CN105302630B - A kind of dynamic adjusting method and its system of virtual machine - Google Patents
A kind of dynamic adjusting method and its system of virtual machine Download PDFInfo
- Publication number
- CN105302630B CN105302630B CN201510703546.8A CN201510703546A CN105302630B CN 105302630 B CN105302630 B CN 105302630B CN 201510703546 A CN201510703546 A CN 201510703546A CN 105302630 B CN105302630 B CN 105302630B
- Authority
- CN
- China
- Prior art keywords
- virtual machine
- host
- server
- server host
- dynamic
- 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
Landscapes
- Debugging And Monitoring (AREA)
Abstract
The present invention provides a kind of dynamic adjusting method of virtual machine and its systems.Wherein, the method includes：Obtain the server host in first and second operating condition；First operating condition is：The Overloaded Servers host of SLA rates of violation cannot be met；Second operating condition is：Load utilization is less than the unloaded server host of predetermined threshold value；Use the globally optimal solution for the virtual machine that the PSO Algorithm based on extremal optimization need to migrate.According to the optimal solution as a result, adjustment virtual machine；In scheduled time window, above-mentioned steps are repeated every the scheduled period to complete the dynamic adjustment of virtual machine.VMDA problems are efficiently solved by using a kind of innovatory algorithm having merged extremal optimization and population.
Description
Technical field
The present invention relates to field of cloud computer technology more particularly to the dynamic adjusting methods and its system of a kind of virtual machine.
Background technology
Cloud data center is usually made of the hardware facility of numerous high configurations.The computing capability of data center has become cloud
ISP's leading indicator of concern, with the appearance for the data center being getting larger and larger, the energy consumption of data center
It is increasing.Large number of high-performance hardware facility consumes a large amount of energy, and the CO2 emission of generation causes greenhouse
Effect generates climatic environment great influence, and largely affects the operation benefits of cloud service provider.Data center
Low energy utilization rate mainly due to server poor efficiency or it is idle when, but still consume huge energy.For example, having
Scholars, which count, finds that the energy of an idle server (be not turned off or switching state) consumption can often account for full load
70% or so of energy is consumed when operation.
In reality, the hardware facility in cloud data center not keeps static constant for a long time, on the contrary, most of hardware shape
State will may be changed often.First, new server may be added in system, before existing then to need to carry out weight
New configuration is repaired or is replaced；Secondly, resource pool may often change operating status to adapt to the money of elastic cloud environment intermittent
Source variation requires；Third, real-time migration (Live Migration, LM) technology make virtual machine (Virtual Machine, VM) exist
It can be realized in different physical nodes and quickly reconfigure and integrate, the targets such as load balance are realized with this；4th, some clothes
Device needs be engaged in virtual machine (vm) migration to other suitable server hosts, shutdown repair is needed to restart after completing dynamic migration
Operation so that this part server is unavailable within a certain period of time.Similar, certain server needs interim open to handle certain
A little uncertain access peak bursts.There is uncertainty in the access for being above server, in addition to this, in each service
There is also the possibility of various change inside device, for example, change server processing unit number (Processing Elements,
PEs), memory size, hard-disc storage, bandwidth etc..In addition, current server generally supports dynamic voltage frequency zoom technology
(Dynamic Voltage Frequency Scaling Technique, DVFS), server can be according to current load dynamics
Change voltage so as to adjust running frequency, reaches energy-efficient purpose.The dynamic and uncertainty of resource are necessary to adopt dynamic
Regulatory mechanism, the hardware resource of dynamic monitoring system detect and find the state change (Energy-aware) of resource, to there is new Shen
VM please rationally place and the VM to Service Level Agreement (Service Level Agreements, SLA) promise breaking or
VM in the extremely low server of cpu busy percentage optimizes configuration, and the configuration of entire VM is made to be optimal as possible.Dynamic regulation mechanism
It must be as far as possible in the efficient management for realizing automatic dynamic without the interference of administrative staff.
In cloud environment, to meet uncertain resource bid peak, usually there is supply (over- in data center
Provisioned thus) state, a large amount of energy waste generate.
Virtualization technology starts to be applied in data center at present, and system support is moved in real time between physical node
Move the virtual machine above it, to realize performance boost or it is energy saving the purpose of.When the resource of virtual machine actual use is less than distribution
When to its resource, virtual machine by adjusting and merge, reconfigure to other server nodes, the server that the free time gets off
Node is switched to energy saving mould by ACPI (Advanced Configuration and Power Interface) interface specification
Formula realizes the purpose for saving energy consumption.Cloud data center resource dispatching strategy focuses primarily upon lifting system performance at present, safeguards
SLA, and seldom from saving energy consumption angle.
DVFS is one of current realization energy-efficient main means of hardware facility.DVFS is the application journey run according to chip
The different of ordered pair computing capability need, and (for same chip, frequency is higher, needs for the running frequency and voltage of dynamic regulation chip
The voltage wanted is also higher), to reach energy-efficient purpose.Power can be reduced by reducing frequency, but merely reduce frequency simultaneously
Energy cannot be saved.Because for a given task, voltage only is reduced while reducing frequency, could veritably be dropped
The consumption of low energy.DVFS can implement depend on success prediction processor next need processing task number and when
Between.And usually in real-time system, clock frequency and voltage are not linear relationships, task execution time, energy expenditure, processing
There are prodigious uncertainties, inappropriate frequent voltage adjustment processor performance can be made to decline instead between device voltage.Big
In most cloud environments, task quantitative forecast is difficult to determine.
DVFS usually requires to carry out power management by BIOS, and the design circuit of different manufacturers has very big difference.For
There are one common power-management interface between operating system and hardware facility, with before improving on power management it is different
What the disunity interface that manufacturer is formulated was brought is difficult to compatibling problem, and the companies such as Intel, Microsoft, Toshiba make jointly
ACPI (Advanced Configuration and Power Interface) specification is determined.ACPI improves original pass through
BIOS carries out the pattern (APM) of power management, provides connecing for a more outstanding powder source management mode and configuration management
Mouth specification.ACPI defines most six kinds of power supply status, and different states corresponds to the energy consumption of different processors, memory and hard disk
Power and operating status.Currently, most processors all support several states such as operation, free time, suspend mode, closing.
