CN104991825B - A kind of excessive distribution of Hypervisor resource and dynamic adjusting method and system based on Load-aware - Google Patents
A kind of excessive distribution of Hypervisor resource and dynamic adjusting method and system based on Load-aware Download PDFInfo
- Publication number
- CN104991825B CN104991825B CN201510140721.7A CN201510140721A CN104991825B CN 104991825 B CN104991825 B CN 104991825B CN 201510140721 A CN201510140721 A CN 201510140721A CN 104991825 B CN104991825 B CN 104991825B
- Authority
- CN
- China
- Prior art keywords
- hypervisor
- resource
- load
- distribution
- host
- 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
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Debugging And Monitoring (AREA)
Abstract
The excessive distribution of Hypervisor resource and dynamic adjusting method and system that the present invention relates to a kind of based on Load-aware, the load and excessive distribution situation that invention passes through monitoring Hypervisor resource, setting, record, and adjust the resource oversubscription distribution coefficient of each Hypervisor host, dynamic adjustment is carried out to each of cluster Hypervisor setting resource oversubscription distribution coefficient, quantify the resource multiplex ability of Hypervisor, a rather than empirical value, by migrating virtual machine in the distribution of Hypervisor cluster, under the premise of guaranteeing that virtual machine operates normally, promote the resource utilization of Hypervisor, to improve the overall utilization rate of cloud, reduce energy consumption, reduce maintenance cost.
Description
Technical field
The present invention relates to cloud computing management domain more particularly to a kind of Hypervisor resource oversubscription based on Load-aware
Match and dynamic adjusting method and system.
Background technique
Hypervisor, a kind of intermediate software layer operated between basic physics server and operating system allow
Multiple operating systems and Application share hardware.It can also be called VMM (virtual machine monitor), i.e., virtual machine monitors
Device.
Hypervisors is a kind of " member " operating system in virtual environment.Include on their accessible servers
All physical equipments including disk and memory.
Hypervisors not only coordinates the access of these hardware resources, also applies between each virtual machine simultaneously anti-
Shield.When server is started and carried out Hypervisor, it can load the operating system of all virtual-machine clients while can divide
Each suitable memory of virtual machine of dispensing, CPU, network and disk.
The resource excessive distribution ability of Hypervisor makes physical machine possess higher resource consolidation rate, while more effective land productivity
With computing resource, it can cut operating costs.Using resource excessive distribution, Hypervisor mentions physical resource in the best way
For virtual machine of good performance.For example, it is virtual that user can operate multiple Windows when a virtual desktop framework is disposed
Machine, each includes a processing application program.There is a possibility that identical virtual with a Hypervisor and excessive offer
The virtual machine of desktop framework.Due to virtual machine include similar operating system and application program, their many memory pages and
Include similar content.
Hypervisor finds and consolidates content memory pages identical with these virtual machines, to save CPU/ memory.
This just preferably utilizes CPU/ memory and the higher integration rate of realization.
Currently, the resource excessive distribution technology of Hypervisor is highly developed in major virtualization manufacturer and product
(Hyper-v, Xenserver, KVM, Esxi etc.).
But when Hypervisor excessive distribution resource is excessive, real resource usage amount is for a long time more than Hypervisor resource
When the upper limit, partial virtual machine can not continue to obtain enough CPU/ memory sources.Occurs the lance of contention for resources inside Hypervisor
Shield can grow in intensity, and eventually lead to virtual-machine fail or even Hypervisor failure.
Summary of the invention
It is a kind of based on Load-aware the technical problem to be solved by the present invention is in view of the deficiencies of the prior art, provide
The excessive distribution of Hypervisor resource and dynamic adjusting method and system, the isolated island for solving Hypervisor resource multiplex technology are asked
Topic, dynamic, deployment of the adjustment virtual machine in Hypervisor cluster of intelligence, distribution situation, reduction resource multiplex technology cause
All kinds of failures, improve the whole resource utilization in cloud.
The technical scheme to solve the above technical problems is that a kind of Hypervisor resource based on Load-aware
Excessive distribution and dynamic adjusting method, include the following steps:
Step 1, Hypervisor resource load threshold, Hypervisor resource excessive distribution threshold value, each is set
The oversubscription distribution coefficient and excessive distribution regulation coefficient of Hypervisor host;
Step 2, the scheduling strategy of Hypervisor cluster virtual machine is set;
Step 3, the loading condition of each Hypervisor host is monitored;
Step 4, it is negative successively to judge whether the load of each Hypervisor host is greater than or equal to Hypervisor resource
Threshold value is carried, if it is reduces the oversubscription distribution coefficient of the Hypervisor host according to excessive distribution regulation coefficient, executes step 5,
It is no to then follow the steps 6;
Step 5, the scheduling strategy according to the oversubscription distribution coefficient of the Hypervisor host adjusted and step 2
Adjust distribution of the virtual machine in Hypervisor cluster, return step 3;
Step 6, the resource excessive distribution situation of the Hypervisor host is detected;
Step 7, judge whether the resource excessive distribution ratio of the Hypervisor host is more than or equal to Hypervisor resource
Excessive distribution threshold value if it is increases the oversubscription distribution coefficient of the Hypervisor host according to excessive distribution regulation coefficient, returns to step
Rapid 6, otherwise terminate process.
