CN104881271B - Cloud dynamic management approach and device - Google Patents
Cloud dynamic management approach and device Download PDFInfo
- Publication number
- CN104881271B CN104881271B CN201410069284.XA CN201410069284A CN104881271B CN 104881271 B CN104881271 B CN 104881271B CN 201410069284 A CN201410069284 A CN 201410069284A CN 104881271 B CN104881271 B CN 104881271B
- Authority
- CN
- China
- Prior art keywords
- load
- sequence
- load value
- value
- cloud
- 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
Abstract
The invention discloses a kind of cloud dynamic management approach and devices, belong to the cloud of computer field, design to solve the problems such as existing cloud management method time delay is big, accuracy is inadequate.The method includes：Obtain the load information of cloud system；The load information includes the first load sequence in chronological sequence sequentially formed；Judge in first load sequence with the presence or absence of the abnormal load value for deviateing the first load sequence overall variation rule；If it exists, according to the normal load value and preset first function relationship in first load sequence, the replacement load value for replacing the abnormal load value is calculated；The replacement load value is replaced into the abnormal load value in first load sequence and forms the second load sequence；Cloud management is carried out according to second load sequence.
Description
Technical field
The present invention relates to the cloud management technology of computer field more particularly to a kind of cloud dynamic management approach and device.
Background technique
Cloud system, which refers to, is distributed in the system completed on a large amount of distributed computer for calculating, can using cloud system platform
To provide information service by network for user --- cloud service.
Relative to traditional software forms, the cloud computing in cloud system have loose couplings, on-demand, cost it is controllable,
The significant advantages such as virtual, the isomery collaboration of resource, make it more adapt to e-commerce, flexible manufacturing, mobile Internet of cash etc.
Using.Cloud computing includes two aspects：First is that the cloud system platform infrastructure of bottom building, is for constructing upper layer application journey
The basis of sequence；Second is that cloud system application program of the building on the basic platform.
Cloud system itself belongs to an emerging technology areas, and relevant technology, theory and method are existing also in the growth stage
Cloud resource administrative skill, there are many deficiencies, such as due to existing cloud resource (physical machine in specific IaaS system) run
When, be mostly system verify access load increase or decrease after, just make the decision for increasing or decreasing resources of virtual machine, have
Apparent hysteresis quality, can not cope with the case where load increases sharply or reduces sharply, and the efficiency of management and management precision are all very low.
Summary of the invention
To solve the above problems, the invention reside in provide a kind of cloud system that can improve to the efficiency of cloud management and management
The cloud dynamic management approach and device of accuracy.
In order to achieve the above objectives, the technical proposal of the invention is realized in this way：
First aspect present invention provides a kind of cloud dynamic management approach, the method includes：
The load information of cloud system is obtained, the load information includes the first load sequence；
Judge in first load sequence with the presence or absence of the exception for deviateing the first load sequence overall variation rule
Load value；
If it exists, it according to the normal load value and preset first function relationship in first load sequence, calculates
Replace the replacement load value of the abnormal load value；
The replacement load value is replaced into the abnormal load value in first load sequence and forms the second load sequence；
Cloud management is carried out according to second load sequence.
Preferably, with the presence or absence of deviation the first load sequence overall variation in judgement first load sequence
Rule abnormal load value include：
Calculate the positive change rate PR of first load sequencemAnd negative change rate NRm；
The load value for meeting following relationship in first load sequence is abnormal load value；
ldj-ldj-1>β×PRmAnd ldj+1-ldj<β×NRm
Or
ldj-ldj-1<β×NRmAnd ldj+1-ldj>β×PRm
The ldjFor j-th of load value in first load sequence；
The ldj-1For -1 load value of jth in first load sequence；
The ldj+1For+1 load value of jth in first load sequence；
The β is coefficient factor；
Wherein, the j is the positive integer greater than 1.
Preferably, the preset first function relationship is：
The ldj1' it is to replace jth in first load sequence1A load value ldj1Replacement load value, and it is described
ldj1For abnormal load value；
The ldkIt for k-th of load value in first load sequence, and is normal load value；
The ldlIt for first of load value in first load sequence, and is normal load value.
Preferably, described to include according to second load sequence progress cloud management：
According to second load sequence and preset second function relationship, computational load increment；
Administrative decision is obtained according to the incremental loading, and carries out cloud management by obtained administrative decision.
Preferably, the second function relationship is：
Wherein, the aincmFor incremental loading；The ldm-x+1For the m-x+1 load in second load sequence
Value；The ldm-xFor the m-x load value in second load sequence；The γ is attenuation coefficient；The n is second negative
Carry the number of load value in sequence；The y is the positive integer less than n；
Administrative decision is obtained according to the incremental loading, and carries out cloud management by obtained administrative decision, is specifically included：
Work as aincm>0 andWhen, then by s in cloud system1A temperature state physical machine is converted into Warm status object
Reason machine；
WhenWhen, then by s in cloud system2A cold state physical machine is converted into temperature state physical machine；
Work as aincm<0 andWhen, then by s in cloud system3A Warm status physical machine is converted into temperature state object
Reason machine；
WhenWhen, then by s in cloud system4A temperature state physical machine is converted into cold state physical machine；
Wherein,
The α1For first threshold；The α2For second threshold, and 0<α1<α2；
The ldmx is preset maximum load value in cloud system；The ldmFor m-th of load in the second load sequence
Value；
The VhotFor the upper limit of Warm status physical machine can be provided in cloud system；
The VwarmFor the upper limit of temperature state physical machine can be provided in cloud system.