Existing scholar is as follows about some progress of the above problem：
Rusu et al. proposes the managing power consumption strategy based on QoS guarantee for server cluster system.System is divided into rear end
Management and two modules of local management.Local management supports DVFS, when back end manager detects that system needs close or open
Some server, local management device controls power supply by DVFS modules, and server is switched to corresponding states.The system is not
Including real-time migration of virtual machine technology, server closing opening or not and the off-line calculation of rear end is depended on, it is limited to save energy consumption.
Also scholar proposes a kind of strategy of available energy dissipation management in distributed cloud computing system, author according to task at
Manage time and energy expenditure relationship, the object function of optimization be defined as relative superiority (Relative Superiority,
RS) expression formula calculates the RS values of task distribution on each server, and this first for will distributing for task
Business finally distributes to the maximum server of RS values.
But algorithm acquiescence Servers-all is in activation and good operating status, does not consider the isomery of system
Property, meanwhile, the assignment problem for newly increasing virtual machine is only considered in text, and real process also needs to consider SLA promise breaking virtual machines
The problems such as adjustment.
The scholars such as Kusic are sorting consistence problem managing power consumption problem definition under virtual isomerous environment and are surpassed using limited
Preceding control filtering (Limited Lookahead Control, LLC) is handled.Processing target is to realize maximum resource clothes
The profit of business supplier, while meeting the requirement of energy expenditure and SLA promise breaking minimums.System is estimated using Kalman filter
The quantity of future customers request and the state of forecasting system are counted, realizes necessary resource consolidation accordingly.However the system is difficult
Realized in IaaS cloud environment, and model too complex, for 15 nodes system every time adjust need 30 minutes when
Between, it is difficult to apply to extensive real-time cloud data center.
Verma et al. models the virtual machine Energy-aware Dynamic Arrangement problem under virtual isomerous environment, it is become
For continuous optimization problems：In each time frame, virtual machine places problem and can be considered that energy expenditure is minimum and performance is maximumlly excellent
Change problem, author solve the bin packing using heuritic approach, each time frame are completed with real-time migration strategy
VM is redistributed.In their follow-up work, they also use static policies (moon, year adjustment), semi-static strategy (day, week
Adjustment) and dynamic strategy (point, hour adjustment) carry out periodic adjustment.However, these algorithms do not consider SLA promise breaking problems：System
System performance can decline with the variation of load, and SLA cannot be guaranteed.
Berral et al. has studied VM Dynamic Integration problems, under the premise of meeting SLA, they using machine learning techniques come
Handle energy expenditure control problem.The processing mode only considered the application of some specific occasions, such as high-performance calculation (High
Performance Computing, HPC) etc. the application scenario with limited constraint, for common mixed load application scenario
And it is not suitable for.
Beloglazov et al. solves the problems, such as the dynamic allocation of VM using optimal adaptation degree sort descending algorithm (MBFD),
However the algorithm can not realize the ideal energy-efficient VM Dynamic Integrations of maximization.
It can be seen from the above, the resource management techniques scheme currently based on energy consumption saving is primarily present following problem：1. energy consumption
Saving technology only considers to save energy mostly, not very well in view of the promise breaking problem of SLA or the two cannot be simultaneous very well
It cares for；2. it is concentrated mainly on the assignment problem for newly increasing virtual machine, and for the operation conditions and correspondence of allocated virtual machine
Server load condition but consider seldom；3. as the VMDA problems of resource management core support technology, existing scheme is big
It mostly uses traditional heuritic approach to solve, solution efficiency is low.In face of the thousands of server host rule of data center
When mould, operation efficiency is insufficient, is extremely difficult to scheduling of resource target rapidly and efficiently.
Existing ideal scheduling (adjustment) aims at：It is realized under the premise of safeguarding SLA, ensureing system performance energy saving
Purpose.
It can realize that the system preferably dispatched can generally face following Railway Project：1. the excessive energy consumption of each server
Adjustment can reduce the stability of server operation；2. closing server in dynamic environment faces the wind that QoS cannot be guaranteed
Danger.Due to merging there are virtual machine and integrating, the service that certain virtual machines are run can not obtain enough in access peak period
Resource, to which the requirement of QoS cannot be met；3. in the virtualized environment that server host quantity is expanded, common scheduling is calculated
Method is it is difficult to ensure that quickly and effectively resource management scheduling, it is ensured that the application performance management design of system SLA has very big challenge.
Realize that the above several points requirement, core are the need for a set of efficiently quick Resource dynamic allocation algorithm, i.e. virtual machine dynamic point
With (Virtual Machine Dynamic Allocation, VMDA) algorithm.
While the algorithm requires system meet energy-saving and emission-reduction, realize that green calculates, also require external service performance excellent,
It can ensure the QoS that user specifies.
It is easy to prove, VMDA problems belong to NP (NP-complete) problems completely, not yet feasible more in the presence of one at present
Formula time algorithm solves np complete problem.All there is the mutually realization for influencing simultaneously decision problem computational complexity in np complete problem
As (phase transition)：All at least there is a control variable in all np complete problems, there are a phase transformations for the variable
Optimizing space is divided into two regions by point (or critical point), the point, and it is existing general can to meet solution for the attachment at area limit
Unexpected transformation has occurred in some critical point of control variable in rate.The phenomenon is known as the phase transition phenomena of np complete problem, in phase
The problem of at height, computation complexity was maximum.There are multiple such transformation temperatures by VMDA, and optimal solution is often located in transformation temperature
Near, therefore, optimizing algorithm such issues that solution must have the good ability for solving Phase-change Problems.
Invention content
Place in view of above-mentioned deficiencies of the prior art, the purpose of the present invention is to provide a kind of dynamic adjustment sides of virtual machine
Method and its system, it is intended to solve in the prior art virtual machine adjustment searching process in solution efficiency it is low, can not largely service
The problem of virtual machine dynamic allocation rapidly and efficiently are realized on device host.
In order to achieve the above object, this invention takes following technical schemes：
A kind of dynamic adjusting method of virtual machine, wherein the method includes：
Obtain the server host in first and second operating condition；
First operating condition is：The Overloaded Servers host of SLA rates of violation cannot be met；
Second operating condition is：Load utilization is less than the unloaded server host of predetermined threshold value；
It migrates in several virtual machines to other server hosts on the Overloaded Servers host；
It unloaded is taken by all virtual machine (vm) migrations to other servers on the unloaded server host, and by described
Business device mian engine changeover is energy saver mode；
Use the globally optimal solution for the virtual machine that the PSO Algorithm based on extremal optimization need to migrate；And
According to the optimal solution as a result, adjustment virtual machine；Also, in scheduled time window, every the scheduled period
Above-mentioned steps are repeated to complete the dynamic adjustment of virtual machine.