The beneficial effects of the present invention are: load and excessive distribution feelings of the invention by monitoring Hypervisor resource
Condition is arranged, record, and adjusts the resource oversubscription distribution coefficient of each Hypervisor host, to each of cluster
Hypervisor is arranged resource oversubscription distribution coefficient and carries out dynamic adjustment, quantifies the resource multiplex ability of Hypervisor, without
It is an empirical value, by migrating virtual machine under the premise of the distribution of Hypervisor cluster, guarantee virtual machine normal operation,
The resource utilization of Hypervisor is promoted, to improve the overall utilization rate of cloud, energy consumption is reduced, reduces maintenance cost.
Based on the above technical solution, the present invention can also be improved as follows.
Further, the virtual machine scheduling policy of Hypervisor cluster described in step 2 using first-fit, best-fit,
Any one of packing, loadbalance, Round-Robin or random.
Further, above-mentioned technical proposal further includes the oversubscription distribution coefficient default value for presetting Hypervisor host, respectively
The oversubscription distribution coefficient of a Hypervisor host is according in excessive distribution regulation coefficient dynamic adjustment process, oversubscription disposition adjusted
Number cannot be below presetting the oversubscription distribution coefficient default value of Hypervisor host.
Further, the specific implementation of step 5 are as follows:
Step 5.1, each Hypervisor host is detected one by one;
Step 5.2, judge whether the load of Hypervisor host to be detected is less than Hypervisor resource load threshold
Value, it is no to then follow the steps 5.5 if it is execution step 5.3;
Step 5.3, it is super to judge whether the resource excessive distribution ratio of the Hypervisor host is less than Hypervisor resource
Allocation threshold, it is no to then follow the steps 5.5 if it is execution step 5.4;
Step 5.4, load is more than or equal to virtual on the Hypervisor host of Hypervisor resource load threshold
Machine moves to Hypervisor host described in step 5.3;
Step 5.5, process is ended processing.
Further, above-mentioned technical proposal further include according to adjustment result adjustment Hypervisor resource load threshold and/or
Hypervisor resource excessive distribution threshold value.
Another technical solution that the present invention solves above-mentioned technical problem is as follows: a kind of Hypervisor based on Load-aware
Resource excessive distribution and dynamic debugging system, including parameter setting module, scheduling strategy setup module, load monitoring module, load
Scheduler module and excessive distribution monitoring module;
The parameter setting module is used to be arranged Hypervisor resource load threshold, Hypervisor resource oversubscription
Oversubscription distribution coefficient and excessive distribution regulation coefficient with threshold value, each Hypervisor host;
The scheduling strategy setup module, is used to be arranged the scheduling strategy of Hypervisor cluster virtual machine;
The load monitoring module, is used to successively judge whether the load of each Hypervisor host to be greater than or equal to
Hypervisor resource load threshold if it is reduces the excessive distribution of the Hypervisor host according to excessive distribution regulation coefficient
Coefficient calls load dispatch module, otherwise calls excessive distribution monitoring module;
The load dispatch module is used for oversubscription distribution coefficient and scheduling according to the Hypervisor host adjusted
Developing Tactics virtual machine Hypervisor cluster distribution, until the Hypervisor host load be less than Hypervisor
Resource load threshold;
The excessive distribution monitoring module, is used to detect the resource excessive distribution situation of the Hypervisor host, and judgement is sentenced
Whether the resource excessive distribution ratio of the Hypervisor host of breaking is more than or equal to Hypervisor resource excessive distribution threshold value, if it is
The oversubscription distribution coefficient for then increasing the Hypervisor host according to excessive distribution regulation coefficient continues to test and detects the Hypervisor
The resource excessive distribution situation of host, until the excessive distribution ratio of the Hypervisor host is less than Hypervisor resource load threshold
Value, otherwise terminates process.
Further, the Hypervisor cluster virtual machine scheduling policy of the scheduling strategy setup module setting uses
Any one of first-fit, best-fit, packing, loadbalance, Round-Robin or random.
Further, the parameter setting module also presets the oversubscription distribution coefficient default value of Hypervisor host, respectively
The oversubscription distribution coefficient of a Hypervisor host is according in excessive distribution regulation coefficient dynamic adjustment process, oversubscription disposition adjusted
Number cannot be below presetting the oversubscription distribution coefficient default value of Hypervisor host.
Further, the load dispatch module includes resources measurement unit, the first judging unit, second judgment unit and void
Quasi- machine migration units;
The resources measurement unit is used to detect each Hypervisor host one by one;
First judging unit, is used to judge whether the load of Hypervisor host to be detected to be less than
Otherwise Hypervisor resource load threshold ends processing process if it is second judgment unit is called;
The second judgment unit, is used to judge whether the resource excessive distribution ratio of the Hypervisor host to be less than
Hypervisor resource excessive distribution threshold value ends processing process if it is virtual machine (vm) migration unit is called;
The virtual machine (vm) migration unit is used to load more than or equal to Hypervisor resource load threshold
Virtual machine (vm) migration on Hypervisor host is to the Hypervisor host.