Second aspect of the present invention provides a kind of cloud dynamic management device, and described device includes：
Load information acquiring unit, to obtain the load information of cloud system；The load information includes the first load sequence
Column；
Judging unit integrally becomes in first load sequence with the presence or absence of deviation first load sequence to judge
The abnormal load value of law；
Replace load value and form unit, in first load sequence there are when abnormal load value, according to described
Normal load value and preset first function relationship in first load sequence, calculate the replacement for replacing the abnormal load value
Load value；
Second load sequence forms unit, the replacement load value is replaced the exception in first load sequence
Load value forms the second load sequence；
Administrative unit, to carry out cloud management according to second load sequence.
Further, the judging unit includes
Computing module, to calculate the positive change rate PR of first load sequencemAnd negative change rate NRm；
Judgment module to the judgement of abnormal load value, and meets negative in first load sequence of following formula
Load value is abnormal load value；
ldj-ldj-1>β×PRmAnd ldj+1-ldj<β×NRm
Or
ldj-ldj-1<β×NRmAnd ldj+1-ldj>β×PRm
The ldjFor j-th of load value in first load sequence；
The ldj-1For -1 load value of jth in first load sequence；
The ldj+1For+1 load value of jth in first load sequence；
The β is coefficient factor；
Wherein, the j is the positive integer greater than 1.
Further, the preset first function relationship is
The ldj1' it is to replace jth in first load sequence1A load value ldj1Replacement load value, and it is described
ldj1For abnormal load value；
The ldkIt for k-th of load value in first load sequence, and is normal load value；
The ldlIt for first of load value in first load sequence, and is normal load value.
Further, the administrative unit includes：
Incremental loading computing module, to calculate according to second load sequence and preset second function relationship
Incremental loading；
Management module to obtain administrative decision according to the incremental loading, and carries out cloud pipe by obtained administrative decision
Reason.
Further, the second function relationship is：
Wherein, the aincmFor incremental loading；The ldm-x+1For the m-x+1 load in second load sequence
Value；The ldm-xFor the m-x load value in second load sequence；The γ is attenuation coefficient；The y is；The n
For the number of load value in the second load sequence；
The management module, specifically to work as aincm>0 andWhen, then by s in cloud system1A temperature state
Physical machine is converted into Warm status physical machine；WhenWhen, by s in cloud system2A cold state physical machine conversion
At temperature state physical machine；Work as aincm<0 andWhen, by s in cloud system3A Warm status physical machine is converted into warm shape
State physical machine；WhenWhen, by s in cloud system4A temperature state physical machine is converted into cold state physical machine；
Wherein, the α1For first threshold；The α2For second threshold, and 0<α1<α2；
The ldmx is preset maximum load value in cloud system；The ldmFor m-th of load in the second load sequence
Value；
The VhotFor the upper limit of Warm status physical machine can be provided in cloud system；
The VwarmFor the upper limit of temperature state physical machine can be provided in cloud system.
The embodiment of the invention provides a kind of cloud dynamic management approach and devices, are carried out down based on cloud system historic load value
The method of one moment cloud management, and for further optimum management, the abnormal load value of historic load value is replaced
Processing, to have, management time delay is small, control efficiency is high, eliminates abnormal load value, so that management control is more accurate.
Detailed description of the invention
Fig. 1 is the flow diagram of cloud dynamic management approach described in first embodiment of the invention；
Fig. 2 is the structural schematic diagram of cloud dynamic management device described in second embodiment of the invention.
Specific embodiment
The application is described in further detail below in conjunction with Figure of description and embodiment.
First embodiment：
The present invention provides a kind of cloud dynamic management approach, as shown in Figure 1, the method includes：
Step S110：The load information of cloud system is obtained, the load information includes the first load sequence；
Here, first load sequence in chronological sequence sequentially forms；
Step S120：Judge in first load sequence with the presence or absence of deviation the first load sequence overall variation rule
The abnormal load value of rule；
Step S130：When in first load sequence there are when abnormal load value, according in first load sequence
Normal load value and preset first function relationship, calculate the replacement load value for replacing the abnormal load value；
Step S140：It is negative that the replacement load value is replaced into the formation second of the abnormal load value in first load sequence
Carry sequence；
Step S150：Cloud management is carried out according to second load sequence.
It wherein, then can be directly according to first load when abnormal duplication value is not present in first load sequence
Sequence carries out cloud management.