The dynamic adjusting method of the virtual machine, wherein the particle cluster algorithm based on extremal optimization is specially：
In the particle cluster algorithm iteration renewal process, extremal optimization local search algorithm is incorporated with probabilistic manner.
The dynamic adjusting method of the virtual machine, wherein in the particle cluster algorithm based on extremal optimization, each group
Member is corresponding with the virtual machine of need migration；Corresponding fitness is assigned for each constituent element, and selects the group of fitness minimum
Member and its neighbours are into row variation.
The dynamic adjusting method of the virtual machine, wherein the fitness of the constituent element is calculated especially by following formula：
λiFitness, h (i) for i constituent elements are the destination server host of virtual machine i distribution；Utlz (h (i)) is destination service
The processor utilization of device host h (i)；Utlz (i) is the processor utilization of virtual machine i；α and β is weight parameter；C1 and c2
For weight primary constants.
The dynamic adjusting method of the virtual machine, wherein the method further includes：Servers-all host is traversed, it will
The virtual machine newly increased, which is assigned to, disclosure satisfy that in SLA rates of violation and the most server host of saving energy consumption.
A kind of dynamic debugging system of virtual machine, wherein the system comprises：
Server host operating condition acquisition module, for obtaining the server master in first and second operating condition
Machine；First operating condition is：The Overloaded Servers host of SLA rates of violation cannot be met；Second operating condition is：It is negative
Carry the unloaded server host that utilization rate is less than predetermined threshold value；
Virtual machine (vm) migration module, for migrating several virtual machines on the Overloaded Servers host to other servers master
In machine and by all virtual machine (vm) migrations to other servers on the unloaded server host, and unloaded taken described
Business device mian engine changeover is energy saver mode；
Computing module is migrated, the overall situation for using virtual machine that the PSO Algorithm based on extremal optimization need to migrate
Optimal solution；And
Virtual machine (vm) migration module is additionally operable to, according to the optimal solution as a result, adjustment virtual machine；
Adjust cycle module, in scheduled time window, every the scheduled period repeat above-mentioned steps to
Complete the dynamic adjustment of virtual machine.
The dynamic debugging system of the virtual machine, wherein the migration computing module is specifically used for：In the population
In algorithm iteration renewal process, extremal optimization local search algorithm is incorporated with probabilistic manner.
The dynamic debugging system of the virtual machine, wherein the migration computing module is specifically used for：It is based on pole described
It is worth in the particle cluster algorithm of optimization, each constituent element is corresponding with the virtual machine that a need migrate；It is assigned for each constituent element corresponding
Fitness, and select fitness minimum constituent element and its neighbours into row variation.
The dynamic debugging system of the virtual machine, wherein the migration computing module is specifically used for：Pass through following formula
Calculate the fitness of the constituent element：
λiFitness, h (i) for i constituent elements are the destination server host of virtual machine i distribution；Utlz (h (i)) is target
The processor utilization of server host h (i)；Utlz (i) is the processor utilization of virtual machine i；α and β is weight parameter；c1
It is weight primary constants with c2.
The dynamic debugging system of the virtual machine, wherein the system also includes：Assignment module, it is all for traversing
User is newly applied for that increased virtual machine is assigned to the clothes that disclosure satisfy that SLA rates of violation and most save energy consumption by server host
It is engaged in device host.
Advantageous effect：The dynamic adjusting method and its system of a kind of virtual machine provided by the invention, have used a kind of fusion
Extremal optimization and the innovatory algorithm of population efficiently solve VMDA problems, to realizing virtual machine in a large amount of servers
Dynamic adjustment in host can realize energy saving maximum while effectively ensureing the customer service quality level of data center
The purpose of change.
Description of the drawings
Fig. 1 is the method flow diagram of the dynamic adjusting method of the virtual machine of the specific embodiment of the invention.
Fig. 2 is the Scheduling Framework schematic diagram of the dynamic adjustment of the virtual machine of the specific embodiment of the invention.
Fig. 3 is the structure diagram of the dynamic debugging system of the virtual machine of the specific embodiment of the invention.
Fig. 4 is that the dynamic adjusting method using the virtual machine of the specific embodiment of the invention and the energy consumption of other algorithms disappear
Consume contrast schematic diagram.
Specific implementation mode
The present invention provides a kind of dynamic adjusting method and its system of virtual machine.To make the purpose of the present invention, technical solution
And effect is clearer, clear, the present invention is described in more detail for the embodiment that develops simultaneously referring to the drawings.It should be appreciated that this
Place is described, and specific examples are only used to explain the present invention, is not intended to limit the present invention.
As shown in Figure 1, the dynamic adjusting method of the virtual machine for the specific embodiment of the invention.The method includes：
S1, the server host in first and second operating condition is obtained.
Wherein, first operating condition is：The Overloaded Servers host of SLA rates of violation cannot be met；Second fortune
Market condition is：Load utilization is less than the unloaded server host of predetermined threshold value.For Overloaded Servers host, need to migrate institute
It states in several virtual machines to other server hosts on Overloaded Servers host.
And for unloaded server host, then it needs all virtual machine (vm) migrations on the unloaded server host arriving it
In his server, and the unloaded server host is switched to energy saver mode (specific as stated in the background art, such as reduces
Server host operation frequency etc.).
In the application of actual cloud data center, for Servers-all host, it is not up to standard that adjustable strategies only select SLA
(being unsatisfactory for rate of violation requirement) server host and the server host of load too low execute VM and integrate, to reduce VM real-time migrations
Quantity.
The server host of SLA promise breakings is selected in strategy, presses formula (1) dynamic adjustment processor utilization upper limit threshold first
Value is so that server service level and the SLA of customization match.
Wherein, upper_th is processor utilization upper threshold when current server meets SLA；RSLAFor server
Host object SLAs；rSLAIt is an increment constant for the currently practical SLAs of server host, τ.