Further, above-mentioned technical proposal further include according to adjustment result adjustment Hypervisor resource load threshold and/or
Hypervisor resource excessive distribution threshold value.
Detailed description of the invention
Fig. 1 is a kind of excessive distribution of Hypervisor resource and dynamic adjusting method stream based on Load-aware of the present invention
Cheng Tu;
Fig. 2 is a kind of excessive distribution of Hypervisor resource and dynamic debugging system frame based on Load-aware of the present invention
Figure;
Fig. 3 is load dispatch modular structure schematic diagram of the present invention.
In attached drawing, parts list represented by the reference numerals are as follows:
1, parameter setting module, 2, scheduling strategy setup module, 3, load monitoring module, 4, load dispatch module, 5, super
Distribute monitoring module, 41, monitoring resource unit, the 42, first judging unit, 4,3, second judgment unit, 44, virtual machine (vm) migration list
Member.
Specific embodiment
The principle and features of the present invention will be described below with reference to the accompanying drawings, and the given examples are served only to explain the present invention, and
It is non-to be used to limit the scope of the invention.
The excessive distribution of Hypervisor resource and dynamic adjusting method and system that the present invention relates to a kind of based on Load-aware.
CPU, memory source excessive distribution management and dynamic is primarily related to adjust.
As shown in Figure 1, a kind of excessive distribution of Hypervisor resource and dynamic adjusting method based on Load-aware, including such as
Lower step:
Step 1, Hypervisor resource load threshold, Hypervisor resource excessive distribution threshold value are set (i.e.
Virtual machine aggregate resource/(Hypervisor resource * oversubscription distribution coefficient), each Hypervisor host on Hypervisor
Oversubscription distribution coefficient and excessive distribution regulation coefficient;
Step 2, the scheduling strategy of Hypervisor cluster virtual machine is set;
Step 3, the loading condition of each Hypervisor host is monitored;
Step 4, it is negative successively to judge whether the load of each Hypervisor host is greater than or equal to Hypervisor resource
Threshold value is carried, if it is reduces the oversubscription distribution coefficient of the Hypervisor host according to excessive distribution regulation coefficient, executes step 5,
It is no to then follow the steps 6;
Step 5, the scheduling strategy according to the oversubscription distribution coefficient of the Hypervisor host adjusted and step 2
Adjust distribution of the virtual machine in Hypervisor cluster, return step 3;
Step 6, the resource excessive distribution situation of the Hypervisor host is detected;
Step 7, judge whether the resource excessive distribution ratio of the Hypervisor host is more than or equal to Hypervisor resource
Excessive distribution threshold value if it is increases the oversubscription distribution coefficient of the Hypervisor host according to excessive distribution regulation coefficient, returns to step
Rapid 6, otherwise terminate process.
The virtual machine scheduling policy of Hypervisor cluster described in step 2 using first-fit, best-fit,
Any one of packing, loadbalance, Round-Robin or random.
Above-mentioned technical proposal further includes the oversubscription distribution coefficient default value for presetting Hypervisor host, each
The oversubscription distribution coefficient of Hypervisor host is according in excessive distribution regulation coefficient dynamic adjustment process, oversubscription distribution coefficient adjusted
It cannot be below presetting the oversubscription distribution coefficient default value of Hypervisor host.
For example the oversubscription distribution coefficient default value of Hypervisor host is set as 1, the oversubscription of each Hypervisor host
Distribution coefficient is respectively set to 2,3,4,5,6, and coefficient adjusted can be 2,3,4,3,2, i.e., cannot be less than default value 1;Again
For example, the oversubscription distribution coefficient default value of Hypervisor host is set as 2, the oversubscription distribution coefficient point of each Hypervisor host
It is not set as 2,4,6,8,10, coefficient adjusted can be 2,4,6,4,2, i.e., cannot be less than default value 2.
The specific implementation of step 5 are as follows:
Step 5.1, each Hypervisor host is detected one by one;
Step 5.2, judge whether the load of Hypervisor host to be detected is less than Hypervisor resource load threshold
Value, it is no to then follow the steps 5.5 if it is execution step 5.3;
Step 5.3, it is super to judge whether the resource excessive distribution ratio of the Hypervisor host is less than Hypervisor resource
Allocation threshold, it is no to then follow the steps 5.5 if it is execution step 5.4;
Step 5.4, load is more than or equal to virtual on the Hypervisor host of Hypervisor resource load threshold
Machine moves to Hypervisor host described in step 5.3;
Step 5.5, process is ended processing.
Above-mentioned technical proposal further include according to adjustment result adjustment Hypervisor resource load threshold and/or
Hypervisor resource excessive distribution threshold value.