The first load sequence formed in the step S110 includes load value corresponding to several moment of cloud system, and
The load value is arranged according to chronological order, and the specific physical meaning of load value can be in cloud system in busy shape
The resource of state accounts for the ratio of total resources in cloud system.In general, the time difference between two load values of arbitrary neighborhood is the value period；
The value period is 50 milliseconds to 1000 milliseconds and differs, specific value can according to the load situation of change of cloud system and
Regulatory requirement is configured, specifically such as：100,200,500,700 or 800 milliseconds etc..
The load value can be denoted as ldi, the i is the ldiSequence in the first load sequence, usual time get over
The forward then described i is smaller.Specific first load sequence can be denoted as ldm-n+1... ..ldm-1, ldm.In general, cloud system
It brings into operation and begins to obtain load value according to the value period, when the Long-term change trend for needing to predict subsequent time, to carry out cloud
When the resource management and scheduling of system, it usually needs the load value in nearest n value period is taken, therefore in the ldm-n+1,
.....ldm-1, ldmIn first load sequence, the ldmThe usually load value in value period corresponding to current time.
The load information for specifically how obtaining the cloud system, can be adopted by equipment such as data collectors to carry out data
Collection or other detection devices carry out the methods of the statistics of each resource busy state in cloud system, obtain the load value.Specific
Implementation process in, the load information not only includes the first load sequence as described above, can also include and cloud system transport
The relevant other information of row.The other information may include the Selecting time of first load sequence and load to described first
The information such as the loading condition analysis of sequence.
When cloud system operates normally, the variation of resource busy state in cloud system, changing rule should have in time
Continuity.If occurring some point in continuous load value all to vary widely with respect to the load value before and after it,
This variation seems very lofty, usually indicates that exception has occurred, corresponding load value is abnormal load value.
Cloud system is caused to occur being broadly divided into two kinds there are many reason of abnormal load value：
The first is system factor, specific such as message exception deferral, the variation of connection bandwidth, calculation resources conflict；
Second is nonsystematic factor, specifically the accidental failure of such as system, software or hardware, information drop-out.
The first load sequence overall variation rule in the step S120, i.e., for according to temporal load value
Variation continuity, specific manifestation can be to show a increasing trend, downward trend or carry out in a certain range one or more each
Kind fluctuation, the fluctuation such as first increases to drop or first drop afterwards to be increased afterwards.The implementation method of the step S120 also there are many, specifically such as
Picture method：
Firstly, ordinate be load value, abscissa be load value serial number coordinate system in form the first load sequence wave
Shape figure；
Secondly, finding out the point of mutation by compared with preset variation range of waveforms figure, the point of the mutation is to deviate from
The abnormal load value of first load sequence changing rule；Or directly observation method finds out lofty load value；Or appointed by calculating
The slope that adjacent two o'clock of anticipating is formed in the coordinate system, then compared with preset threshold, if a certain load value it is adjacent thereto two
Slope between a load value is both greater than threshold value, then it is believed that corresponding load value is abnormal load value.
There are many acquisition methods of the specific abnormal load value, just no longer repeats one by one herein.
The step S130 is the replacement load value that the abnormal load value is obtained by calculating, and second so as to formation is negative
It carries sequence and is more in line with cloud system actual loading variation.There are many acquisition methods of the replacement load value, specific as by institute
Mean value is taken after stating the summation of former and later two normal load values of abnormal load value, the modifying factors such as coefficient factor can also be passed through as needed
Son to the mean value computation, during concrete implementation also there are many, also no longer repeat one by one herein.
In the step S140, the abnormal load value is replaced with the replacement load value being calculated in step S130,
The second load sequence is just obtained, second load sequence eliminates the abnormal load value in first load sequence, protects
Normal load value has been stayed, has run consumed resource so as to properer reflection cloud system, and according to the second load sequence
The cloud management of cloud system is carried out, scheduling of resource is more reasonable, and resource distribution is more in line with demand.
In summary, cloud dynamic management approach described in the present embodiment, firstly, predicting that cloud system obtains according to historic load value
The variation tendency of scheduling of resource is taken, and cloud scheduling is carried out according to prediction, so as to timely carry out scheduling of resource, and can be quickly
The situation that reply load increases sharply and reduces sharply；Secondly, it is contemplated that the abnormal load value of load value, and using corresponding replacement load value
The abnormal load value is replaced, so that cloud management precision caused by avoiding because of abnormal load value is low to ask when carrying out cloud management
Topic, has advanced optimized the resource distribution of cloud management.
It is obtained in the step S120 in the first load sequence there are many kinds of the modes of abnormal load value, presented below one
Kind specifically includes following steps relative to easier, quick, the accurate method of the methods of the picture method：
Step 1：Calculate the positive change rate PR of first load sequencemAnd negative change rate NRm；
The ldvFor v-th of load value in first load sequence；The lduFor the u in the first load sequence
A load value；The n is the number of load value in the first load sequence；
Step 2：The load value for meeting following relationship in first load sequence is then abnormal load value；
ldj-ldj-1>β×PRmAnd ldj+1-ldj<β×NRm
Or
ldj-ldj-1<β×NRmAnd ldj+1-ldj>β×PRm
The ldjFor j-th of load value in first load sequence；The ldj-1For in first load sequence
- 1 load value of jth；The ldj+1For+1 load value of jth in first load sequence；
The β is coefficient factor；Usual β is greater than 1；
Wherein, the j is the positive integer greater than 1.