In the present embodiment example, estimate new processor threshold value approximation direct ratio with present threshold value and target SLA and practical SLA
Product between ratio, and new threshold value is substituted old threshold value.It is worth noting that upper_th values cannot infinitely reduce or
Increase (is set as in the present embodiment：
upper_th∈[0.5,0.95])。
Then the SLA rates of violation of current server host are assessed and judge whether to meet specified SLA requirements, if discontented
It is sufficient then select part VM by formula (2) and migration list is added and prepares they migrating out host.
Wherein, hvs is that the virtual machine that current server host undertakes is all；Utlz (h) is that current hosts processor is overall
Utilization rate is horizontal, and Utlz (v) is the average utilization that virtual machine v occupies current processor within the scope of time window.
The present embodiment selects to need to migrate out in Overloaded Servers using the strategy for minimizing virtual machine (vm) migration quantity
Virtual machine, keep the virtual machine quantity that needs migrate minimum, bring system performance to decline when it is possible thereby to reduce virtual machine (vm) migration
The problem of.
After virtual machine (vm) migration is gone out, the sum of the processor utilization that preceding server host carries remaining virtual machine is less than
Defined threshold value, to reach the requirement for meeting regulation SLA.
Specific migration position for the virtual machine in the server host of first and second operating condition, migrated
It sets, can be solved and be obtained by step S2 operations.
In actual application, meet the important requirement that QoS is cloud computing system.QoS is usually presented by SLA.
SLA typically specifies the indexs such as system computing capacity, reaction time, system maximum visit capacity, and different application service institutes is right
These indexs answered have prodigious difference again.
Specifically, some server host SLA rates of violation (SLA in certain time window can be defined by formula (4)
Violation,SLAV)：
Previous item is the SLA violations caused by server host processor overload operation in formula (4), wherein
TvThe duration of 100% utilization rate, T are undergone for current server host in time windowaFor time window size, when host reaches 100%
Host overload operation, α are normalized parameter at this time for representative when utilization rate.It is transported at full capacity due to usually reaching in server host
When row, it is difficult to timely respond to new access application, SLA promise breakings is caused to occur.
Latter is the SLA promise breakings caused by performance caused by real-time migration declines in formula (4), because in real-time migration
In the process, the access performance for needing the service run on the VM migrated external can be declined, wherein CvSpent for migration
Processor resource (MIPS), usual CvValue account for 10% or so, C of real-time migration process controller total resourcesaFor virtual machine application
Total processor resource (MIPS).
Also that is, as shown in Fig. 2, using virtual machine of the present invention dynamic adjusting method Scheduling Framework specifically such as
Under：
User is to cloud system application VM resources, and specified SLA.Scheduling Framework is required according to application and server host is transported
Row situation is arranged to the VM of application on suitable server host.(it is worth noting that, to ensure response speed, the VM's
Instrumentation is completed within the very fast time)
Scheduling Framework monitors hardware facility (such as a large amount of server host of data center) operation conditions in real time, one
Window of fixing time is interior to carry out dynamic VM integration according to the operating condition of host.The time window specifically can be according to actual use feelings
Condition is determined.
Dynamic Integration is carried out for the server of following two situations：(1) for the server host of SLA cannot be met,
It is horizontal to reduce server host processor utilization upper threshold, certain (a) VM selected on the server host move to it
Its server reduces its load, it is ensured that SLA is ensured；(2) the server master for utilization rate load less than certain lower limit
Whole VM on the server are moved to other servers, and the server are switched to energy saving (suspend mode) state by machine, are realized
Energy saving purpose.
The globally optimal solution of S2, the virtual machine that need to be migrated using the PSO Algorithm based on extremal optimization.
In the specific implementation of the present invention, the particle cluster algorithm based on extremal optimization is specially：In the population
Extremal optimization local search algorithm is incorporated with probabilistic manner in algorithm iteration renewal process.
Existing basic particle group algorithm ability of searching optimum is fine, but its algorithmic statement efficiency is not high.To improve grain
The efficiency of evolution (convenient for being applied on a large amount of server hosts) of swarm optimization keeps sample diversity, can ensure centainly
Better local search technique is incorporated under the premise of sample space is multifarious to improve algorithm evolution efficiency.
More specifically, in the particle cluster algorithm based on extremal optimization, the virtual machine of each constituent element and the migration of a need
It is corresponding；Assign corresponding fitness for each constituent element, and select fitness minimum constituent element and its neighbours into row variation.
Particle cluster algorithm (PSO) is a kind of group hunting evolution algorithm based on mould because of evolution.It is according to flock of birds airflight
When, every bird constantly self-position can be adjusted according to the position of bird around with the principle for the position being optimal and
A kind of bionic Algorithm put forward.The particle cluster algorithm includes the following important feature：
(1) in PSO, each particle (bird) will carry out mould in their flights because evolving.Each particle flies according to itself
The flight path of row track and companion dynamically adjust its state of flight.
(2) PSO constantly updates several particles randomly generated, in iteration renewal process, always leads to
Cross the optimal solution p that particle itself is foundidThe optimal solution p found by the end of current iteration with entire populationgdThe two " extreme values "
Each particle is updated, in some occasions, entire population can also be used as with part neighbours, in neighbours region
Extreme value is local extremum.
(3) in current kth time iteration, a certain particle i once finds the two optimal values, is updated according to following formula
The speed of oneself and new position：
In formula (5) and (6),Speed when iteration secondary for particle i kth；W is inertia weight；It is current particle
Position；pidFor the optimal solution found by particle itself, pgdFor the optimal solution found by entire population, r1And r2It is between 0
And the random number between 1；c1And c2It is Studying factors.
Wherein, w is great for the convergence effect of particle cluster algorithm.W values are bigger, then global optimizing ability is stronger, part
Optimizing ability is weaker.Conversely, then local optimal searching ability enhances, and global optimizing ability weakens.
It can become by adjusting w size to control influence of the former speed to present speed and take into account the overall situation and search
One compromise of rope and local search.Therefore, in practical application, the algorithm starting stage, w was larger, and w is gradual with iterations going on
Become smaller.Since w is big, then speed v is just big, is conducive to the space of particle search bigger, it may be found that new solution domain；And w is smaller, then
Speed w is just small, is conducive to excavate preferably solution in current solution space
The pseudocode of the above-mentioned particle cluster algorithm based on extremal optimization is specific as follows shown：
It is described based on the particle cluster algorithm of extremal optimization evolve when only there are one by several group of components at individual, each
Constituent element corresponds to each component of solution vector.The algorithm thinks that each internal constituent element is different to the contribution of individual good and bad degree
, and fitness is assigned for each constituent element to the contribution of individual goal functional value according to each constituent element, fitness is minimum
Constituent element be worst constituent element.The each iteration of extremal optimization always selects worst constituent element and its neighbours into row variation.
The basic framework of the particle cluster algorithm based on extremal optimization is as described below：
First, an individual X=(x are randomly generated1,x2,...,xt), if optimal solution is Xbest=X, target function value are denoted as C
(X)；
Then, to individual X, following operation is executed：
A. evaluation constituent element xiFitness and be denoted as λi, i ∈ { 1,2 ..., t }；
B. t constituent element is ranked up according to fitness size, finds out the worst constituent element x of fitnessj, i.e. λj≤λi, i=
1,2 ..., t, then xjAs worst constituent element；
C. a neighbours X' is selected in the neighborhood of X, by variation so that worst constituent element xjIt changes；
D. more new individual X=X'；
If e. C (X)<C(Xbest), then Xbest=X；
Then the step of fitness worst constituent element, is found out in repetitions, until end condition meets；
Last returns to gained optimal solution Xbest。
In the specific embodiment of the dynamic adjusting method of virtual machine of the present invention, the fitness of the constituent element is specific
It is calculated by following formula：
λiFitness, h (i) for i constituent elements are the destination server host of virtual machine i distribution；Utlz (h (i)) is target
The processor utilization of server host h (i)；Utlz (i) is the processor utilization of virtual machine i；α and β is weight parameter；c1
It is weight primary constants with c2.
It is preferred that for maintenance algorithm global search, avoid being absorbed in local optimum, when selection needs the constituent element to make a variation,
It is selected by the probability distribution of power-law according to fitness sequence.If shared t needs the VM migrated, each VM (constituent element)
It chooses probability and is obeyed from big to small by constituent element fitness value and is distributed as shown in formula (8)：
p(r)∝r-τ1≤r≤t,τ≥0 (8)
S3, according to the optimal solution as a result, adjustment virtual machine.
S4, in scheduled time window, repeat above-mentioned steps every the scheduled period to complete the dynamic of virtual machine
State adjusts.
The scheduled period can be determined according to actual conditions.In specific implementation process, due in cloud data
Each server host software and hardware of the heart is all in the service in continually changing environment, run on the VM of each server carrying
It is also at continually changing state, Scheduling Framework real-time perception host operation conditions, and with window period certain time according to host
Operation conditions carry out dynamic VM integration.
Specifically, as shown in Figure 1, newly applying for that increased virtual machine, the method further include for user：
S5, traversal Servers-all host, the virtual machine newly increased is assigned to and disclosure satisfy that SLA rates of violation and most save
About in the server host of energy consumption.
Corresponding, in above-mentioned Scheduling Framework, whenever data center all allows have user to propose new service Shen
Please (VM), after reaching SLA, cloud data center is assigned to suitable server host (Host) to its quick response and VM.
The scheduling strategy for the virtual machine newly applied for user is：It will newly apply for that increased VM is assigned to meet SLA and can most save energy
On the server host of consumption.It need to only traverse a server host list and can obtain a result, and have the advantages that quick response.
In terms of the specific calculating of energy consumption, the server host energy expenditure of existing cloud data center is mainly by CPU, interior
It deposits, the consumption of the hardware modules such as disk and refrigeration system is constituted.As multi-core CPU is more and more universal, the energy expenditure of CPU accounts for
Major part, it is existing researches show that：Server energy expenditure can approximately linear direct ratio with the energy expenditure of CPU, and one
The energy of the server consumption of a free time can account for consuming the 70% of energy when full load operation.
In addition, we also need to consider virtual machine (vm) migration when energy expenditure, real time virtual machine migrating technology permission taking
Virtual machine is reset to fast and flexible between business device host, and hang-up need not be serviced and can be completed.In real-time migration technology, virtually
The image file and data real time backup of machine at network attached storage (Network Attached Storage, NAS), because
This, when migration, does not need to copy virtual machine itself, it is only necessary to copy virutal machine memory.The virtual machine (vm) migration time is memory
Size divided by network bandwidth.However, real time virtual machine transition process still can be next unfavorable to the service band run on virtual machine
It influences, existing research has been discussed in detail influence of the real-time migration technology to system performance and has carried out mathematical modeling, it is indicated that
Performance impairment depends primarily on virtual in-fight service memory pages newer quantity when time and the migration of migration.For usual
Application service, for example web page server, real-time migration process about expend 10%CPU utilization rates.
Defining unit interval self-energy consumption power as a result, can be specifically defined as by following formula：
In formula (3), Emax(h) energy expenditure power when being run for server host h full load, Utlz (h) are unit
Time server host-processor average utilization, v are the virtual machine set migrated in unit time window, and T (i) is virtual machine i
Transit time.
As shown in figure 3, a kind of dynamic debugging system of virtual machine for the specific embodiment of the invention, wherein the system
Including：
Server host operating condition acquisition module 100, for obtaining the server in first and second operating condition
Host；First operating condition is：The Overloaded Servers host of SLA rates of violation cannot be met；Second operating condition is：
Load utilization is less than the unloaded server host of predetermined threshold value.
Virtual machine (vm) migration module 200, for migrating several virtual machines on the Overloaded Servers host to other services
In device host and by all virtual machine (vm) migrations to other servers on the unloaded server host, and by the sky
It carries server host and is switched to energy saver mode.
Computing module 300 is migrated, for using virtual machine that the PSO Algorithm based on extremal optimization need to migrate
Globally optimal solution；And
Virtual machine (vm) migration module 300 is additionally operable to, according to the optimal solution as a result, adjustment virtual machine.
Adjust cycle module 400, in scheduled time window, every the scheduled period repeat above-mentioned steps from
And complete the dynamic adjustment of virtual machine.
Specifically, for the virtual machine that user newly applies, the system also includes：Assignment module 500, it is all for traversing
User is newly applied for that increased virtual machine is assigned to the clothes that disclosure satisfy that SLA rates of violation and most save energy consumption by server host
It is engaged in device host.
In a specific embodiment of the present invention, the migration computing module 300 is specifically used for：It changes in the particle cluster algorithm
For in renewal process, extremal optimization local search algorithm is incorporated with probabilistic manner.