Above-mentioned technical proposal further includes presetting detection cycle, is distributed in each detection cycle to virtual machine in cluster
Situation is adjusted.
As shown in Fig. 2, a kind of excessive distribution of Hypervisor resource and dynamic debugging system based on Load-aware, including ginseng
Number setup module 1, scheduling strategy setup module 2, load monitoring module 3, load dispatch module 4 and excessive distribution monitoring module 5;Institute
Parameter setting module 1 is stated, is used to be arranged Hypervisor resource load threshold, Hypervisor resource excessive distribution threshold value, each
The oversubscription distribution coefficient and excessive distribution regulation coefficient of a Hypervisor host;The scheduling strategy setup module 2, is used to be arranged
The scheduling strategy of Hypervisor cluster virtual machine;The load monitoring module 3 is used to successively judge each Hypervisor
Whether the load of host is greater than or equal to Hypervisor resource load threshold, is if it is reduced according to excessive distribution regulation coefficient
The oversubscription distribution coefficient of the Hypervisor host calls load dispatch module 4, otherwise calls excessive distribution monitoring module 5;It is described negative
Scheduler module 4 is carried, is used to be adjusted according to the oversubscription distribution coefficient and scheduling strategy of the Hypervisor host adjusted virtual
Machine Hypervisor cluster distribution, until the Hypervisor host load be less than Hypervisor resource load threshold
Value;The excessive distribution monitoring module 5 is used to detect the resource excessive distribution situation of the Hypervisor host, judges that judgement should
Whether the resource excessive distribution ratio of Hypervisor host is more than or equal to Hypervisor resource excessive distribution threshold value, if it is root
The oversubscription distribution coefficient for increasing the Hypervisor host according to excessive distribution regulation coefficient continues to test and detects the Hypervisor host
Resource excessive distribution situation, until the Hypervisor host excessive distribution ratio be less than Hypervisor resource load threshold,
Otherwise terminate process.
The Hypervisor cluster virtual machine scheduling policy that the scheduling strategy setup module 2 is arranged using first-fit,
Any one of best-fit, packing, loadbalance, Round-Robin or random.
The parameter setting module 1 also presets the oversubscription distribution coefficient default value of Hypervisor host, each
The oversubscription distribution coefficient of Hypervisor host is according in excessive distribution regulation coefficient dynamic adjustment process, oversubscription distribution coefficient adjusted
It cannot be below presetting the oversubscription distribution coefficient default value of Hypervisor host.
As shown in figure 3, the load dispatch module 4 includes resources measurement unit 41, the judgement of the first judging unit 42, second
Unit 43 and virtual machine (vm) migration unit 44;The resources measurement unit 41 is used to detect each Hypervisor host one by one;
First judging unit 42, is used to judge whether the load of Hypervisor host to be detected to be less than Hypervisor money
Otherwise source load threshold ends processing process if it is second judgment unit 43 is called;The second judgment unit 43 is used
In judging whether the resource excessive distribution ratio of the Hypervisor host is less than Hypervisor resource excessive distribution threshold value, if it is
Virtual machine (vm) migration unit 44 is called, process is ended processing;The virtual machine (vm) migration unit 44, is used to load and is more than or equal to
Virtual machine (vm) migration on the Hypervisor host of Hypervisor resource load threshold is to the Hypervisor host.
Assuming that the cluster being made of 3 Hypervisor hosts, configuration is equal are as follows: 16CPU*2.0Ghz, 32G memory,
Hypervisor is individually identified as H1, H2, H3.
100 1CPU*1.0Ghz, 2G memory virtual machine are created, virtual machine is divided into 4 groups, successively creates, T1, T2 are used below,
T3, T4 are indicated:
T1:30 platform, the load of virtual machine average resource are 10%.
T2:30 platform, the load of virtual machine average resource are 30%
T3:30 platform, the load of virtual machine average resource are 70%
T4:10 platform, the load of virtual machine average resource are between 30~70%
For easy understanding, example is saved as within the present invention to be described in detail:
Total memory source number that virtual machine needs is 200G, is far longer than the physical memory (96G) of Hypervisor.It is necessary
It can just be met the requirements using the resource multiplex technology of Hypervisor.
Traditional resource multiplex technology merely provides resource excessive distribution ability, after the completion of virtual machine creating, virtual machine
Distribution often will following table:
Table 1
From table 1 as can be seen that H1, the memory that the memory usage of H2 has had reached 96% or more, H3 are up to
56.8%.When the fluctuation of the memory of virtual machine, H1, H2 just will appear serious contention for resources, and be unable to satisfy at all actual
Use demand often will cause virtual machine response slowly, notably the problems such as Hypervisor failure.And H1 deploys 65 virtually
Machine, and H3 only has 13 virtual machines, is distributed also unreasonable.
By the invention it is possible to quantify the method for resource allocation, dynamic adjusts the distribution of virtual machine, the stabilization of lifting system
Property.