By repeatedly it is demonstrated experimentally that using the first load sequence entirety is deviateed in the above method the first load sequence of acquisition
The method of the abnormal load value of changing rule has the advantages that realize that easy, quick and accuracy rate is high.
As the further improvement of the present embodiment, the present embodiment additionally provides a kind of calculate and replaces the preferred of load value
Preset first function relationship, the preset first function relationship are：
The ldj1' it is to replace jth in first load sequence1A load value ldj1Replacement load value, and it is described
ldj1For abnormal load value；
The ldkIt for k-th of load value in first load sequence, and is normal load value；
The ldlIt for first of load value in first load sequence, and is normal load value.
Using above-mentioned preset first function relationship, calculated replacement load value can follow the change of load value well
Law, and can be well close to the actual loading value at corresponding moment, to there is the resource optimization for being further conducive to cloud management
Management and scheduling.
Further, the present embodiment is embodied to described according to second load sequence progress cloud management：
Step A：According to second load sequence and preset second function relationship, computational load increment；
Step B：Administrative decision is obtained according to the incremental loading, and carries out cloud management by obtained administrative decision.
The incremental loading can be positive incremental loading and be also possible to negative sense incremental loading.In specific application process
In, it, then can be according to the first load sequence and preset second when abnormal load value is not present in first load sequence
Functional relation calculates incremental loading；Cloud management is carried out further according to the incremental loading.Cloud management herein is cloud system medium cloud
The management of resource.
Cloud system can be divided into three classes according to service type：Using infrastructure as service (IaaS, Infrastructure
As a Service), using platform as service (PaaS, Platform as a Service) and using software as service
(SaaS, Software as a Service).
IaaS mode cloud system is by system virtualization, multiprocessor virtualization, internal memory virtualization, I/O virtualization etc.
Physical resource is virtualized into resource pool by virtualization technology, these resources are carried out unified manage and dispatch by IaaS system again.
IaaS system user oriented provides computing resource and storage resource.
The advantages of IaaS system, is：Flexible and dynamic management is carried out to resource, makes full use of the injustice of application request
Weighing apparatus property, the speciality such as uncertainty, sparsity, make less computing resource (such as host, storage, network bandwidth) can support compared with
Large-scale application demand, to realize benefit.
In order to obtain balance in terms of performance and energy consumption, physical machine that IaaS system is possessed institute's system (PM,
Physical Machine) state is respectively set to cold (cold), warm (warm) and heat (hot) three kinds of states.Warm status PM's
Energy consumption highest, temperature state PM take second place, and cold state PM is minimum.And the speed of performing task highest (full speed running) of Warm status PM, temperature
State PM takes second place (lower than full speed running), and the speed of service of cold state PM is zero.
It is to carry out service operation and demand sound from the physical host (PM) that can be seen that in above-mentioned introduction in IaaS cloud system
The center answered, the loading condition of system directly determine by the loading condition of its all PM managed, PM load management strategy
Superiority and inferiority, directly determine the operational efficiency of cloud system.In this context, how dynamically negative according to the task of IaaS cloud system
It carries to safeguard the state of PM, to realize energy conservation and the two-win of performance, becomes the hot and difficult issue for research.
Cloud dynamic management approach described in the present embodiment, the IaaS system being particularly suitable in above-mentioned three kinds of systems.For
IaaS system, cloud dynamic management approach described in the present embodiment proposes the preferred of the second function relationship, specific as follows：
Wherein, the aincmFor incremental loading；The ldm-x+1For the m-x+1 load in second load sequence
Value；The ldm-xFor the m-x load value in second load sequence；The γ is attenuation coefficient, and usual value is small
In 1 positive number, and preferably, value is 0.5；The y is；The n is the load value sum of second load sequence.
It is described to be specifically included according to incremental loading progress cloud management：
Work as aincm>0 andWhen, then by s in cloud system1A temperature state physical machine is converted into Warm status object
Reason machine；
WhenWhen, then by s in cloud system2A cold state physical machine is converted into temperature state physical machine；
Work as aincm<0 andWhen, then by s in cloud system3A Warm status physical machine is converted into temperature state object
Reason machine；
WhenWhen, then by s in cloud system4A temperature state physical machine is converted into cold state physical machine；
Wherein, the α1For first threshold；The α2For second threshold, and 0<α1<α2；
The ldmx is preset maximum load value in cloud system；The usual ldmx is less than 100%, usual value range
Arbitrary value between 90%~99%；The value of the second threshold is not more than 0.1, according to usually setting for existing IaaS system
It sets, preferably takes the arbitrary value between 0.03~0.06, such as 0.05.
The ldmFor m-th of load value in second load sequence；
The VhotFor the upper limit of Warm status physical machine can be provided in the cloud system；
The VwarmFor the upper limit of temperature state physical machine can be provided in the cloud system.