More specifically, the migration computing module 300 is specifically used for：In the particle cluster algorithm based on extremal optimization
In, each constituent element is corresponding with the virtual machine that a need migrate；Corresponding fitness is assigned for each constituent element, and selects to adapt to
Minimum constituent element and its neighbours are spent into row variation.
The migration computing module is specifically used for：The fitness of the constituent element is calculated by following formula：
Wherein, λiFitness, h (i) for i constituent elements are the destination server host of virtual machine i distribution；Utlz (h (i)) is
The processor utilization of destination server host h (i)；Utlz (i) is the processor utilization of virtual machine i；α and β joins for weight
Number；C1 and c2 is weight primary constants.As detailed above.
Embodiment 1：
Emulation experiment：
Document HPCS 2009, ISBN (is originated from using cloud computing emulation platform CloudSim toolkit:978-1-4244-
49071) it is emulated.
CloudSim is the grid experiment room of Univ Melbourne Australia and the cloud that Gridbus projects were released in 2009
Computer sim- ulation platform, it inherits the programming model of GridSim, supports the research and development of cloud computing, provides cloud computing
Characteristic supports resource management and the dispatching simulation of cloud computing.The CIS (Cloud Information Service) of CloudSim
Resource discovering and information exchange are realized with DataCenterBroker, are the cores of operation simulation.The independently developed scheduling of user
Algorithm can be realized in the method for DataCenterBroker, to realize the simulation of dispatching algorithm.Latest edition has been propped up at present
The simulated experiment of server host Energy-aware is held, this programme tests algorithm using CloudSim platforms.
The emulation cloud platform of test is configured to 1000 server hosts, including 500 ProLiant DL360 G4p masters
Machine (is configured to 3400MHz*2core, 6GB memory, 1GB network bandwidths), and 500 ProLiant ML110 G3 hosts (are configured to
3000MHz*2core, 4GB memory, 1GB network bandwidths).The energy expenditure of every server is calculated according to formula (3).
According to the SPECpower benchmark test statistical averages of fourth quarter in 2010 as a result, by EmaxValue be set as 259Wh.
The virtual machine task load data source of this emulation experiment is in CoMon projects.The project belongs to PlanetLab monitorings
A part for facility.Data are more than mainly that the upper CPU of 1000 virtual machines is utilized comprising 500 different places that spread all over the world
Rate, these data were obtained every detection in 5 minutes.
Select measured data on March 3rd, 2011 as final test data.It is 5 minutes to adjust the period.When experiment, test
Each VM loads on collection are applied as new VM, need in data center's distribution resource and operation service.Using institute of the present invention
The Scheduling Framework stated loads all VM and carries out intelligent scheduling.The period VM dynamic executed at regular intervals adjusts, and adjusts every time
The continuous G=250 iteration of optimal solution is not improved and then thinks to reach the condition of convergence when spending, and algorithm exits.Each scheduling scheme weight
Running 10 times again takes its average value as the performance of evaluation algorithms.
Mainly it is compared from data center's SLA rates of violation (SLAV) and energy expenditure (Energy) angle.Experiment knot
Fruit is as shown in Table 1：
1 several scheduling strategy results contrasts of table
Wherein, first it is classified as target SLAV, experiment tests SLAV values from 1% to 5% (× E-3) all kinds of algorithms respectively
Performance.Secondary series LTH is the minimum utilization rate of server host, and every group of SLAV value all includes the corresponding host profit from 0.1 to 0.5
With rate.
In algorithm for comparing, the first NPA is that non-energy perceives scheduling strategy, allows Servers-all master under the strategy
Machine is run under ceiling capacity consumption patterns；Second is DVFS patterns, under the pattern to VM without any resetting integration at
Reason；The third strategy is the corresponding adjustable strategies of virtual machine dynamic adjusting method of the present invention.
Each algorithm is tested under a variety of different SLAV and LTH combinations, obtains energy expenditure (kWh) and virtual machine
Migrate the value of number (Migr.).
As shown in Table 1, the energy expenditure of cloud data center can significantly be reduced using Energy-aware strategy.Using this hair
Energy expenditure only accounts for 3% or so of NPA strategies when the bright described VM allocation strategies, is greatly saved the consumption of energy.
Meanwhile in this experiment test, as the minimum 86kWh of energy consumption, the SLA of 5% × E-3 still can be kept to break a contract
Rate, i.e. system still can keep extraordinary service quality while saving energy.Therefore, of the present invention to be based on extreme value
The particle cluster algorithm of optimization, the scheduling that population is can further improve using the good local searching strategy of extremal optimization technology are imitated
Rate realizes preferably energy saving purpose in the case where virtual machine (vm) migration number is not much different.Such as in SLAV=5% × E-3,
When LTH=0.3, population extremal optimization scheme only expends the lowest power consumption of 86kWh.
On the other hand, as shown in Table 1, LTH generally can not be too small, though LTH is too small avoidable too much server master
Machine-cut enters dormant state, but there is also the excessively high problems of energy consumption；Meanwhile LTH values cannot be too big, LTH too conferences cause server
Host frequently switches between suspend mode and wake-up states, causes additional energy expenditure, if LTH is too big until approaching Upper
Threshold, or even the problem of can cause virtual machine that can not distribute.The usual comparatively ideal values of LTH should be set as between 0.2~0.4.
Document " Beloglazov A., Buyya R..Optimal Online Deterministic Algorithms
and Adaptive Heuristics for Energy and Performance Efficient Dynamic
Consolidation of Virtual Machines in Cloud Data Centers[J].Concurrency and
Computation:Practice and Experience,2011,0:1-24 " describes a kind of Energy-aware algorithm to handle
The scheduling problem of cloud data center, the algorithm select the virtual machine for needing to migrate using the strategy such as minimum transition time (MMT),
And using MBFD algorithms come the placement problem of processing virtual machine.
The algorithm emulates certain incoming task under using optimal MMT strategies under CloudSim platforms
Simulation, it is 87.67kWh to obtain least energy consumption, and the SLAV obtained is 4.65% × E-3.
The comparison of above-mentioned algorithm and algorithm of the present invention：
Algorithm target SLAV is set as 4.65% by (test platform is identical with task input condition) under the same test conditions
× E-3 is tested.As shown in figure 4, under identical SLAV levels, 82.35 are only needed using algorithm of the present invention
Energy expenditure can achieve the purpose that preferably to save energy consumption.