Precondition:
1. the load of pair each Hypervisor host and resource allocation conditions are monitored;
2. it is 5 minutes that detection cycle, which is arranged,;
3. the CPU/ memory default oversubscription distribution coefficient that each Hypervisor host is arranged is 1;
4. it is 1 that excessive distribution regulation coefficient, which is arranged,;
5. it is 90% that Hypervisor resource load threshold, which is arranged,;
6. it is 85% that Hypervisor resource excessive distribution threshold value, which is arranged,;
7. setting Hypervisor cluster virtual machine scheduling policy is best free, i.e., preferentially select resources occupation rate most
Low Hypervisor.
In order to which the present invention can be allowed more convincing, using the Hypervisor and virtual machine with 1 same size of table,
Within save as example and be illustrated.
Application scenarios of the invention, mainly in terms of adjustment two after the completion of the creation process of virtual machine carries out and creates
It is described, elaborates separately below.
Describe how the creation process of virtual machine applies method of the invention, for easy understanding, illustrate below first
Ignore the reserved consumption of memory of Hypervisor and virtual machine itself.Original state is as shown in table 2.
Table 2
Hypervisor | It can storage allocation | Memory average load | Virtual machine |
H1 | 32G*1=32G | 0% | Nothing |
H2 | 32G*1=32G | 0% | Nothing |
H3 | 32G*1=32G | 0% | Nothing |
Successively create 100 virtual machines:
T1:30 platform virtual machine, average resource load are 10%;
T2:30 platform virtual machine, average resource load are 30%;
T3:30 platform virtual machine, average resource load are 70%;
T4:10 platform virtual machine, average resource load are between 30~70%.
The 1st virtual machine: memory 2G is created, virtual machine loads 10%, according to setting cluster virtual machine scheduling policy, preferentially
The H1 for selecting a memory usage minimum, due to load (virtual machine) not yet, memory loads occupancy and is less than threshold value 90%,
Continuing to judge Memory Allocation situation, 2G/32G=6.2% after distribution is less than resource excessive distribution threshold value 85%, terminate process judgement,
Create virtual machine.
Since each Hypervisor resource is more idle, according to the scheduling strategy being arranged at present, the deployment of virtual machine is successively
With H1, H2, H3 is uniformly distributed, and until resource load or excessive distribution reach threshold value, virtual machine exists when the first time of this example adjusts
The state of Hypervisor is as shown in table 3.
Table 3
Since default excessive distribution threshold value is 85%, before adjusting oversubscription distribution coefficient, can not allow to distribute new virtual machine.It is super
Allocation proportion threshold value is that 85%, H1 distributes 10 T1, and when 4 T2, occupancys storage allocation is 28G, reach virtual machine can distribute it is interior
87.5% deposited.Since memory average load only has 13.7%, it is lower than load threshold, need to be increased according to excessive distribution regulation coefficient
The oversubscription distribution coefficient of Hypervisor, in conjunction with precondition, after adjustment as shown in Table 4.
Table 4
Since the Hypervisor cluster virtual machine scheduling policy of this example setting is best free, next virtual machine
Creation, can select H2, rather than can the biggish H1 of storage allocation.So also will increase the oversubscription distribution coefficient of H2.And so on, directly
100 virtual machines are completed to creation, as a result such as table 5.
Table 5
Above description can be appreciated that the present invention has the following advantages:
The excessive distribution ability of 1.Hypervisor can quantitative statistics, and adjust automatically;
2. keeping the distribution of virtual machine more balanced in conjunction with scheduling strategy;
3. load threshold controls, so that Hypervisor is not distributed excess resource leads to failure.
Above-described embodiment only describes a fairly simple application scenarios, using the virtual machine of relatively simple specification,
And orderly creation, virtual machine load all in a certain range.In practical applications, the specification of virtual machine is varied, especially
Load is unpredictalbe.It is single from virtual machine creating angle be using method of the invention it is inadequate, need to also after the completion of creation,
The load of each Hypervisor of periodic detection, oversubscription distribution coefficient carry out dynamic adjustment.
Next, the problem of how we use process of the invention to have a look, and adjustment table 1 has existed.Initially
Situation such as table 6.
Table 6
In table 6, the reality that the oversubscription distribution coefficient of each Hypervisor host is 1, H1, H2 storage allocation far more than mistake
Can apportioning cost, but since the actual loading of memory has been above the setting of load threshold, so will not influence process of the present invention
Final result.Newly created virtual machine will not continue to be assigned to H1, on H2, increase the burden of Hypervisor.