Cloud management is carried out to IaaS system using cloud dynamic management approach described in the present embodiment, relative to first in existing method
After particular exam increases or decreases number to load, the change of hot, cold, the warm three kinds of states of physical machine is carried out, retardance is smaller；Together
When this method also change in traditional IaaS system according to previously given ratio and quantity to it is described have physical machine carry out heat,
Switching between cold, warm three kinds of states, the system resource for switching consumption are obviously smaller, it is clear that operation is more finely more in line with tool
Body actual needs.
In the present embodiment, in each state conversion for carrying out physical machine, the method also includes statistics need to currently be turned
The quantity of the physical machine of state is changed, to judge whether to complete the conversion of physical machine state, realizes that cloud system dynamic manages.Tool
Body as needed s2A cold state physical machine is converted into temperature state physical machine, needs to count in advance before specifically being converted
Whether s is had2For conversion, if judging, result has a cold state physical machine, can directly carry out the conversion of physical machine state, no
Then, need to further determine whether occur fortune resource consumption excessively etc. abnormal phenomenon, from and entered abnormality processing step.
In the specific implementation process, cloud dynamic management approach described in the present embodiment is real-time management method or period
Property management method, i.e., it is duplicate to execute the step S110 to step S150, specifically can be execute be step S150 after it is vertical
It is transferred to the step S110；Or a cycle period is waited to enter back into the step S110 after executing the step S150；Specifically
, select any method can be according to the load pace of change of cloud system, cloud system stock number and to the fine of cloud management
Degree requires to set.
In summary, a kind of cloud dynamic management approach that difference is previous is present embodiments provided, the method is by obtaining cloud
The historic load value of system carries out the dynamic management of current cloud system cloud resource, since the acquisition of load value has in time
Continuity so that the speed obtained is fast, and determines that the cloud resource of subsequent time is dispatched with historic load value, it is clear that manage and dispatch
Decision is generated prior to subsequent time, thus it is directly all right according to the manage and dispatch decision execution formed when dispatching, relative to
The existing management moment obtains current load condition in advance, then is managed, it is clear that and the efficiency of management is higher, and time delay is smaller, thus
It can be very good the fierce of reply load to increase and acutely reduce；It manages, eliminates to advanced optimize cloud dynamic simultaneously
Abnormal load value in historic load value replaces to meet the replacement load value of the first load sequence changing rule, thus
Improve the accuracy to fortune resource management and scheduling.
Second embodiment：
A kind of cloud dynamic management device is present embodiments provided, can be used for cloud dynamic manager described in first embodiment
Method, as shown in Fig. 2, described device includes：
Load information acquiring unit 210, to obtain the load information of cloud system；The load information includes the first load
Sequence；
Judging unit 220, it is whole with the presence or absence of first load sequence is deviateed in first load sequence to judge
The abnormal load value of body changing rule；
Replace load value and form unit 230, in first load sequence there are when abnormal load value, according to institute
The normal load value and preset first function relationship in the first load sequence are stated, calculates and replaces replacing for the abnormal load value
Change load value；
Second load sequence forms unit 240, to replace the replacement load value in first load sequence
Abnormal load value forms the second load sequence；
Administrative unit 250, to carry out cloud management according to second load sequence.
The specific structure of the load information acquiring unit 210 can be data collector or cloud system resource spare time busy shape
Detection statistics equipment of state etc., to obtain load information, the load information includes at least the first load sequence, and described the
Load value in one load sequence is usually arranged according to chronological order, when the load sequence is by first n of current time
The load value at quarter is constituted, specific composition such as ldm-n+1... ..ldm-1, ldm。
The specific structure of the replacement load value stars unit 230 can be any device with computing function, specifically
Such as calculator.
The administrative unit 250 can be any one managing device or management module in the prior art.
A kind of difference cloud dynamic management device in the prior art is present embodiments provided, it is negative according to the history of cloud system
Load value predicts subsequent time load variation to carry out cloud management, carries out cloud management again relative to current load condition is obtained
Managing device, have management time delay it is small, can quickly cope with the acute variation of load；Carrying out subsequent time load at the same time
When the prediction of variation, the abnormal load value in historic load value is proposed, eliminates mismanagement caused by abnormal load value, is provided
The problem of source configuration not enough optimizes.
As further improvement of this embodiment, the judging unit includes computing module and judgment module.
The computing module, to calculate the positive change rate PR of the first load sequencemAnd negative change rate NRm；
The judgment module to the judgement of abnormal load value, and meets negative in the first load sequence of following formula
Load value is abnormal load value；
ldj-ldj-1>β×PRmAnd ldj+1-ldj<β×NRm
Or
ldj-ldj-1<β×NRmAnd ldj+1-ldj>β×PRm
The ldjFor j-th of load value in the first load sequence；The ldj-1For the jth -1 in the first load sequence
A load value；The ldj+1For+1 load value of jth in the first load sequence；The β is coefficient factor；The usual β is
It is preset be not less than 1 number.
Wherein, the j is the positive integer greater than 1.
Specifically, the computing module, can be used to carry out the positive change rate PR according to the following formulamAnd negative change rate
NRmCalculating；
The ldvFor v-th of load value in the first load sequence；The lduIt is negative for u-th in the first load sequence
Load value；The n is the number of load value in the first load sequence.
The generic module that calculates can be the functional module of any one achievable above-mentioned calculating, specific such as calculator；
The judging unit can be one and realize that above-mentioned logical relation compares and the equipment such as the comparator that judges.
There are many specific structures of the judging unit, realizes that quick, easy and accuracy is high there is provided herein a kind of
Structure, other structures are just no longer described in further detail herein.
As the further improvement of the present embodiment, there are many preset specific expressions of first function relationship, with
Calculation amount is few in lower offer one, calculates simple relation formula, specific as follows：
The ldj1' it is to replace jth in first load sequence1A load value ldj1Replacement load value, and it is described
ldj1For abnormal load value；
The ldkIt for k-th of load value in first load sequence, and is normal load value；
The ldlIt for first of load value in first load sequence, and is normal load value.
In addition, the present embodiment additionally provides a kind of preferred administrative unit, the administrative unit includes：
Incremental loading computing module, to calculate according to second load sequence and preset second function relationship
Incremental loading；
Management module to obtain administrative decision according to the incremental loading, and carries out cloud pipe by obtained administrative decision
Reason.
Specifically the second function relationship is：
Wherein, the aincmFor incremental loading；The ldm-x+1For the m-x+1 load in second load sequence
Value；The ldm-xFor the m-x load value in second load sequence；The γ is attenuation coefficient；The y is less than n
Positive integer；
The management module, specifically to work as aincm>0 andWhen, then by s in cloud system1A temperature state
Physical machine is converted into Warm status physical machine；WhenWhen, then by s in cloud system2A cold state physical machine turns
Change temperature state physical machine into；Work as aincm<0 andWhen, then by s in cloud system3A Warm status physical machine is converted into
Temperature state physical machine；WhenWhen, then by s in cloud system4A temperature state physical machine is converted into cold state object
Reason machine；
Wherein, the α1For first threshold；The α2For second threshold, and 0<α1<α2；
The ldmx is preset maximum load value in cloud system；The ldmFor m-th of load in the second load sequence
Value；
The VhotFor the upper limit of Warm status physical machine can be provided in cloud system；
The VwarmFor the upper limit of temperature state physical machine can be provided in cloud system.
Administrative unit described in the present embodiment passes through " heat " temperature to physical machine primarily directed in IaaS system " and
What the cloud management mode of " cold " three kinds of states switching was proposed, using the administrative unit to cloud management in IaaS system, relatively
After successive acquisition present load, according to pre-set ratio or quantity, above-mentioned three are carried out to physical machine in IaaS system
The conversion of kind situation, not only time delay is small, and has many advantages, such as that resource consumed by management is small.
Cloud dynamic management device described in the present embodiment can be integrated in the cloud system one or more server
Managing device, is also possible to the managing device of the independent cloud system, then passes through wired or wireless connection type and the cloud system
System is attached, to obtain load information and outgoing management information etc..
The cloud dynamic management device can be with another embodiment include processor, at least one communication interface,
The equipment of storage medium and bus.Software or program are stored in the medium.The processor can be central processing unit
CPU, single-chip microprocessor MCU, digital signal processor DSP and programmable logic array etc..The bus connects the processor, deposits
Storage media and the communication interface.When the processor runs the software or program, following functions can be at least completed：
Obtain the load information of cloud system；The load information includes the first load sequence in chronological sequence sequentially formed
Column；
Judge in first load sequence with the presence or absence of the exception for deviateing the first load sequence overall variation rule
Load value；
If it exists, it according to the normal load value and preset first function relationship in first load sequence, calculates
Replace the replacement load value of the abnormal load value；
The replacement load value is replaced into the abnormal load value in first load sequence and forms the second load sequence；
Cloud management is carried out according to second load sequence.
In summary, a kind of cloud dynamic management device is present embodiments provided, is carried out down based on cloud system historic load value
The method of one moment cloud management, and for further optimum management, and the abnormal load value of historic load value is replaced
Processing is changed, management time delay is small, control efficiency is high, eliminates abnormal load value to have, so that management control is more smart
Really, in combination with by the processing to historic load value, using the cloud management method of dynamic self-adapting, have it is intelligent high,
The performance of cloud system is improved, and power-saving balanced property is good.
The foregoing is only a preferred embodiment of the present invention, is not intended to limit the scope of the present invention.
Claims (8)
1. a kind of cloud dynamic management approach, which is characterized in that the method includes：
The load information of cloud system is obtained, the load information includes the first load sequence；
Judge in first load sequence with the presence or absence of the abnormal load for deviateing the first load sequence overall variation rule
Value, the first load sequence overall variation rule are the variation continuity according to temporal load value；
If it exists, according to the normal load value and preset first function relationship in first load sequence, replacement is calculated
The replacement load value of the abnormal load value；
The replacement load value is replaced into the abnormal load value in first load sequence and forms the second load sequence；
Cloud management is carried out according to second load sequence；
Wherein, the preset first function relationship is：
It is describedTo replace jth in first load sequence1A load valueReplacement load value, and it is describedFor exception
Load value；
The ldkIt for k-th of load value in first load sequence, and is normal load value；
The ldlIt for first of load value in first load sequence, and is normal load value.
2. cloud dynamic management approach according to claim 1, which is characterized in that in judgement first load sequence
Include with the presence or absence of the abnormal load value for deviateing the first load sequence overall variation rule：
Calculate the positive change rate PR of first load sequencemAnd negative change rate NRm；
The load value for meeting following relationship in first load sequence is abnormal load value；
ldj-ldj-1>β×PRmAnd ldj+1-ldj<β×NRm
Or
ldj-ldj-1<β×NRmAnd ldj+1-ldj>β×PRm
The ldjFor j-th of load value in first load sequence；
The ldj-1For -1 load value of jth in first load sequence；
The ldj+1For+1 load value of jth in first load sequence；
The β is coefficient factor；
Wherein, the j is the positive integer greater than 1.
3. cloud dynamic management approach according to claim 1 or 2, which is characterized in that described according to the second load sequence
Column carry out cloud management：
According to second load sequence and preset second function relationship, computational load increment；
Administrative decision is obtained according to the incremental loading, and carries out cloud management by obtained administrative decision.
4. cloud dynamic management approach according to claim 3, which is characterized in that the second function relationship is：
Wherein, the aincmFor incremental loading；The ldm-x+1For the m-x+1 load value in second load sequence；
The ldm-xFor the m-x load value in second load sequence；The γ is attenuation coefficient；The n is the second load
The number of load value in sequence；The y is the positive integer less than n；
Administrative decision is obtained according to the incremental loading, and carries out cloud management by obtained administrative decision, is specifically included：
Work as aincm>0 andWhen, then by s in cloud system1A temperature state physical machine is converted into Warm status physical machine；
WhenWhen, then by s in cloud system2A cold state physical machine is converted into temperature state physical machine；
Work as aincm<0 andWhen, then by s in cloud system3A Warm status physical machine is converted into temperature state physical machine；
WhenWhen, then by s in cloud system4A temperature state physical machine is converted into cold state physical machine；
Wherein,
The α1For first threshold；The α2For second threshold, and 0<α1<α2；
The ldmx is preset maximum load value in cloud system；The ldmFor m-th of load value in the second load sequence；
The VhotFor the upper limit of Warm status physical machine can be provided in cloud system；
The VwarmFor the upper limit of temperature state physical machine can be provided in cloud system.
5. a kind of cloud dynamic management device, which is characterized in that described device includes：
Load information acquiring unit, to obtain the load information of cloud system；The load information includes the first load sequence；
Judging unit, to judge in first load sequence with the presence or absence of deviation the first load sequence overall variation rule
The abnormal load value of rule, the first load sequence overall variation rule are the variation continuity according to temporal load value；
Replace load value and form unit, in first load sequence there are when abnormal load value, according to described first
Normal load value and preset first function relationship in load sequence calculate the replacement load for replacing the abnormal load value
Value；
Second load sequence forms unit, the replacement load value is replaced the abnormal load in first load sequence
Value forms the second load sequence；
Administrative unit, to carry out cloud management according to second load sequence；
Wherein, the preset first function relationship is
It is describedTo replace jth in first load sequence1A load valueReplacement load value, and it is describedFor exception
Load value；
The ldkIt for k-th of load value in first load sequence, and is normal load value；
The ldlIt for first of load value in first load sequence, and is normal load value.
6. cloud dynamic management device according to claim 5, which is characterized in that the judging unit includes
Computing module, to calculate the positive change rate PR of first load sequencemAnd negative change rate NRm；
Judgment module to the judgement of abnormal load value, and meets the load value in first load sequence of following formula
For abnormal load value；
ldj-ldj-1>β×PRmAnd ldj+1-ldj<β×NRm
Or
ldj-ldj-1<β×NRmAnd ldj+1-ldj>β×PRm
The ldjFor j-th of load value in first load sequence；
The ldj-1For -1 load value of jth in first load sequence；
The ldj+1For+1 load value of jth in first load sequence；
The β is coefficient factor；
Wherein, the j is the positive integer greater than 1.
7. cloud dynamic management device according to claim 5, which is characterized in that the administrative unit includes：
Incremental loading computing module, to according to second load sequence and preset second function relationship, computational load
Increment；
Management module to obtain administrative decision according to the incremental loading, and carries out cloud management by obtained administrative decision.
8. cloud dynamic management device according to claim 7, which is characterized in that the second function relationship is：
Wherein, the aincmFor incremental loading；The ldm-x+1For the m-x+1 load value in second load sequence；
The ldm-xFor the m-x load value in second load sequence；The γ is attenuation coefficient；The y is；The n is
The number of load value in second load sequence；
The management module, specifically to work as aincm>0 andWhen, then by s in cloud system1A temperature state physics
Machine is converted into Warm status physical machine；WhenWhen, by s in cloud system2A cold state physical machine is converted into temperature
State physical machine；Work as aincm<0 andWhen, by s in cloud system3A Warm status physical machine is converted into temperature state object
Reason machine；WhenWhen, by s in cloud system4A temperature state physical machine is converted into cold state physical machine；
Wherein, the α1For first threshold；The α2For second threshold, and 0<α1<α2；
The ldmx is preset maximum load value in cloud system；The ldmFor m-th of load value in the second load sequence；
The VhotFor the upper limit of Warm status physical machine can be provided in cloud system；
The VwarmFor the upper limit of temperature state physical machine can be provided in cloud system.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410069284.XA CN104881271B (en) | 2014-02-27 | 2014-02-27 | Cloud dynamic management approach and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410069284.XA CN104881271B (en) | 2014-02-27 | 2014-02-27 | Cloud dynamic management approach and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104881271A CN104881271A (en) | 2015-09-02 |
| CN104881271B true CN104881271B (en) | 2018-11-23 |
Family
ID=53948780
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410069284.XA Active CN104881271B (en) | 2014-02-27 | 2014-02-27 | Cloud dynamic management approach and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104881271B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111683132B (en) * | 2020-06-04 | 2022-07-19 | 重庆金窝窝网络科技有限公司 | Business distribution method based on micro-service architecture and related device |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103748993B (en) * | 2009-05-31 | 2011-02-16 | 北京理工大学 | A kind of host load prediction method based on multisequencing combination |
| CN103442087A (en) * | 2013-09-12 | 2013-12-11 | 重庆大学 | Web service system access volume control device and method based on response time trend analysis |
| CN103607459A (en) * | 2013-11-21 | 2014-02-26 | 东北大学 | Dynamic resource monitoring and scheduling method of cloud computing platform IaaS layer |
-
2014
- 2014-02-27 CN CN201410069284.XA patent/CN104881271B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103748993B (en) * | 2009-05-31 | 2011-02-16 | 北京理工大学 | A kind of host load prediction method based on multisequencing combination |
| CN103442087A (en) * | 2013-09-12 | 2013-12-11 | 重庆大学 | Web service system access volume control device and method based on response time trend analysis |
| CN103607459A (en) * | 2013-11-21 | 2014-02-26 | 东北大学 | Dynamic resource monitoring and scheduling method of cloud computing platform IaaS layer |
Non-Patent Citations (1)
| Title |
|---|
| 基于时间序列的服务器负载预测;杨伟 等;《计算机工程》;20061031;第32卷(第19期);第143-148页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104881271A (en) | 2015-09-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhu et al. | A three-dimensional virtual resource scheduling method for energy saving in cloud computing | |
| CN104991830B (en) | YARN resource allocations and energy-saving scheduling method and system based on service-level agreement | |
| WO2018137402A1 (en) | Cloud data centre energy-saving scheduling implementation method based on rolling grey prediction model | |
| CN102004671B (en) | Resource management method of data center based on statistic model in cloud computing environment | |
| US8001403B2 (en) | Data center power management utilizing a power policy and a load factor | |
| Arshad et al. | Utilizing power consumption and SLA violations using dynamic VM consolidation in cloud data centers | |
| US9037880B2 (en) | Method and system for automated application layer power management solution for serverside applications | |
| US8230249B2 (en) | Dynamic selection of server states for servers in a cluster of servers | |
| CN105302630B (en) | A kind of dynamic adjusting method and its system of virtual machine | |
| CN103916438B (en) | Cloud testing environment scheduling method and system based on load forecast | |
| CN104035542B (en) | A kind of virtual machine migration method and system calculated with energy consumption for cooling balance | |
| CN104065745A (en) | Cloud computing dynamic resource scheduling system and method | |
| CN102759984A (en) | Power supply and performance management system for virtualization server cluster | |
| CN101719081A (en) | Method for scheduling virtual machines | |
| Sampaio et al. | Towards high-available and energy-efficient virtual computing environments in the cloud | |
| CN108287749A (en) | A kind of data center's total management system cloud resource dispatching method | |
| CN102339233A (en) | Cloud computing centralized management platform | |
| CN109005223A (en) | Internet of Things resource regulating method and system, computer readable storage medium and terminal | |
| Zhang et al. | An energy and SLA-aware resource management strategy in cloud data centers | |
| Mehta et al. | Energy conservation in cloud infrastructures | |
| CN103970256A (en) | Energy saving method and system based on memory compaction and CPU dynamic frequency modulation | |
| Li et al. | Load prediction-based automatic scaling cloud computing | |
| CN103617090A (en) | Energy saving method based on distributed management | |
| Yuan et al. | Energy aware resource scheduling algorithm for data center using reinforcement learning | |
| Singh et al. | Value and energy optimizing dynamic resource allocation in many-core HPC systems |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |