponding to the similarity group of the curve to be analyzed; multiplying the matching correction factor by the average curve similarity of the curve similarity group to be analyzed to obtain initial correlation parameters of the first virtual machine and the second virtual machine; and changing the first virtual machine or the second virtual machine to obtain initial correlation parameters between each virtual machine and other virtual machines.
The calculation formula of the initial correlation parameter includes:
wherein,representing the first virtual machine->And a second virtual machine->Is used for the initial correlation parameters of the (a); />Representing the first virtual machine->And a second virtual machine->Is a matching correction factor of (a); />Representing the total number of data in the similarity group of the curves to be analyzed; />A sequence number representing data in a similarity group of the curve to be analyzed; />Indicate->And the similarity of the curves.
In the calculation formula of the initial correlation parameter, the larger the average curve similarity of the curve similarity group to be analyzed is, the larger the similarity degree of the first virtual machine and the second virtual machine is, and the larger the initial correlation parameter is; the larger the matching correction factor is, the more similar parts between the two virtual machines are, the longer the duration is, the greater the similarity degree between the first virtual machine and the second virtual machine is, and the greater the initial correlation parameter is; in one embodiment of the present invention, use is made ofThe K-means clustering algorithm is a well known technical means for those skilled in the art, and will not be described in detail herein.
The slope change characteristic in the CPU utilization rate change curve can reflect the change of the working load and the performance of the virtual machine; the clustering algorithm can accurately segment the change curves of different stages to obtain a plurality of segments, so in a preferred embodiment of the present invention, the method for obtaining the segment change curves includes: clustering by using a DBSACN clustering algorithm according to the slope change of the CPU utilization rate change curve of each virtual machine, and connecting adjacent data points in the same cluster and time domain to obtain a sectional change curve. The DBSACN clustering algorithm is a well-known technical means for those skilled in the art, and will not be described in detail herein.
Preferably, in one embodiment of the present invention, in consideration of a situation that two virtual machines are highly similar in a shorter period of time but are dissimilar in other longer periods of time, a matching correction factor is obtained by using a duration and number of piecewise change curves of a matching result corresponding to a similarity group of curves to be analyzed, so as to further improve accuracy of an initial correlation parameter, and the method for obtaining the matching correction factor includes: obtaining a matching correction factor by using a matching correction factor calculation formula; the calculation formula of the matching correction factor includes:
wherein,representing a matching correction factor; />Representing the number of elements in the similarity group of the curve to be analyzed; />A sequence number representing a matching piecewise variation curve; />Indicate->Time difference values of the segment change curves are matched; />Representing the average value of the time difference values of all the matching piecewise variation curves; />、/>Respectively represent +.>Starting time and ending time of each matched sectional change curve; />、/>Respectively represent +.>The start and end times of the segment change curves are matched.
In a calculation formula of the matching correction factor, the more the number of elements in the curve similarity group to be analyzed is, the more similar segments between two virtual machines are described; The larger the duration of the similar segment is explained, the more similar parts are, the longer the duration is, the greater the degree of similarity between virtual machines is, and the greater the matching correction factor is.
Preferably, in one embodiment of the present invention, it is contemplated that CPU usage of a virtual machine is also affected by other virtual machines and running tasks on the physical host. If the CPU resources on the physical host are tense, the CPU utilization rate of the virtual machine may be limited, so that in order to reduce the influence of the CPU high load of the virtual machine on other virtual machines, the similarity among the virtual machines is influenced, and the initial correlation is further inaccurate, so that the CPU utilization rate change of the virtual machine on other physical servers at the same position in the CPU high load time period of the virtual machine is analyzed, and the correlation suppression factor is obtained. The method for acquiring the correlation suppression factor comprises the following steps: obtaining a correlation suppression factor according to a calculation formula of the correlation suppression factor; the calculation formula of the correlation suppression factor includes:
wherein,the expression number is->A correlation suppression factor of a multi-machine server corresponding to the virtual machine; />Representation +.>The number of virtual machines actually operated under the same physical server; / >Representing virtual machine +.>Is>Virtual machine under high load time period +.>A CPU utilization rate change parameter of the CPU utilization rate; />Indicate->The number of acquisitions during a high load period;indicate->Virtual machine under high load time period +.>Is>CPU utilization rate corresponding to each moment; />Indicate->Virtual machine under high load time period +.>CPU utilization corresponding to time 1; />Representing virtual machine +.>Is a number of high load time periods; />Representing virtual machine +.>For other virtual machinesInfluence parameters of CPU usage.
In the calculation formula of the correlation suppression factor, in the virtual machineIn the high-load time period of the (2), the larger the fluctuation of the CPU utilization rate of other virtual machines is, the explanation of the (5) virtual machines is->The larger the influence on other virtual machines in the same physical server is, the larger the influence parameters are, and the larger the correlation suppression factors are; statistical virtual machine->All influencing parameters of the virtual machine which really runs under the same physical server, and the larger the summation result is, the explanation of the virtual machine is +.>In the corresponding physical server, the more common the influence among the virtual machines is, the higher the reliability of the correlation suppression factors is, and the larger the correlation suppression factors are, the lower the reliability of the initial correlation parameters is.
In one embodiment of the present invention, the preset threshold value of the high load time period is selected to be 0.8, and the time period in which the CPU utilization rate of the virtual machine is greater than the preset threshold value is regarded as the high load time period; the method comprises the steps of identifying physical servers with two or more virtual machines in the physical servers as multi-machine servers; and the correlation inhibition factors corresponding to all virtual machines in the same multi-machine server are the same, and repeated calculation is not needed.
The method for acquiring the initial correlation parameter and the correlation suppression factor can know that the initial correlation parameter and the corrected correlation parameter have positive correlation; the correlation suppression factor is inversely related to the corrected correlation parameter, so in a preferred embodiment of the invention, the quotient of the initial correlation parameter and the normalized correlation suppression factor is taken as the corrected correlation parameter. The correction correlation parameter is obtained by using a simple division relation, and in other embodiments of the invention, the practitioner can select other mathematical operation modes conforming to the correlation relation. The calculation formula for correcting the correlation parameter includes:
wherein,representing virtual machine +.>And virtual machine->Is used for correcting the correlation parameter; />Representing virtual machine +. >And virtual machine->Is used for the initial correlation parameters of the (a); />Representing virtual machine +.>And virtual machine->The correlation suppression factors corresponding to the multi-machine servers are located; />Representing the normalized sign.
In the calculation formula for correcting the correlation parameter, the larger the initial correlation parameter is, the virtual machine is describedAnd virtual machine->The greater the degree of correlation, the greater the corrected correlation parameter; the larger the correlation suppression factor is, the virtual machine is described as +>And virtual machine->The more likely the correlation is that caused by the influence on other virtual machines under high load, the lower the reliability of the correlation, the smaller the correction correlation parameter.
Preferably, in one embodiment of the present invention, in order not to limit to the single similarity determination criterion that the CPU utilization is similar, the correction correlation parameter is further adjusted in combination with the similarity of other allocated resources, so as to obtain a more convincing target correlation parameter integrating multiple similarity factors. The method for acquiring the target correlation parameter comprises the following steps: and obtaining the memory difference and the bandwidth difference distributed between the two virtual machines, mapping the standard deviation negative correlation of the memory difference and the bandwidth difference, multiplying the standard deviation negative correlation with the corrected correlation parameter, and normalizing the product to obtain the target correlation parameter. The calculation formula of the target correlation parameter includes:
Wherein,representing virtual machine +.>And virtual machine->Target correlation parameters of (2); />Representing a normalized symbol;representing virtual machine +.>And virtual machine->Is used for correcting the correlation parameter; />Representing virtual machine +.>And virtual machine->Memory differences of (a);representing virtual machine +.>And virtual machine->Is provided.
In the calculation formula of the target correlation parameter, the smaller the memory difference and the bandwidth difference distributed between the two virtual machines are, the smaller the obtained standard deviation is, the larger the standard deviation is after the negative correlation mapping is, which means that the higher the similarity of the two virtual machines on the memory and the bandwidth is, the larger the target correlation parameter is; the larger the correction correlation parameter is, the higher the similarity of the two virtual machines on the CPU utilization rate is, and the larger the target correlation parameter is.
In one embodiment of the present invention, in order to facilitate subsequent adjustment of the inertial weights, a mean value of target correlation parameters of each virtual machine and all other virtual machines in each multi-machine server is used as a location correlation feature.
Step S3: the method comprises the steps of combining initial correlation parameters of each virtual machine and the virtual machines at different positions and position correlation characteristics of each virtual machine, adjusting preset inertia weights, and obtaining self-adaptive inertia weights of each virtual machine; and predicting the state parameters of each physical server after dispatching according to the state parameters of each physical server, a preset state parameter change curve and the CPU utilization rate of the virtual machine, and combining the corresponding migration loss parameters to obtain a dispatching objective function.
After the initial correlation parameters between each virtual machine and other virtual machines and the target correlation parameters of the virtual machines in each multi-machine server are obtained in the step S2, the inertia weight can be adaptively adjusted, so that the self-adaptive inertia weight is more consistent with the actual situation of the virtual machines, the final scheduling scheme can achieve the expected effect that the related virtual machines in the same position are not scheduled as much as possible, the virtual machines with strong correlations in different positions are scheduled to the physical servers in the same position as much as possible, and the scheduling objective function is provided for each iteration scheduling by utilizing the state parameters of each physical server, the preset state parameter change curve, the CPU utilization rate of the virtual machine and the corresponding migration loss parameters, so that the optimal scheduling scheme is convenient to find.
Preferably, in one embodiment of the present invention, in order to enable the adaptive inertia weight to achieve the expected effect that the relevant virtual machines in the same location are not scheduled as much as possible, the virtual machines with strong relevance in different locations are scheduled to the physical server in the same location as much as possible, the target relevance parameters of the virtual machines and other virtual machines in the same location are mapped in a positive correlation manner, the initial relevance parameters of the virtual machines and other virtual machines in the same location are mapped in a negative correlation manner, and taking account of that the average of a plurality of data can ignore the volatility of a single data, it is more convincing to perform average processing on the target relevance parameters and the initial relevance parameters, and the influence of the iteration times on the adaptive inertia weight is increased to accelerate the algorithm convergence speed. The method for acquiring the adaptive inertia weight comprises the following steps: acquiring an adaptive inertia weight according to an adaptive inertia weight calculation formula; the adaptive inertial weight calculation formula includes:
Wherein,representing the current virtual machine +.>Average target correlation parameters of other virtual machines on the same physical server where the target correlation parameters and the other virtual machines are located, namely the correlation characteristics of the positions; />Representing the current virtual machine +.>The number of other virtual machines on the same physical server as the virtual machine; />Representing the current virtual machine +.>The serial numbers of other virtual machines on the same physical server as the virtual machine; />Representing the current virtual machine +.>Virtual machine on the same physical server as it is +.>Target correlation parameters of (2); />Representing the current virtual machine +.>Average initial correlation parameters of other virtual machines not on the same physical server; />Representing the current virtual machine +.>The number of other virtual machines that are not on the same physical server as it; />Representing the current virtual machine +.>Serial numbers of other virtual machines which are not on the same physical server; />Current virtual machine->Virtual machine not on the same physical server as it +.>Is used for the initial correlation parameters of the (a); />、/>Respectively representing the maximum value and the minimum value of preset inertia weight; />Representing the current iteration times;represents the maximum number of iterations,/->Representing virtual machine +.>Is provided.
In the self-adaptive inertia weight calculation formula, the larger the average target correlation parameter between the current virtual machine and other virtual machines on the same physical server, the larger the correlation degree between the current virtual machine and other virtual machines on the same physical server, the smaller the self-adaptive inertia weight is; the larger the average initial correlation parameter of the current virtual machine and other virtual machines which are not on the same physical server, the larger the correlation degree of the current virtual machine and other virtual machines which are not on the same physical server, and the smaller the self-adaptive inertia weight is when the current virtual machine and other virtual machines are scheduled to be moved to the same physical server as much as possible.
Preferably, in one embodiment of the present invention, considering that the number of scheduling virtual machines corresponding to different scheduling schemes and the generated migration costs are different in the iterative process of the particle swarm algorithm, and the changes of the physical servers after the scheduling is completed are different, a scheduling objective function is constructed according to the migration costs of each iterative scheduling scheme and the prediction results of the changes of the physical servers after the scheduling, so as to evaluate the scheduling scheme and find the optimal scheduling scheme.
In one embodiment of the invention, the sum of migration loss parameters of the migration virtual machine is taken as a migration cost parameter according to the scheme of migrating the virtual machine in each iteration; predicting the state parameters of each physical server after migration according to the state parameters of each physical server before migration, a preset state parameter change curve and the CPU utilization rate of the virtual machine, and obtaining predicted state parameters of each physical server; multiplying the power in the predicted state parameters by the utilization efficiency of the physical servers to obtain predicted energy consumption parameters of each physical server; and summing the predicted energy consumption parameters corresponding to all the physical servers and then summing the predicted energy consumption parameters with the migration cost parameters to obtain the scheduling objective function. The calculation formula of the scheduling objective function includes:
Wherein,representing a scheduling objective function; />Representing the number of physical servers; />Representing a physical server serial number; />Indicate->Predicted power of the individual physical servers; />Indicate->Predictive utilization efficiency of the individual physical servers; />Representing the number of scheduled virtual machines; />A sequence number representing a scheduled virtual machine; />Indicate->Migration loss of the individual scheduled virtual machines, +.>Representing a migration cost parameter.
In a calculation formula of the scheduling objective function, the larger the migration loss of the virtual machines is, the larger the number of the migration virtual machines is, the larger the migration cost parameter is, and the larger the scheduling objective function is; the smaller the power and the utilization efficiency in the predicted state parameters of the physical server after the prediction scheduling, the better the energy-saving effect brought by the scheduling is, and the smaller the scheduling objective function is. The smaller the scheduling objective function, the better the scheduling scheme, the closer to the expected effect.
In one embodiment of the present invention, the particle swarm algorithm is configured as follows: the number of the virtual machines is recorded as the population number, the maximum iteration number is 100, the maximum value and the minimum value of the preset inertia weight are respectively 2 and 0.4, the individual learning factor is 2, and the population learning factor is 2; the preset state parameter change curve is related to the actual physical server configuration and is obtained by the actual change of the physical server in actual operation, and is not limited; the method for predicting the state parameters is to utilize the CPU computing power corresponding to the CPU utilization rate of the virtual machine and the state parameters of the current physical server, increase or decrease the corresponding values of the CPU computing power of the corresponding physical server after the virtual machine is simulated and scheduled, and predict the state parameters of the physical server by means of a preset state parameter change curve, wherein the particle swarm algorithm is a technical means well known to those skilled in the art and is not described herein.
Step S4: obtaining an optimal scheduling scheme by utilizing a particle swarm algorithm and a scheduling objective function according to the self-adaptive inertia weight; and controlling the migration of the virtual machine according to the optimal scheduling scheme to complete the energy-saving optimal scheduling of the data center.
After the self-adaptive inertia weight and the scheduling objective function are obtained, a particle swarm algorithm can be used for obtaining a scheduling scheme, when the objective function is minimum, an optimal scheduling scheme is obtained, the migration of the virtual machine is controlled according to the optimal scheduling scheme, the energy-saving optimal scheduling of the data center is completed, the data center is helped to achieve the expected effects of energy conservation, environmental protection and operation cost reduction, and the energy utilization efficiency is improved.
An embodiment of the present invention further provides a data center optimized energy-saving scheduling system, which includes a memory, a processor, and a computer program, where the memory is configured to store a corresponding computer program, and the processor is configured to execute the corresponding computer program, and the computer program is configured to implement a data center optimized energy-saving scheduling method described in steps S1 to S4 when the computer program is executed in the processor.
In summary, the invention provides a data center optimization energy-saving scheduling method and system, which take account of the fact that the scheduling scheme of a data center is not proper enough due to the fact that inertia parameter setting in the existing particle swarm algorithm is not proper, and the operation cost and the resource utilization rate of the data center are affected. Firstly, obtaining initial similarity between virtual machines through the change similarity of CPU utilization rate; further correcting the initial similarity parameters by using the correlation suppression factors obtained according to the mutual influence of the virtual machines under the same physical server, and readjusting by means of the similarity of other allocated resources to obtain target correlation parameters; adjusting the inertia weight according to the obtained correlation parameters, and constructing a scheduling objective function; and finally, an optimal scheduling scheme is obtained, and scheduling of the virtual machine of the data center is completed. According to the invention, the data center is optimally scheduled through the self-adaptive particle swarm algorithm, so that the energy consumption of the data center can be reduced, and the use efficiency of each server of the data center is improved.
It should be noted that: the sequence of the embodiments of the present invention is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments.
Claims (9)
1. A data center optimized energy-saving scheduling method, characterized in that the method comprises:
acquiring state parameters corresponding to each physical server in a data center, CPU utilization rate of a virtual machine, the position of the physical server where the virtual machine is located and migration loss parameters of the virtual machine according to preset acquisition frequency;
analyzing the variation similarity of the CPU utilization rate between the virtual machines to obtain initial correlation parameters between each virtual machine and other virtual machines; screening out a high-load time period of the virtual machine according to the CPU utilization rate of the virtual machine; screening out a plurality of servers according to the position of the physical server where each virtual machine is located; in each multi-machine server, obtaining a correlation suppression factor of each multi-machine server according to the fluctuation influence of CPU utilization rate on other virtual machines in the high-load time period of all the virtual machines; correcting the initial correlation parameters between each virtual machine and other virtual machines in each multi-machine server by using the corresponding correlation suppression factors to obtain corrected correlation parameters between each virtual machine and other virtual machines; in each multi-machine server, analyzing the similarity of the allocated resources between each virtual machine and other virtual machines, and combining the corrected correlation parameters to obtain target correlation parameters; obtaining position related features according to all target related parameters of each virtual machine;
The preset inertia weight is adjusted according to the initial correlation parameters of each virtual machine and the virtual machines in the physical servers at different positions and the position correlation characteristics of each virtual machine, so that the self-adaptive inertia weight of each virtual machine is obtained; predicting the state parameters of each physical server after dispatching according to the state parameters of each physical server, a preset state parameter change curve and the CPU utilization rate of the virtual machine, and combining the corresponding migration loss parameters to obtain a dispatching objective function;
obtaining an optimal scheduling scheme by utilizing a particle swarm algorithm and the scheduling objective function according to the self-adaptive inertia weight; controlling the migration of the virtual machine according to the optimal scheduling scheme to complete energy-saving optimal scheduling of the data center;
the method for acquiring the self-adaptive inertia weight comprises the following steps:
acquiring an adaptive inertia weight according to an adaptive inertia weight calculation formula; the adaptive inertial weight calculation formula comprises:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein->；；/>Representing the current virtual machine +.>Average target correlation parameters of other virtual machines on the same physical server where the average target correlation parameters and the other virtual machines are located; />Representing the current virtual machine +.>The number of other virtual machines on the same physical server as the virtual machine; / >Representing the current virtual machine +.>The serial numbers of other virtual machines on the same physical server as the virtual machine; />Representing the current virtual machine +.>Virtual machine on the same physical server as it is +.>Target correlation parameters of (2); />Representing the currentVirtual machine->Average initial correlation parameters of other virtual machines not on the same physical server; />Representing the current virtual machine +.>The number of other virtual machines that are not on the same physical server as it; />Representing the current virtual machine +.>Serial numbers of other virtual machines which are not on the same physical server; />Current virtual machine->Virtual machine not on the same physical server as it +.>Is used for the initial correlation parameters of the (a); />、Respectively representing the maximum value and the minimum value of preset inertia weight; />Representing the current iteration times; />Represents the maximum number of iterations,/->Representing virtual machine +.>Is provided.
2. The method for optimizing energy-saving scheduling of a data center according to claim 1, wherein the method for acquiring the initial correlation parameter comprises:
acquiring a CPU utilization rate change curve between the current moment of each virtual machine and the starting moment of the last scheduling; segmenting the CPU utilization rate change curve to obtain a segmented change curve;
Selecting any two different virtual machines as a first virtual machine and a second virtual machine, acquiring the curve similarity of each segment change curve of the first virtual machine and the second virtual machine by using a dynamic time warping algorithm, constructing all the curve similarity into a similarity matrix, wherein the transverse length of the similarity matrix is the segment change curve number of the first virtual machine, the longitudinal length is the segment change curve number of the second virtual machine, and the elements in the matrix are the corresponding curve similarity of the segment change curve of the first virtual machine and the segment change curve of the second virtual machine;
processing the similarity matrix by using a Hungary algorithm to obtain a matching result; clustering the segmented change curves by using a clustering algorithm according to the matched curve similarity, and taking the element in the cluster with the maximum average curve similarity as a curve similarity group to be analyzed;
obtaining a matching correction factor according to the duration time of the segmented change curve of the matching result corresponding to the similarity group of the curve to be analyzed; multiplying the matching correction factor by the average curve similarity of the curve similarity group to be analyzed to obtain initial correlation parameters of the first virtual machine and the second virtual machine; and changing the first virtual machine or the second virtual machine to obtain initial correlation parameters between each virtual machine and other virtual machines.
3. The method for optimizing energy-saving scheduling of a data center according to claim 2, wherein the method for acquiring the piecewise variation curve comprises:
clustering by using a clustering algorithm according to the slope change of the CPU utilization rate change curve of each virtual machine, and connecting adjacent data points in the same clustering cluster and the time domain to obtain a segmented change curve.
4. The method for optimizing energy-saving scheduling of a data center according to claim 2, wherein the method for obtaining the matching correction factor comprises:
obtaining a matching correction factor by using a matching correction factor calculation formula; the calculation formula of the matching correction factor includes:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>；/>Representing a matching correction factor; />Representing the number of elements in the similarity group of the curve to be analyzed; />A sequence number representing a matching piecewise variation curve; />Indicate->Time of matching the piecewise change curveA difference value; />、/>Respectively represent +.>Starting time and ending time of each matched sectional change curve; />、/>Respectively represent +.>The start and end times of the segment change curves are matched.
5. The method for optimizing energy-saving scheduling of a data center according to claim 1, wherein the method for obtaining the correlation suppression factor comprises:
Obtaining a correlation suppression factor according to a calculation formula of the correlation suppression factor; the calculation formula of the correlation suppression factor includes:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>，/>，/>The expression number is->A correlation suppression factor corresponding to the virtual machine; />Representation +.>The number of virtual machines actually operated at the same position; />Representing virtual machine +.>Is>Virtual machine under high load time period +.>A CPU utilization rate change parameter of the CPU utilization rate; />Indicate->The number of acquisitions during a high load period; />Indicate->Virtual machine under high load time period +.>Is>Corresponding to each momentCPU utilization; />Indicate->Virtual machine under high load time period +.>CPU utilization corresponding to time 1; />Representing virtual machine +.>Is set, the number of high load time periods of (a); />Representing virtual machine +.>Influence parameters on CPU utilization rate of other virtual machines.
6. The method for optimizing energy-saving scheduling of a data center according to claim 1, wherein the method for acquiring the correction correlation parameter comprises:
and taking the quotient of the initial correlation parameter and the normalized correlation suppression factor as a corrected correlation parameter.
7. The method for optimizing energy-saving scheduling of a data center according to claim 1, wherein the method for acquiring the target correlation parameter comprises:
And obtaining the memory difference and the bandwidth difference distributed between the two virtual machines, mapping the standard deviation negative correlation of the memory difference and the bandwidth difference, multiplying the standard deviation negative correlation with the corrected correlation parameter, and normalizing the product to obtain the target correlation parameter.
8. The data center optimized energy-saving scheduling method according to claim 1, wherein the method for obtaining the scheduling objective function comprises:
according to the scheme of migrating the virtual machine in each iteration, taking the sum of the migration loss parameters of the migrating virtual machine as a migration cost parameter; predicting the state parameters of each physical server after migration according to the state parameters of each physical server before migration, a preset state parameter change curve and the CPU utilization rate of the virtual machine, and obtaining predicted state parameters of each physical server; multiplying the power in the predicted state parameters by the utilization efficiency of the physical servers to obtain predicted energy consumption parameters of each physical server; and summing the predicted energy consumption parameters corresponding to all the physical servers and then summing the predicted energy consumption parameters with the migration cost parameters to obtain a scheduling objective function.
9. A data center optimized energy saving scheduling system, the system comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the steps of a data center optimized energy saving scheduling method according to any one of claims 1-8 when executing the computer program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311558255.5A CN117271143B (en) | 2023-11-22 | 2023-11-22 | Data center optimization energy-saving scheduling method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311558255.5A CN117271143B (en) | 2023-11-22 | 2023-11-22 | Data center optimization energy-saving scheduling method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117271143A CN117271143A (en) | 2023-12-22 |
| CN117271143B true CN117271143B (en) | 2024-01-26 |
Family
ID=89218145
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311558255.5A Active CN117271143B (en) | 2023-11-22 | 2023-11-22 | Data center optimization energy-saving scheduling method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117271143B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117573373B (en) * | 2024-01-15 | 2024-03-26 | 中诚华隆计算机技术有限公司 | CPU virtualization scheduling method and system based on cloud computing |
| CN117828389B (en) * | 2024-03-05 | 2024-06-04 | 深圳市互盟科技股份有限公司 | Distributed system monitoring method based on intelligent operation and maintenance |
| CN118059428B (en) * | 2024-04-17 | 2024-07-26 | 大连欣洋电子设备有限公司 | Intelligent pressure monitoring method for vehicle-mounted superfine dry powder fire extinguisher |
| CN118297285B (en) * | 2024-06-03 | 2024-08-16 | 北京盈泽世纪科技发展有限公司 | Energy efficiency optimization method and system for data center |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109067834A (en) * | 2018-06-29 | 2018-12-21 | 南京邮电大学 | Discrete particle cluster dispatching algorithm based on oscillatory type inertia weight |
| CN110795208A (en) * | 2019-10-11 | 2020-02-14 | 南京航空航天大学 | Mobile cloud computing self-adaptive virtual machine scheduling method based on improved particle swarm |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110851272B (en) * | 2019-10-30 | 2022-02-11 | 内蒙古农业大学 | Cloud task scheduling method based on phagocytic particle swarm genetic hybrid algorithm |
-
2023
- 2023-11-22 CN CN202311558255.5A patent/CN117271143B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109067834A (en) * | 2018-06-29 | 2018-12-21 | 南京邮电大学 | Discrete particle cluster dispatching algorithm based on oscillatory type inertia weight |
| CN110795208A (en) * | 2019-10-11 | 2020-02-14 | 南京航空航天大学 | Mobile cloud computing self-adaptive virtual machine scheduling method based on improved particle swarm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117271143A (en) | 2023-12-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN117271143B (en) | Data center optimization energy-saving scheduling method and system | |
| CN110096349B (en) | Job scheduling method based on cluster node load state prediction | |
| CN109714400B (en) | Container cluster-oriented energy consumption optimization resource scheduling system and method thereof | |
| CN113515351B (en) | Resource scheduling implementation method based on energy consumption and QoS (quality of service) cooperative optimization | |
| CN110321222B (en) | Decision tree prediction-based data parallel operation resource allocation method | |
| KR101113006B1 (en) | Apparatus and method for clustering using mutual information between clusters | |
| CN113627871A (en) | Workflow scheduling method, system and storage medium based on multi-target particle swarm algorithm | |
| CN110334026B (en) | CS-SPSO algorithm-based combined test case generation method | |
| CN108171550B (en) | Behavior data analysis method and device and storage medium | |
| CN112346866B (en) | GPU scheduling method and system based on asynchronous data transmission | |
| US11551095B2 (en) | Sharing preprocessing, computations, and hardware resources between multiple neural networks | |
| CN114237884A (en) | Cloud computing data center energy consumption optimization method based on gray Markov chain | |
| CN117435451A (en) | Method for establishing power consumption and performance model of virtual computing unit in mobile edge computing | |
| CN112035234A (en) | Distributed batch job distribution method and device | |
| CN112053006A (en) | Migration learning-based optimization time acceleration method and system for combined cooling heating and power system | |
| CN111813512B (en) | High-energy-efficiency Spark task scheduling method based on dynamic partition | |
| GB2593355A (en) | Reservoir simulation systems and methods to dynamically improve performance of reservoir simulations | |
| CN107277824B (en) | Virtual base station clustering method under C-RAN architecture | |
| CN112052087A (en) | Deep learning training system and method for dynamic resource adjustment and migration | |
| CN113487215A (en) | Intelligent evaluation and regulation method for indexes of green building group | |
| Zhang et al. | A new meta-heuristic task scheduling algorithm for optimizing energy efficiency in data centers | |
| CN108829500B (en) | Dynamic energy-saving scheduling method for modular parallel operation in cloud environment | |
| US20240095081A1 (en) | Optimizing operation of high-performance computing systems | |
| Zhang et al. | An improved speculative strategy for heterogeneous spark cluster | |
| Chen | Effective scheduling simulation of Internet of things load balanced sharing of resources |
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 |