Hypervisor host is detected one by one according to the method for the present invention:
Step 1: H1 being detected with the period that preset condition is arranged;
Step 2: detecting that the resource load of H1 is greater than threshold value 90% and does not reduce since the oversubscription distribution coefficient of H1 has been 1
Oversubscription distribution coefficient;
Step 3: virtual machine being migrated according to the scheduling strategy of setting and load threshold, H3 load most loads most in the cluster
Low, after the T1 virtual machine for migrating a H1, load still is below load threshold 90%, selects H3 to migrate target
Hypervisor;
Step 4: migrating a T1 virtual machine to H3 from H1;
Step 5: detecting the memory loading condition of H1 again, loading is 96.2%, is greater than load threshold 90%, repeats step
2-4, until the load of H1 is lower than in threshold value 90% or cluster without suitable Hypervisor;
Step 6: when the load of H1 is lower than 90%, virtual machine is distributed when being 89% are as follows: T1,18;T2,30;T3,5;
Step 7:H1 can storage allocation be 32G, it is 106G that real resource, which distributes occupancy situation, is greater than and super matches threshold value
85%, increase the super distribution coefficient of H1, the super distribution coefficient of H1 is 2 after increase, i.e. 32*2=64G;
Step 8: detection distribution again, resource allocation conditions 165.5% are greater than to surpass and match threshold value 85%, continue growing H1
Super distribution coefficient, super match threshold value until resource allocation conditions are less than.
By the adjustment in 1 period, Hypervisor and virtual machine distribution such as table 7.
Table 7
From table 7 as can be seen that the load of each Hypervisor host is loaded already below the Hypervisor of setting
Threshold value, the distribution of virtual machine are opposite balanced with table 1.It can also continue dynamic by modifying Hypervisor load threshold and drop
The load of low H1, H2 improve the load of H3.
For table 5 compared with table 7, the oversubscription distribution coefficient of each Hypervisor and virtual distribution be not identical.But all meet
Preset condition, it is seen that adaptability and scalability of the invention.
The foregoing is merely presently preferred embodiments of the present invention, is not intended to limit the invention, it is all in spirit of the invention and
Within principle, any modification, equivalent replacement, improvement and so on be should all be included in the protection scope of the present invention.
Claims (10)
1. a kind of excessive distribution of Hypervisor resource and dynamic adjusting method based on Load-aware, which is characterized in that including such as
Lower step:
Step 1, Hypervisor resource load threshold, Hypervisor resource excessive distribution threshold value, each Hypervisor are set
The oversubscription distribution coefficient and excessive distribution regulation coefficient of host;
Step 2, the scheduling strategy of Hypervisor cluster virtual machine is set;
Step 3, the loading condition of each Hypervisor host is monitored;
Step 4, successively judge whether the load of each Hypervisor host is greater than or equal to Hypervisor resource load threshold
Value if it is reduces the oversubscription distribution coefficient of the Hypervisor host according to excessive distribution regulation coefficient, executes step 5, otherwise
Execute step 6;
Step 5, the scheduling strategy according to the oversubscription distribution coefficient of the Hypervisor host adjusted and step 2 adjusts
Distribution of the virtual machine in Hypervisor cluster, return step 3;
Step 6, the resource excessive distribution situation of the Hypervisor host is detected;
Step 7, judge whether the resource excessive distribution ratio of the Hypervisor host is more than or equal to Hypervisor resource oversubscription
With threshold value, if it is increase the oversubscription distribution coefficient of the Hypervisor host according to excessive distribution regulation coefficient, return step 6,
Otherwise terminate process.
2. a kind of excessive distribution of Hypervisor resource and dynamic adjusting method based on Load-aware according to claim 1,
It is characterized in that, the virtual machine scheduling policy of Hypervisor cluster described in step 2 using first-fit, best-fit,
Any one of packing, loadbalance, Round-Robin or random.
3. a kind of excessive distribution of Hypervisor resource and dynamic adjusting method based on Load-aware according to claim 1,
It is characterized in that, further including the oversubscription distribution coefficient default value for presetting Hypervisor host, each Hypervisor host
Oversubscription distribution coefficient according in excessive distribution regulation coefficient dynamic adjustment process, oversubscription distribution coefficient adjusted cannot be below setting in advance
Set the oversubscription distribution coefficient default value of Hypervisor host.
4. a kind of excessive distribution of Hypervisor resource and dynamic adjusting method based on Load-aware according to claim 1,
It is characterized in that, the specific implementation of step 5 are as follows:
Step 5.1, each Hypervisor host is detected one by one;
Step 5.2, judge whether the load of Hypervisor host to be detected is less than Hypervisor resource load threshold, such as
Fruit is to execute step 5.3, no to then follow the steps 5.5;
Step 5.3, judge whether the resource excessive distribution ratio of the Hypervisor host is less than Hypervisor resource excessive distribution
Threshold value, it is no to then follow the steps 5.5 if it is execution step 5.4;
Step 5.4, virtual machine load being more than or equal on the Hypervisor host of Hypervisor resource load threshold moves
Move on to Hypervisor host described in step 5.3;
Step 5.5, process is ended processing.
5. a kind of excessive distribution of Hypervisor resource and dynamic adjusting method based on Load-aware according to claim 1,
It is characterized in that, further including super according to adjustment result adjustment Hypervisor resource load threshold and/or Hypervisor resource
Allocation threshold.
6. a kind of excessive distribution of Hypervisor resource and dynamic debugging system based on Load-aware, which is characterized in that including ginseng
Number setup module, scheduling strategy setup module, load monitoring module, load dispatch module and excessive distribution monitoring module;
The parameter setting module is used to be arranged Hypervisor resource load threshold, Hypervisor resource excessive distribution threshold
The oversubscription distribution coefficient and excessive distribution regulation coefficient of value, each Hypervisor host;
The scheduling strategy setup module, is used to be arranged the scheduling strategy of Hypervisor cluster virtual machine;
The load monitoring module, is used to successively judge whether the load of each Hypervisor host to be greater than or equal to
Hypervisor resource load threshold if it is reduces the excessive distribution of the Hypervisor host according to excessive distribution regulation coefficient
Coefficient calls load dispatch module, otherwise calls excessive distribution monitoring module;
The load dispatch module is used for oversubscription distribution coefficient and scheduling strategy according to the Hypervisor host adjusted
Virtual machine is adjusted in the distribution of Hypervisor cluster, until the load of the Hypervisor host is less than Hypervisor resource
Load threshold;
The excessive distribution monitoring module is used to detect the resource excessive distribution situation of the Hypervisor host, judges that judgement should
Whether the resource excessive distribution ratio of Hypervisor host is more than or equal to Hypervisor resource excessive distribution threshold value, if it is root
The oversubscription distribution coefficient for increasing the Hypervisor host according to excessive distribution regulation coefficient continues to test and detects the Hypervisor host
Resource excessive distribution situation, until the Hypervisor host excessive distribution ratio be less than Hypervisor resource load threshold,
Otherwise terminate process.
7. a kind of excessive distribution of Hypervisor resource and dynamic debugging system based on Load-aware according to claim 6,
It is characterized in that, the Hypervisor cluster virtual machine scheduling policy of the scheduling strategy setup module setting uses first-
Any one of fit, best-fit, packing, loadbalance, Round-Robin or random.
8. a kind of excessive distribution of Hypervisor resource and dynamic debugging system based on Load-aware according to claim 6,
It is characterized in that, the parameter setting module also presets the oversubscription distribution coefficient default value of Hypervisor host, it is each
The oversubscription distribution coefficient of Hypervisor host is according in excessive distribution regulation coefficient dynamic adjustment process, oversubscription distribution coefficient adjusted
It cannot be below presetting the oversubscription distribution coefficient default value of Hypervisor host.
9. a kind of excessive distribution of Hypervisor resource and dynamic debugging system based on Load-aware according to claim 6,
It is characterized in that, the load dispatch module includes resources measurement unit, the first judging unit, second judgment unit and virtual machine
Migration units;
The resources measurement unit is used to detect each Hypervisor host one by one;
First judging unit, is used to judge whether the load of Hypervisor host to be detected to be less than Hypervisor
Otherwise resource load threshold ends processing process if it is second judgment unit is called;
The second judgment unit is used to judge that the load to be less than the Hypervisor master of Hypervisor resource load threshold
Whether the resource excessive distribution ratio of machine is less than Hypervisor resource excessive distribution threshold value, if it is call virtual machine (vm) migration unit,
End processing process;
The virtual machine (vm) migration unit is used to load the Hypervisor more than or equal to Hypervisor resource load threshold
Virtual machine (vm) migration on host is less than the Hypervisor host of Hypervisor resource load threshold to the load.
10. a kind of excessive distribution of Hypervisor resource and dynamic debugging system based on Load-aware according to claim 6,
It is characterized in that, further including super according to adjustment result adjustment Hypervisor resource load threshold and/or Hypervisor resource
Allocation threshold.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510140721.7A CN104991825B (en) | 2015-03-27 | 2015-03-27 | A kind of excessive distribution of Hypervisor resource and dynamic adjusting method and system based on Load-aware |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510140721.7A CN104991825B (en) | 2015-03-27 | 2015-03-27 | A kind of excessive distribution of Hypervisor resource and dynamic adjusting method and system based on Load-aware |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104991825A CN104991825A (en) | 2015-10-21 |
CN104991825B true CN104991825B (en) | 2019-07-05 |
Family
ID=54303641
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510140721.7A Active CN104991825B (en) | 2015-03-27 | 2015-03-27 | A kind of excessive distribution of Hypervisor resource and dynamic adjusting method and system based on Load-aware |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104991825B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106059940B (en) * | 2016-05-25 | 2019-07-09 | 新华三信息技术有限公司 | A kind of flow control methods and device |
CN107707424A (en) * | 2017-09-11 | 2018-02-16 | 厦门集微科技有限公司 | The control method and system of load condition |
CN110930291B (en) * | 2019-11-15 | 2022-06-17 | 山东英信计算机技术有限公司 | GPU video memory management control method and related device |
CN113722100B (en) * | 2021-09-02 | 2023-12-15 | 上海仪电(集团)有限公司中央研究院 | Dynamic super-allocation method and equipment for cloud platform memory resources based on trend prediction |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101968749A (en) * | 2010-09-26 | 2011-02-09 | 华中科技大学 | Method for receiving message passing interface (MPI) information under circumstance of over-allocation of virtual machine |
CN102508718A (en) * | 2011-11-22 | 2012-06-20 | 杭州华三通信技术有限公司 | Method and device for balancing load of virtual machine |
CN103502944A (en) * | 2011-11-28 | 2014-01-08 | 华为技术有限公司 | Method and device for adjusting memories of virtual machines |
CN104050008A (en) * | 2013-03-15 | 2014-09-17 | 中兴通讯股份有限公司 | Excessive distribution management system and method for memory |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8104039B2 (en) * | 2006-08-07 | 2012-01-24 | International Business Machines Corporation | Method for balancing resource sharing and application latency within a data processing system |
US20080307188A1 (en) * | 2007-06-06 | 2008-12-11 | International Business Machines Corporation | Management of Guest OS Memory Compression In Virtualized Systems |
US9069890B2 (en) * | 2011-04-20 | 2015-06-30 | Cisco Technology, Inc. | Ranking of computing equipment configurations for satisfying requirements of virtualized computing environments based on an overall performance efficiency |
US9176787B2 (en) * | 2011-08-18 | 2015-11-03 | Lenovo Enterprise Solutions (Singapore) Pte. Ltd. | Preserving, from resource management adjustment, portions of an overcommitted resource managed by a hypervisor |
-
2015
- 2015-03-27 CN CN201510140721.7A patent/CN104991825B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101968749A (en) * | 2010-09-26 | 2011-02-09 | 华中科技大学 | Method for receiving message passing interface (MPI) information under circumstance of over-allocation of virtual machine |
CN102508718A (en) * | 2011-11-22 | 2012-06-20 | 杭州华三通信技术有限公司 | Method and device for balancing load of virtual machine |
CN103502944A (en) * | 2011-11-28 | 2014-01-08 | 华为技术有限公司 | Method and device for adjusting memories of virtual machines |
CN104050008A (en) * | 2013-03-15 | 2014-09-17 | 中兴通讯股份有限公司 | Excessive distribution management system and method for memory |
Non-Patent Citations (1)
Title |
---|
Tesseract:reconciling guest IO and hypervisor swapping in a VM;Kapil Arya,Yury Baskakov,Alex Garthwaite;《Proceedings of the 10th ACM SIGPLAN/SIGOPS international conference on Virtual execution environments》;20140301;第15-28页 |
Also Published As
Publication number | Publication date |
---|---|
CN104991825A (en) | 2015-10-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104991825B (en) | A kind of excessive distribution of Hypervisor resource and dynamic adjusting method and system based on Load-aware | |
CN103605613B (en) | Cloud computing environment dynamically adjusts the method and system of virutal machine memory | |
US9891953B2 (en) | Dynamic resource configuration based on context | |
US8667496B2 (en) | Methods and systems of managing resources allocated to guest virtual machines | |
CN104834569B (en) | A kind of cluster resource dispatching method and system based on application type | |
Salfner et al. | Downtime analysis of virtual machine live migration | |
CN108037994A (en) | Scheduling mechanism supporting multi-core parallel processing in heterogeneous environment | |
US7533385B1 (en) | Virtualization and server imaging system for allocation of computer hardware and software | |
CN101788920A (en) | CPU virtualization method based on processor partitioning technology | |
CN107066332A (en) | Distributed system and its dispatching method and dispatching device | |
CN102594861A (en) | Cloud storage system with balanced multi-server load | |
WO2019153697A1 (en) | Cluster resource adjustment method and device, and cloud platform | |
CN102339229A (en) | Virtualization method based on operating system layer | |
CN109558216B (en) | Single root I/O virtualization optimization method and system based on online migration | |
CN107346264A (en) | A kind of method, apparatus and server apparatus of virtual machine load balance scheduling | |
CN103502944A (en) | Method and device for adjusting memories of virtual machines | |
CN106686136A (en) | Cloud resource scheduling method and device | |
CN103870332A (en) | Virtual machine processor resource adjusting method and device and virtual machine system | |
CN108123980A (en) | A kind of resource regulating method and system | |
CN104461731B (en) | Resources of virtual machine recovery method and device in a kind of dynamic resource extension | |
CN102654841A (en) | Method and device for allocating computing resource of virtual machine based on fine granularity | |
CN106713375A (en) | Method and device for allocating cloud resources | |
CN105183563B (en) | A kind of cpu resource dynamic self-configuration method towards key task computer | |
CN107463430A (en) | A kind of virutal machine memory dynamic management system and method based on internal memory and Swap spaces | |
WO2021002961A1 (en) | Harvest virtual machine for utilizing cloud-computing resources |
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 | ||
PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: A method and system of hypervisor resource over allocation and dynamic adjustment based on load perception Effective date of registration: 20210510 Granted publication date: 20190705 Pledgee: Beijing ustron Tongsheng financing Company limited by guarantee Pledgor: BEIJING SKYCLOUD RONGCHUANG SOFTWARE TECHNOLOGY Co.,Ltd. Registration number: Y2021990000410 |
|
PE01 | Entry into force of the registration of the contract for pledge of patent right |