It, can according to the technique and scheme of the present invention and this hair it is understood that for those of ordinary skills
Bright design is subject to equivalent substitution or change, and all these changes or replacement should all belong to the guarantor of appended claims of the invention
Protect range.
Claims (4)
1. a kind of dynamic adjusting method of virtual machine, which is characterized in that the method includes：
Obtain the server host in first and second operating condition；
First operating condition is：The Overloaded Servers host of SLA rates of violation cannot be met；
Second operating condition is：Load utilization is less than the unloaded server host of predetermined threshold value；
It migrates in several virtual machines to other server hosts on the Overloaded Servers host；
By in all virtual machine (vm) migrations to other servers on the unloaded server host, and by the unloaded server
Mian engine changeover is energy saver mode；
Use the globally optimal solution for the virtual machine that the PSO Algorithm based on extremal optimization need to migrate；And according to it is described most
Excellent solution as a result, adjustment virtual machine；
In scheduled time window, above-mentioned steps are repeated every the scheduled period to complete the dynamic adjustment of virtual machine；
The particle cluster algorithm based on extremal optimization is specially：
In the particle cluster algorithm iteration renewal process, extremal optimization local search algorithm is incorporated with probabilistic manner；
In the particle cluster algorithm based on extremal optimization, each constituent element is corresponding with the virtual machine that a need migrate；
Corresponding fitness is assigned for each constituent element, and
Select fitness minimum constituent element and its neighbours into row variation；
The fitness of the constituent element is calculated especially by following formula：
λiFitness, h (i) for i constituent elements are the destination server host of virtual machine i distribution；Utlz (h (i)) is destination server
The processor utilization of host h (i)；Utlz (i) is the processor utilization of virtual machine i；α and β is weight parameter；C1 and c2 are
Weight primary constants.
2. the dynamic adjusting method of virtual machine according to claim 1, which is characterized in that the method further includes：
Servers-all host is traversed, the virtual machine newly increased is assigned to and disclosure satisfy that SLA rates of violation and most save energy consumption
Server host in.
3. a kind of dynamic debugging system of virtual machine, which is characterized in that the system comprises：
Server host operating condition acquisition module, for obtaining the server host in first and second operating condition；
First operating condition is：The Overloaded Servers host of SLA rates of violation cannot be met；
Second operating condition is：Load utilization is less than the unloaded server host of predetermined threshold value；
Virtual machine (vm) migration module, for migrating in several virtual machines to other server hosts on the Overloaded Servers host
And by all virtual machine (vm) migrations to other servers on the unloaded server host, and by the unloaded server
Mian engine changeover is energy saver mode；
Computing module is migrated, the global optimum for using virtual machine that the PSO Algorithm based on extremal optimization need to migrate
Solution；And
Virtual machine (vm) migration module is additionally operable to, according to the optimal solution as a result, adjustment virtual machine；
Cycle module is adjusted, in scheduled time window, above-mentioned steps to be repeated to complete every the scheduled period
The dynamic of virtual machine adjusts；
The migration computing module is specifically used for：
In the particle cluster algorithm iteration renewal process, extremal optimization local search algorithm is incorporated with probabilistic manner；
The migration computing module is specifically used for：
It is in the particle cluster algorithm based on extremal optimization, each constituent element is corresponding with the virtual machine that a need migrate；
Corresponding fitness is assigned for each constituent element, and
Select fitness minimum constituent element and its neighbours into row variation；
The migration computing module is specifically used for：The fitness of the constituent element is calculated by following formula：
λiFitness, h (i) for i constituent elements are the destination server host of virtual machine i distribution；Utlz (h (i)) is destination server
The processor utilization of host h (i)；Utlz (i) is the processor utilization of virtual machine i；α and β is weight parameter；C1 and c2 are
Weight primary constants.
4. the dynamic debugging system of virtual machine according to claim 3, which is characterized in that the system also includes：It assigns
User is newly applied for that increased virtual machine is assigned to and disclosure satisfy that SLA rates of violation simultaneously by module for traversing Servers-all host
And it most saves in the server host of energy consumption.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510703546.8A CN105302630B (en) | 2015-10-26 | 2015-10-26 | A kind of dynamic adjusting method and its system of virtual machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510703546.8A CN105302630B (en) | 2015-10-26 | 2015-10-26 | A kind of dynamic adjusting method and its system of virtual machine |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105302630A CN105302630A (en) | 2016-02-03 |
| CN105302630B true CN105302630B (en) | 2018-07-13 |
Family
ID=55199924
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510703546.8A Active CN105302630B (en) | 2015-10-26 | 2015-10-26 | A kind of dynamic adjusting method and its system of virtual machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105302630B (en) |
Families Citing this family (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106936905B (en) * | 2017-03-07 | 2020-04-28 | 中国联合网络通信集团有限公司 | Openstack-based scheduling method and scheduling system of Nova component virtual machine |
| CN108306780B (en) * | 2017-09-07 | 2021-07-20 | 上海金融期货信息技术有限公司 | Cloud environment-based virtual machine communication quality self-optimization system and method |
| CN107861802A (en) * | 2017-10-24 | 2018-03-30 | 郑州云海信息技术有限公司 | Method for managing resource and device in cloud data system |
| CN108469983B (en) * | 2018-04-02 | 2021-10-19 | 西南交通大学 | Particle swarm algorithm-based virtual machine deployment method in cloud environment |
| CN110362388B (en) * | 2018-04-11 | 2021-08-31 | 中移(苏州)软件技术有限公司 | Resource scheduling method and device |
| CN109491760B (en) * | 2018-10-29 | 2021-10-19 | 中国科学院重庆绿色智能技术研究院 | High-performance data center cloud server resource autonomous management method |
| CN109766190A (en) * | 2019-01-15 | 2019-05-17 | 无锡华云数据技术服务有限公司 | Cloud resource dispatching method, device, equipment and storage medium |
| CN109960568A (en) * | 2019-02-18 | 2019-07-02 | 深圳大学 | A kind of dispatching method and electronic equipment |
| CN109901932B (en) * | 2019-03-12 | 2023-04-07 | 东北大学 | Server integration method based on virtual machine |
| CN113448718A (en) * | 2020-03-26 | 2021-09-28 | 安徽寒武纪信息科技有限公司 | Method, apparatus and computer-readable storage medium for frequency modulation of a chip |
| CN112395161A (en) * | 2020-11-26 | 2021-02-23 | 国网天津市电力公司 | Big data center energy consumption analysis method and computing equipment |
| CN113821313A (en) * | 2020-12-28 | 2021-12-21 | 京东科技控股股份有限公司 | Task scheduling method and device and electronic equipment |
| CN113075995B (en) * | 2021-04-26 | 2024-02-27 | 华南理工大学 | Virtual machine energy-saving integration method, system and storage medium based on hybrid group intelligence |
| CN116881085A (en) * | 2023-09-05 | 2023-10-13 | 北京华鲲振宇智能科技有限责任公司 | Method for optimizing energy consumption of server |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1764127A (en) * | 2004-10-18 | 2006-04-26 | 国际商业机器公司 | Method for distributing resource for high-grade user and data processor |
| CN101504620A (en) * | 2009-03-03 | 2009-08-12 | 华为技术有限公司 | Load balancing method, apparatus and system of virtual cluster system |
| CN103488539A (en) * | 2013-09-23 | 2014-01-01 | 北京交通大学 | Data center energy saving method based on central processing unit (CPU) dynamic frequency modulation technology |
| CN103576829A (en) * | 2012-08-01 | 2014-02-12 | 复旦大学 | Hybrid genetic algorithm based dynamic cloud-computing virtual machine scheduling method |
| CN103699446A (en) * | 2013-12-31 | 2014-04-02 | 南京信息工程大学 | Quantum-behaved particle swarm optimization (QPSO) algorithm based multi-objective dynamic workflow scheduling method |
| CN104679594A (en) * | 2015-03-19 | 2015-06-03 | 成都艺辰德迅科技有限公司 | Middleware distributed calculating method |
| CN104793993A (en) * | 2015-04-24 | 2015-07-22 | 江南大学 | Cloud computing task scheduling method of artificial bee colony particle swarm algorithm based on Levy flight |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7437730B2 (en) * | 2003-11-14 | 2008-10-14 | International Business Machines Corporation | System and method for providing a scalable on demand hosting system |
| US7650464B2 (en) * | 2007-03-26 | 2010-01-19 | Intel Corporation | Object relocation guided by data cache miss profile |
| US20150074222A1 (en) * | 2013-09-12 | 2015-03-12 | Guanfeng Liang | Method and apparatus for load balancing and dynamic scaling for low delay two-tier distributed cache storage system |
-
2015
- 2015-10-26 CN CN201510703546.8A patent/CN105302630B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1764127A (en) * | 2004-10-18 | 2006-04-26 | 国际商业机器公司 | Method for distributing resource for high-grade user and data processor |
| CN101504620A (en) * | 2009-03-03 | 2009-08-12 | 华为技术有限公司 | Load balancing method, apparatus and system of virtual cluster system |
| CN103576829A (en) * | 2012-08-01 | 2014-02-12 | 复旦大学 | Hybrid genetic algorithm based dynamic cloud-computing virtual machine scheduling method |
| CN103488539A (en) * | 2013-09-23 | 2014-01-01 | 北京交通大学 | Data center energy saving method based on central processing unit (CPU) dynamic frequency modulation technology |
| CN103699446A (en) * | 2013-12-31 | 2014-04-02 | 南京信息工程大学 | Quantum-behaved particle swarm optimization (QPSO) algorithm based multi-objective dynamic workflow scheduling method |
| CN104679594A (en) * | 2015-03-19 | 2015-06-03 | 成都艺辰德迅科技有限公司 | Middleware distributed calculating method |
| CN104793993A (en) * | 2015-04-24 | 2015-07-22 | 江南大学 | Cloud computing task scheduling method of artificial bee colony particle swarm algorithm based on Levy flight |
Non-Patent Citations (1)
| Title |
|---|
| "一种求解云计算资源优化的改进蝙蝠算法";张燕 等;《科技通报》;20141130;第30卷(第11期);第117-130页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105302630A (en) | 2016-02-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105302630B (en) | A kind of dynamic adjusting method and its system of virtual machine | |
| Ranjbari et al. | A learning automata-based algorithm for energy and SLA efficient consolidation of virtual machines in cloud data centers | |
| Malekloo et al. | An energy efficient and SLA compliant approach for resource allocation and consolidation in cloud computing environments | |
| Li et al. | Energy-aware and multi-resource overload probability constraint-based virtual machine dynamic consolidation method | |
| Zhu et al. | A three-dimensional virtual resource scheduling method for energy saving in cloud computing | |
| Esfandiarpoor et al. | Structure-aware online virtual machine consolidation for datacenter energy improvement in cloud computing | |
| Luo et al. | Hybrid shuffled frog leaping algorithm for energy-efficient dynamic consolidation of virtual machines in cloud data centers | |
| Li et al. | Bayesian network-based virtual machines consolidation method | |
| Zakarya | Energy, performance and cost efficient datacenters: A survey | |
| Li et al. | Holistic energy and failure aware workload scheduling in Cloud datacenters | |
| Al-Dulaimy et al. | Type-aware virtual machine management for energy efficient cloud data centers | |
| CN104657215A (en) | Virtualization energy-saving system in Cloud computing | |
| CN109684074A (en) | Physical machine resource allocation methods and terminal device | |
| Wang et al. | An energy-aware resource allocation heuristics for VM scheduling in cloud | |
| Sampaio et al. | Towards high-available and energy-efficient virtual computing environments in the cloud | |
| Xu et al. | Resource pre-allocation algorithms for low-energy task scheduling of cloud computing | |
| Zakarya et al. | Managing energy, performance and cost in large scale heterogeneous datacenters using migrations | |
| Supreeth et al. | VM Scheduling for Efficient Dynamically Migrated Virtual Machines (VMS-EDMVM) in Cloud Computing Environment. | |
| Li et al. | Dynamic virtual machine consolidation algorithm based on balancing energy consumption and quality of service | |
| Armenta-Cano et al. | Heterogeneous job consolidation for power aware scheduling with quality of service | |
| Ali et al. | Profit-aware DVFS enabled resource management of IaaS cloud | |
| Madireddy et al. | Dynamic virtual machine relocation system for energy‐efficient resource management in the cloud | |
| Arabas | Modeling and simulation of hierarchical task allocation system for energy-aware HPC clouds | |
| CN110308991A (en) | A kind of data center's energy conservation optimizing method and system based on Random Task | |
| Guo et al. | Heuristic algorithms for energy and performance dynamic optimization in cloud computing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |