CN102546379A - Virtualized resource scheduling method and system - Google Patents
Virtualized resource scheduling method and system Download PDFInfo
- Publication number
- CN102546379A CN102546379A CN2010106218420A CN201010621842A CN102546379A CN 102546379 A CN102546379 A CN 102546379A CN 2010106218420 A CN2010106218420 A CN 2010106218420A CN 201010621842 A CN201010621842 A CN 201010621842A CN 102546379 A CN102546379 A CN 102546379A
- Authority
- CN
- China
- Prior art keywords
- resource
- physical server
- scheduling
- virtual
- information
- 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.)
- Granted
Links
Images
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a virtualized resource scheduling method and a system. A request from an external management user is received, or a scheduling request is triggered according to a preset strategy; according to the resource requirement information of a virtual machine and a physical server in the current system, a virtualized resource scheduling scheme is calculated and obtained; according to physical server resource flowing information contained in the received resource scheduling scheme, the resource of the relevant physical server is scheduled; according to the volume information contained in the virtualized resource scheduling scheme, the resource channel of the volume is built with the corresponding physical server; and scheduling resource is obtained from a mapping physical server to provide scheduled virtual resource for an external business entity. According to the virtualized resource scheduling method and the system, the resource scheduling efficiency can be improved, and the resource can be optimally scheduled in the whole situation.
Description
Technical field
The present invention relates to virtual scheduling of resource technology, particularly a kind of method of virtual scheduling of resource and virtual resource scheduling system.
Background technology
Since Amazon and release elasticity calculating cloud (EC2 in 2006; Elastic Compute Cloud) platform and highly successful after; Industry has started one and based on virtual flexible resource pond shared data center infrastructure is provided, and internally integrates the whirlwind of brand-new business model (the publicly-owned cloud of the IaaS pattern) research of (privately owned cloud) shared resource or externally lease service.
The combination that Intel Virtualization Technology and elasticity are calculated the cloud platform has brought brand-new resource consolidation and use pattern; Wherein, the distribution according to need of resource and dynamic flow have crucial meaning for the service quality of the utilance that improves elasticity calculating cloud platform resource, raising elasticity calculating cloud service and the total cost of ownership that reduction elasticity is calculated the cloud user.
In the cloud computing resource pool, each physical server passes through the operation virtualization software, thereby is virtualized into some separate virtual machines, as the unit of Business Entity carrying.Share the hardware resource of this physical server between each virtual machine of virtualized this physical server correspondence, that is to say, the hardware resource of this physical server can be shared by each virtual machine of correspondence, promptly can between local resource, dispatch.In the prior art, local resource mainly comprises cpu resource and memory source, down in the face of local resource scheduling carrying out brief description.
The local cpu resource regulating method:
Fig. 1 is the existing virtual machine monitor structural representation that carries out the cpu resource scheduling.Referring to Fig. 1, virtual machine monitor comprises:
Intercept and capture module, be used to intercept and capture the frequency adjustment instruction that a plurality of client operating systems send, and obtain all each self-corresponding expected frequencies of frequency adjustment instruction;
Acquisition module is used for obtaining all expected frequencies load information of corresponding virtual CPU separately according to expected frequency;
Distribution module is used for distributing true CPU resource according to the load information of virtual cpu, and further, the true CPU resource that the heavy more virtual cpu of load is assigned to is many more.
Local memory source dispatching method:
Be different from the local cpu resource regulating method and pay attention to optimized dispatching strategy design characteristic; On virtual platform, carry out local memory source scheduling; Also be faced with virtual machine internal memory operating position and obtain practical difficulties such as difficult and memory requirements prediction, therefore, on scheduling strategy; The hypothesis that has identical service priority based on each virtual machine; And be optimization aim to minimize local and interruption times, carry out the heuristic search algorithm of iteration in twos through between the corresponding a plurality of virtual machines of this physical server, being provided with, carry out the memory source scheduling according to the heuristic search algorithm result.
Along with deepening continuously of Intel Virtualization Technology and resource-sharing research; Cross over the physical server border and realize that in global scope dynamically sharing with Real-Time Scheduling of resource becomes virtual resource-sharing scheduling Development Trend; But by above-mentioned visible, existing virtual resource regulating method is confined to the corresponding a plurality of virtual machines inside of a physical server basically and carries out scheduling of resource; Virtual scheduling of resource model is too simple; Do not take all factors into consideration in the global resource scheduling scheme, factors such as very important performance cost and network capacity restriction during the long-range use of resource lack virtual global resource, optimize the ability of global resource scheduling.
With the resource to be that object carries out the visual angle of fine granularity optimized dispatching between Business Entity different with traditional scheduling of resource; The virtual machine (vm) migration dispatching method adopts the mode of Business Entity migration to realize the global configuration of all kinds of resources; Virtual resource system is thread according to each physical server resource situation with the virtual machine, and virtual machine is dispatched between each physical server; Like this, resource can be shared between a plurality of physical servers.But in this virtual machine (vm) migration dispatching method, be thread with the virtual machine, the thread granularity is thicker, and for example, the resource of a thread of less than is not participated in scheduling, makes dispatching efficiency lower, and resource can not get the effective optimization scheduling in the overall situation; And each scheduling of resource possibly reschedule the resource of original scheduled; Scheduling is comparatively complicated; Make the single dispatching office relate to the resource type complicacy, for example, need relate to the integrated decision-making of cpu resource, memory source, disk resource etc.; And receiving many condition restriction such as physical server resource concrete configuration, its optimized scheduling problem can not directly be modeled as continuous planning problem; Further, above-mentioned prior art all adopts the Static Design thinking, does not consider resource distribution and service deployment scheme in resource pool construction and the operation life cycle as a whole, and dynamically changeable factor such as business load pressure is for the influence of dispatching effect.
Summary of the invention
In view of this, main purpose of the present invention is to propose a kind of method of virtual scheduling of resource, improves scheduling of resource efficient, realizes the optimized dispatching of resource in the overall situation.
Another object of the present invention is to propose a kind of virtual resource scheduling system, improve scheduling of resource efficient, realize the optimized dispatching of resource in the overall situation.
For achieving the above object, the invention provides a kind of method of virtual scheduling of resource, this method comprises:
Reception is asked from the external management user, or, according to the dispatch request of Provisioning Policy triggering in advance,, calculate and obtain virtual resource scheduling scheme according to the resource requirement information of virtual machine and physical server in the current system;
According to the physical server resource flow information that comprises in the virtual resource scheduling scheme that receives, the resource of scheduling related physical server;
The capacity information that comprises according to virtual resource scheduling scheme is set up the chnnels of resources of this capacity with the respective physical server;
Obtain scheduling resource from the physical server of mapping, the virtual resource of scheduling is provided for the outside service entity.
What the resource requirement information of physical server comprised the physical server node can confession amount or demand information, adopts greedy matching algorithm to calculate and obtain virtual resource scheduling scheme.
The greedy matching algorithm of said employing calculates and obtains virtual resource scheduling scheme and specifically comprises:
Obtain resource pool physical server node set and resource production and marketing relation information s
i, wherein, s
iThat representes i physical server node can confession amount or demand;
All local resources physical server node that supply exceed demand is rearranged formation O from big to small by output;
All local resources physical server node that supply falls short of demand is rearranged formation I from big to small by demand;
From formation O and formation I, take out physics server node i and physical server node j respectively:
If | s
i|>| s
j|, with { i → j:s
jAdd virtual resource scheduling scheme, upgrade s
i=s
i+ s
j, i is inserted formation I again;
If | s
i|<| s
j|, with { i → j:s
iAdd virtual resource scheduling scheme, upgrade s
j=
Si+ s
j, j is inserted formation O again;
Export virtual resource scheduling scheme.
Further comprise the long-range allocation and transportation cost information of resource in the resource requirement information of said physical server, adopt the transportation optimized Algorithm to calculate and obtain virtual resource scheduling scheme.
Said employing transportation optimized Algorithm is calculated and is obtained virtual resource scheduling scheme and specifically comprises:
But obtain the resource physical server node funding source a that supply exceed demand
i
Obtain the resource physical server node b that supply falls short of demand
j
Obtain the long-range allocation and transportation cost of the resource c of i resource physical server node that supply exceed demand and j the resource physical server node that supply falls short of demand
Ij
M is the resource physical server node number that supply exceed demand, and n is the resource physical server node number that supply falls short of demand.
Further comprise the long-range allocation and transportation cost information of resource in the resource requirement information of said physical server, adopt shortest path algorithm to calculate and obtain virtual resource scheduling scheme.
Further comprise long-range allocation and transportation cost information of resource and network system load information in the resource requirement information of said physical server, adopt the minimum cost flow algorithm computation and obtain virtual resource scheduling scheme.
When resource bottleneck is the physical server end, obtain the long-range allocation and transportation cost information of resource through logic pond method.
When resource bottleneck is that network connects, obtain the long-range allocation and transportation cost information of resource through physics pond method.
Further comprise network bandwidth limitations information in the resource requirement information of said physical server.
Said employing minimum cost flow algorithm computation is also obtained virtual resource scheduling scheme and is specifically comprised:
Each physical server in the resource pool is categorized as resource physical server more than needed and the nervous physical server of resource;
Network between the physical server in the resource pool is connected bandwidth restriction in real time be mapped as in the network flow upper limit of arc between node, obtain and definite capacity limit c
Ij
With bandwidth consumption for the influence of operation system as node in the network diagram to arc (i, the resource cost of transportation b between j)
Ij
When communication performance is subject to physical server, adopt logic pond method to obtain resource cost of transportation information.
When communication performance is subject to network topology and interconnect architecture, adopt physics pond method to obtain resource cost of transportation information.
Augmenting chain through unit of adjustment's flow cost is minimum obtains said minimum value as the minimum augmenting chain of expense.
A kind of virtual resource scheduling system, this system comprises: distributed business network DSN multi-service resource manager, scheduling of resource decision-making device, distributed virtual machine Resource Scheduler, a plurality of virtual machine and a plurality of physical server, wherein,
DSN multi-service resource manager; Be used for business interface being provided according to the Business Entity of type of service to the outside; The resource mapping of maintenance service type corresponding virtual machine and physical server receives the request from the external management user, or; According to the dispatch request of Provisioning Policy triggering in advance, export the scheduling of resource decision-making device to;
The scheduling of resource decision-making device is used to provide user interface, and virtual resource scheduling scheme is calculated and obtained to the receiving scheduling request according to the resource requirement information of virtual machine and physical server in the current system,, is sent to the distributed virtual machine Resource Scheduler;
The distributed virtual machine Resource Scheduler is used for the physical server resource flow information that comprises according to the virtual resource scheduling scheme that receives, the resource of scheduling related physical server;
Virtual machine is used for obtaining scheduling resource from the physical server of mapping, and the virtual resource of scheduling is provided for the outside service entity through DSN multi-service resource manager;
Physical server is used for being virtualized into a plurality of separate virtual machines through the operation virtualization software, and the capacity information that comprises according to virtual resource scheduling scheme is set up the chnnels of resources of this capacity with the respective physical server.
Said DSN multi-service resource manager is further used for obtaining the Business Entity load information, implements the user and asks shunting and adjustment.
Said virtual resource scheduling scheme comprises: do not shut down virtual machine (vm) migration, local resource and flow and the strange land resource flow.
Said scheduling of resource decision-making device comprises: mapping block and computing module, wherein,
Mapping block; Be used to provide user interface, the receiving scheduling request is mapped as the fundamental plan model with the actual schedule scene; According to the resource requirement information of virtual machine and physical server in the current system, confirm optimizing scheduling target and corresponding network description parameter sets;
Computing module is used for the network description parameter sets definite according to mapping block, calculates and obtain the virtual resource scheduling scheme of resource flow, is sent to the distributed virtual machine Resource Scheduler.
Said distributed virtual machine Resource Scheduler comprises: monitoring modular and enforcement module, wherein,
Monitoring modular is used for extracting the required parameter of fundamental plan model and being used to determine that mapping block adopts the state parameter of fundamental plan problem model from the real network environment;
Implement module, be used for the physical server resource flow information that comprises according to the virtual resource scheduling scheme that receives, the resource of scheduling related physical server.
Said monitoring modular comprises parameter acquiring submodule and status monitoring submodule, wherein,
The parameter acquiring submodule is used for extracting required limiting parameter and the cost parameter of plan model from the real network environment;
The status monitoring submodule is used for accomplishing from the real network environment and extracts the state parameter that is used to determine mapping block employing fundamental plan problem model.
Visible by above-mentioned technical scheme; The method of a kind of virtual scheduling of resource provided by the invention and virtual resource scheduling system; Reception is asked from the external management user, or, according to the dispatch request of Provisioning Policy triggering in advance; According to the resource requirement information of virtual machine and physical server in the current system, calculate and obtain virtual resource scheduling scheme; According to the physical server resource flow information that comprises in the virtual resource scheduling scheme that receives, the resource of scheduling related physical server; The capacity information that comprises according to virtual resource scheduling scheme is set up the chnnels of resources of this capacity with the respective physical server; Obtain scheduling resource from the physical server of mapping, the virtual resource of scheduling is provided for the outside service entity.Like this, make full use of cross-platform fine granularity resource-sharing mechanism, improved resource utilization and scheduling of resource efficient, realized the optimized dispatching of resource in the overall situation.
Description of drawings
Fig. 1 is the existing virtual machine monitor structural representation that carries out the cpu resource scheduling.
Fig. 2 is the virtual resource scheduling system structural representation of the embodiment of the invention.
Fig. 3 is the method first embodiment schematic flow sheet of the virtual scheduling of resource of the present invention.
Fig. 4 is the method second embodiment schematic flow sheet of the virtual scheduling of resource of the present invention.
Fig. 5 adopts the schematic flow sheet of greedy matching algorithm calculation optimization scheduling scheme for embodiment of the invention resource global registration problem.
Fig. 6 finds the solution minimum cost flow rudimentary algorithm min_flow (G) schematic flow sheet for the embodiment of the invention.
Fig. 7 is case study of global resource optimized scheduling and the solving model structural representation in the embodiment of the invention resource pool life cycle.
Embodiment
For making the object of the invention, technical scheme and advantage clearer, will combine accompanying drawing and specific embodiment that the present invention is done to describe in detail further below.
The purpose of the embodiment of the invention is to provide and suits distributed business network (DSN; Distributed Service Network) the global resource scheduling strategy of privately owned cloud platform resource pond actual environment; Make full use of cross-platform fine granularity resource-sharing mechanism; When improving resource utilization and scheduling of resource efficient; Through to the condition analysis of remote resource access restriction, utilize the optimization planning theory, balance scheduling strategy performance boost (resource utilization optimization effect) and cost expense with loss; And combine dynamical feedback for resource pool resource distribution and operating position, self adaptation and the adjustable resource regulating method of complexity that can progressively refine according to actual needs are provided.Its core concept is:
In the virtual resource dispatching patcher, at first calculate and obtain virtual resource scheduling scheme according to the resource requirement information of virtual machine and physical server in the current system;
Then, take all factors into consideration under the prerequisite of concrete restriction of resource pool network environment and remote resource access cost, utilize optimization planning thinking to solve global resource optimized scheduling problem.The concrete restriction of resource pool network environment comprises: the real-time network bandwidth constraints between the physical server etc.; The remote resource access cost is used to reflect the Business Entity resource consumption demand cost that global resource flows, for example, and long-distance inner resource access, disk resource visit etc.;
Further; To resource pool construction resource distribution and the service deployment demand corresponding with the different phase of operation; In conjunction with composite factors such as resource load pressure real-time change; For the related service entity is provided with different state, dynamically the scheduling problem that adopted of adjustment is analyzed granularity and stipulations derivation algorithm, thereby obtains balance in dispatching effect and response time, between assessing the cost;
At last, to the difference of current grid state analysis resource bottleneck present position,, dynamically adjust the limiting parameter acquisition methods that is adopted, thereby between modeling effect and intermediate computations cost, obtain balance for the related service entity is provided with different state.
Fig. 2 is the virtual resource scheduling system structural representation of the embodiment of the invention.Referring to Fig. 2, this system comprises: DSN multi-service resource manager, scheduling of resource decision-making device, distributed virtual machine Resource Scheduler, a plurality of virtual machine and a plurality of physical server, wherein,
DSN multi-service resource manager; Be used for business interface being provided according to the Business Entity of type of service to the outside; The resource mapping of maintenance service type corresponding virtual machine and physical server receives the request from the external management user, or; According to the dispatch request of Provisioning Policy triggering in advance, export the scheduling of resource decision-making device to;
In the embodiment of the invention; In order to improve resource-sharing efficient; Reduce the complexity of scheduling of resource; DSN multi-service resource manager is classified to the Business Entity of outside according to type of service, for each type Business Entity is safeguarded corresponding virtual machine resource, and the mapping of the resource of managing virtual machines and physical server.
The outside service entity is positioned at the Core Feature layer, is used to provide the service application scene, to all kinds of telecommunication applications softwares calling interface is provided on the Core Feature course.For example, with the networking telephone (VoIP, Voice over Internet Protocol) be the content sharing class calling service interface etc. of representative for the voice class calling service interface of representative, with Streaming Media (Streaming).
Core Feature layer inner utilization point-to-point transmission (P2P; Peer to Peer) distributed technology; The a large number of users request of self terminal in the future is distributed to each peer node in stack (Overlay) network that comprises distributed sound exchange and distributed content exchange; Each peer node is used the virtual resource that is provided by infrastructure layer, and promptly virtual machine provides service.
Type of service comprises: the distributed reciprocal exchange of business of DSN and content exchange are professional.
DSN multi-service resource manager is positioned at the infrastructure layer below the Core Feature layer; Infrastructure course Core Feature layer provides abstract network abilities such as calculating, storage, scheduling; Utilize system-level Intel Virtualization Technology to realize flexible division to the physical server resource with the virtual machine mode; And can realize fine-grained resource dynamic scheduling decision and enforcement between different business, different virtual machine, the different physical server by technology such as no shutdown virtual machine (vm) migration, resource flows, form the virtual resource scheduling system of the embodiment of the invention.
DSN multi-service resource manager also is used to obtain the Business Entity load information, implements the user and asks shunting and adjustment.That is to say that DSN multi-service resource manager is in charge of the dynamic adding of peer node in the Core Feature layer/withdraw from; According to the resource dispatching strategy that receives,, implement the user and ask shunting and adjustment simultaneously, realize the load balancing and the disaster tolerance mechanism of application layer according to the real-time load of each peer node.
The scheduling of resource decision-making device is used to provide user interface, and virtual resource scheduling scheme is calculated and obtained to the receiving scheduling request according to the resource requirement information of virtual machine and physical server in the current system,, is sent to the distributed virtual machine Resource Scheduler;
In the embodiment of the invention, dispatch request is from system manager or Core Feature layer, and virtual resource scheduling scheme is the mapping adjustment strategy of physical resource to resources of virtual machine.
The scheduling of resource decision-making device is through controlled trigger mechanism and execution mechanism; In operator is each physical server of disposing of core net or implement virtual resource scheduling scheme between the physical server, concrete virtual resource scheduling scheme comprises: local resource flows and strange land resource flow etc.The strange land resource flow is just under the prerequisite that keeps existing system virtual machine and physical server mapping relations; The resource that physical server is idle is dispatched between each virtual machine; For example, can the resource that physical server A is idle between physical server A and physical server B corresponding virtual machine, dispatch.
Virtual resource scheduling scheme in the embodiment of the invention can be used as the part of infrastructure layer in the DSN fusion architecture (virtual resource scheduling system), is used to accomplish P2P business such as DSN VoIP, DSN Streaming are striden physical server on the unified resource pond that the DSN fusion architecture provides global resource dynamic dispatching.
The scheduling of resource decision-making device comprises: mapping block and computing module, wherein,
Mapping block; Be used to provide user interface, the receiving scheduling request is mapped as the fundamental plan model with the actual schedule scene; According to the resource requirement information of virtual machine and physical server in the current system, confirm optimizing scheduling target and corresponding network description parameter sets;
Computing module is used for the network description parameter sets definite according to mapping block, calculates and obtain the virtual resource scheduling scheme of resource flow, is sent to the distributed virtual machine Resource Scheduler.
The distributed virtual machine Resource Scheduler is used for the physical server resource flow information that comprises according to the virtual resource scheduling scheme that receives, the resource of scheduling related physical server;
In the embodiment of the invention,, specifically can repeat no more at this referring to the correlation technique document according to virtual resource scheduling scheme scheduling resource.
The distributed virtual machine Resource Scheduler comprises: monitoring modular and enforcement module, wherein,
Monitoring modular is used for extracting the required parameter of fundamental plan model and being used to determine that mapping block adopts the state parameter of fundamental plan problem model from the real network environment, comprises parameter acquiring submodule and status monitoring submodule,
The parameter acquiring submodule is used for extracting required limiting parameter and the cost parameter of plan model from the real network environment;
The status monitoring submodule is used for accomplishing from the real network environment and extracts the state parameter that is used to determine mapping block employing fundamental plan problem model, i.e. current available resource distribution situation.
Implement module, be used for the physical server resource flow information that comprises according to the virtual resource scheduling scheme that receives, the resource of scheduling related physical server.
Virtual machine is used for obtaining scheduling resource from the physical server of mapping, and the virtual resource of scheduling is provided for the outside service entity through DSN multi-service resource manager;
Physical server is positioned at physical layer, is used for being virtualized into a plurality of separate virtual machines through the operation virtualization software, and the capacity information that comprises according to virtual resource scheduling scheme is set up the chnnels of resources of this capacity with the respective physical server.
In the embodiment of the invention, a plurality of virtual machines of operation on each physical server, operation is used for logic function softwares such as operation system Request Processing in each virtual machine.
In the embodiment of the invention, can be that the system manager is artificial initiate the dispatch request that triggers virtual scheduling of resource, also can be to trigger automatically according to Provisioning Policy in advance.
Fig. 3 is the method first embodiment schematic flow sheet of the virtual scheduling of resource of the present invention.Referring to Fig. 3, this flow process comprises:
Step 301, the keeper determines the artificial global resource allocation optimized dispatch request of initiating according to actual needs, initiates global resource to DSN multi-service resource manager and dynamically adjusts request (GloSch_req ());
Step 302; DSN multi-service resource manager receives global resource and dynamically adjusts request (GloSch_cal_req (P)), and request resource scheduling decision device calculates global resource dynamic dispatching scheme (overall adjustment scheme D) according to current resource pool available resources distribution situation;
In this step, the available resources distribution situation is meant the available resources on the physical server, can upwards report to DSN multi-service resource manager by the distributed virtual machine Resource Scheduler.Certainly, in the practical application, can also the combined with virtual machine and the resource requirement information of physical server, or other relevant information calculates, about calculating global resource dynamic dispatching scheme, follow-uply be described in detail again.
Step 303, scheduling of resource decision-making device are called the optimized dispatching algorithm that is provided with in advance, calculate global resource dynamic dispatching scheme, and are sent to the distributed virtual machine Resource Scheduler;
In this step, about the optimized dispatching algorithm, follow-uply be described in detail, the scheduling of resource decision-making device is sent to the distributed virtual machine Resource Scheduler through GloFlow_req () message with global resource dynamic dispatching scheme again.
Step 304, distributed virtual machine Resource Scheduler be according to the global resource dynamic dispatching scheme that receives, and initiates the corresponding physical server resource instruction (VM_flow (u, v, vol)) that flows to physical server;
In this step, the distributed virtual machine Resource Scheduler with global resource dynamic dispatching scheme split into to the physical server between the physical resource instruction set that flows, send to corresponding physical server to carrying out.In the resource flow instruction, u, v are the physical server sign, and vol is a capacity of setting up resource flow between physical server u and the physical server v.
Step 305; Physical server receives resource flow instruction (VM_flow (u; The information of v, vol)), carrying according to instruction; Between physical server u and physical server v, setting up capacity is the resource flow channel of vol, and to the information of distributed virtual machine Resource Scheduler feedback operation success or not (Success/Failure);
In this step, feed back information to the distributed virtual machine Resource Scheduler through (VM_flow_confirm ()) message.
Step 306, the distributed virtual machine Resource Scheduler receives information, revises the available resources statistics, and returns the adjustment object information to DSN multi-service resource manager;
In this step, the adjustment object information is the information of reception.
Step 306, DSN multi-service resource manager receive the adjustment object information, revise the available resources statistics, and return the adjustment object information to the keeper.
Fig. 4 is the method second embodiment schematic flow sheet of the virtual scheduling of resource of the present invention.Referring to Fig. 4, this flow process comprises:
Step 401, DSN multi-service resource manager are periodically monitored each physical server available resources situation P of resource pool, computational resource allocation-business demand matching degree;
In this step, periodically its physical server resource allocation of monitoring and operating position matching degree are judged by DSN multi-service resource manager and to be triggered global resource dynamic dispatching request whether automatically.The resource allocation that calculates-business demand matching degree is a dispatching effect, can take all factors into consideration through resource utilization and professional benefit two aspects.Concrete COMPREHENSIVE CALCULATING method is not limit, and can manually be confirmed according to the resource pool concrete condition by the keeper, specifically can repeat no more at this referring to the correlation technique document.
Step 402 judges that resource allocation-business demand matching degree less than the threshold value that is provided with in advance, triggers global resource allocation optimized dispatch request (GloSch_cal_req (P));
Step 403, scheduling of resource decision-making device are called the optimization optimized dispatching algorithm that is provided with in advance, calculate global resource dynamic dispatching scheme, and are sent to the distributed virtual machine Resource Scheduler;
In this step, the scheduling of resource decision-making device is sent to the distributed virtual machine Resource Scheduler through GloFlow_req () message with global resource dynamic dispatching scheme.
Step 404, distributed virtual machine Resource Scheduler be according to the global resource dynamic dispatching scheme that receives, and initiates the corresponding physical server resource instruction (VM_flow (u, v, vol)) that flows to physical server;
Step 405; Physical server receives resource flow instruction (VM_flow (u; The information of v, vol)), carrying according to instruction; Between physical server u and physical server v, setting up capacity is the resource flow channel of vol, and to the information of distributed virtual machine Resource Scheduler feedback operation success or not (Success/Failure);
In this step, feed back information to the distributed virtual machine Resource Scheduler through (VM_flow_confirm ()) message.
Step 406, the distributed virtual machine Resource Scheduler receives information, revises the available resources statistics, and returns the adjustment object information to DSN multi-service resource manager.
After the resource requirement information of virtual machine and physical server is confirmed virtual resource scheduling scheme in according to current resource pool available resources distribution situation or current system, if a plurality of physical server can both provide resource requirement, as previously mentioned; The embodiment of the invention is because resource bottleneck present position different in the resource pool under heterogeneous networks scale and the network environment; For example, for the bigger situation of Business Entity resources requirement, its resource bottleneck present position possibly be the physical server side; The different stipulations modes that possibly cause set planning problem; Thereby obtain the different special cases of this problem, and, even network environment is given; Along with the difference of bearer service type and the dynamic change of load in real time, the resource bottleneck present position also possibly moved.Therefore, in the embodiment of the invention, further provide a kind of, thereby conciliate realization balance between the annual reporting law complexity in the expectation dispatching effect for enough accurate abstract problem modeling method under the specific environment.The expectation dispatching effect can be carried out scheduling of resource resource-demand matching degree afterwards according to result of calculation and weighed; Can take all factors into consideration through resource utilization and professional benefit two aspects; Concrete COMPREHENSIVE CALCULATING method is not limit, and can manually be confirmed according to the resource pool concrete condition by the keeper.
For this reason; The mapping block of the embodiment of the invention and the computing module core concept when accomplishing the particular problem stipulations and finding the solution is in the dispatch application of real resource pond, at first selects for use with the planning that the actual match degree can be accepted and complexity is minimum and separates annual reporting law, makes every effort to shorten scheduling scheme computing time; When in case dispatching effect surpasses the tolerable bottom line that is provided with in advance; Dispatching effect can ask reject rate to characterize through physical resource service efficiency and user, just under the prerequisite that the model input parameter does not change, carry out the overall scheduling operation repeatedly after; Dispatching effect still be lower than be provided with in advance threshold value more progressively refinement for the scheduling problem of resource pool environment portrayal model; Perhaps, through the keeper when the resource pool environment takes place significantly to change, for example; New business online implementing, number of servers variation, network environment adjustment etc., manual adjustment problem portrayal model.Thereby come the optimized dispatching effect through the higher planning derivation algorithm of the pairing complexity of problem model after switching to refinement.
For example, can suppose that in small-scale, single service resources pond network interconnection resource is infinitely sufficient, capacity (bandwidth c; Import-restriction) with cost (b; Optimization aim) be 0, at this moment, in considering current system on the basis of the resource requirement information of virtual machine and physical server; Carry out the resource global registration, then promptly adopt greedy matching algorithm calculation optimization scheduling scheme.
Fig. 5 adopts the schematic flow sheet of greedy matching algorithm calculation optimization scheduling scheme for embodiment of the invention resource global registration problem.Referring to Fig. 5, this flow process comprises:
In this step, with each physical server as the network diagram node set, resource production and marketing relation information s
iThat representes i physical server node can confession amount or demand.
Step 502 rearranges formation O with all local resources output node that supply exceed demand by output from big to small;
In this step, the resource physical server node that supply exceed demand is called the output node.
Step 503 rearranges formation I with all local resources consumption node that supply falls short of demand by demand from big to small;
If | s
i|>| s
j|, with { i → j:s
jAdd resource overall scheduling scheme D, upgrade s
i=s
i+ s
j, i is inserted formation I again;
In this step; If the available resources of node i satisfy the remote access demand of node j; Then the decision with " to j j requirement resource being provided by i " writes resource overall scheduling scheme D (j being shifted out the demand formation); Again upgrade the available resources quantity of i, still insert supply formation (formation is taken out with it in the front).
If | s
i|<| s
j|, with { i → j:s
iAdd resource overall scheduling scheme D, upgrade s
j=s
i+ s
j, j is inserted formation O again.
In this step; If the node i available resources are not enough to the remote access demand of support node j; Then the decision with " providing i that the quantity resource can be provided to j by i " writes resource overall scheduling scheme D (i being shifted out the supply formation); Again upgrade the resource quantity that needs of j, still insert demand formation (formation is taken out with it in the front).
Along with the expansion of networking scale, in extensive, single service resources pond of network environment complicacy, further, can consider the cost optimization of the long-range allocation and transportation of resource, divide two kinds of situation of work to consider below.
First kind of situation: network system underloading
When the network system underloading; Can not consider the capacity limit of the long-range allocation and transportation of resource; And select logic pond or physics pond method according to the optimizing scheduling effect; Logic pond method is not considered the influence of switch to grid, and physics pond method is considered the influence of switch to grid, only the allocation and transportation cost is carried out monitoring and estimating.
Like this, can adopt general allocation and transportation cost acquisition methods, at this moment, the further refinement of resource global registration problem, refinement obtains transportation problem model.
Mathematical Modeling and derivation algorithm in the face of transportation problem describes down.
Suppose to have certain goods and materials to need allocation and transportation, the dosage unit of this goods and materials can be weight, packing unit or other.This goods and materials can be supplied in the known m of a having place (physical server), and promptly i=1 is used in the place of production, 2 ..., m representes have n place to need this kind goods and materials, promptly j=1 is used on pin ground, 2 ..., m representes, if m the place of production can the confession amount, promptly output is a
i(i=1,2 ..., m), the requirement on n pin ground, promptly sales volume is b
j(j=1,2 ..., n), the unit goods and materials freight rate from i the place of production to j pin ground is c
IjIf use x
IjThe Board Lot to the goods and materials on j pin ground is allocated and transported in representative from i the place of production, then under the condition of the co-ordination of supply and marketing, make total freight charges expenditure minimum, can be expressed as following transportation problem Mathematical Modeling:
The classic algorithm of transportation problem optimal solution is a table dispatching method, and the O of approximate optimal solution (m+n) greedy algorithm comprises least member method and Vogel method etc.Specific algorithm can repeat no more at this referring to the correlation technique document.
In addition, in the practical application, can also be according to more simple other planning problem models of actual conditions comprehensive selection.For example, simply use between the physical server local area network (LAN) to connect jumping figure and simplify logic pond method, just obtain another special case " critical path problem ".
Mathematical Modeling and derivation algorithm in the face of critical path problem describes down.
The most frequently used algorithm of asking shortest path has two kinds: the one, ask certain any dijkstra's algorithm to beeline between other points; Another kind is to ask on the network diagram matrix algorithm of beeline between any 2, wherein,
Dijkstra's algorithm:
If with d
IjThe distance of two consecutive points i and j supposes that i and j are non-conterminous in the expression network diagram, makes d
Ij=∞, obviously, d
Ii=0, establish Ls
iThe beeline that expression is ordered to i from the s point is then obtained the shortest path from the s point to certain 1 t, promptly ask specified point to (s, t) between shortest path Dijstra algorithm min_distance_pair (t), step is following during with dijkstra's algorithm for G, s:
A, from a s, because of L
Ss=0, this value is labeled in the other little square frame of s, expression s point is label;
B, from the s point, find out in the point adjacent with s, minimum one of distance is made as r, with L
Sr=L
Ss+ d
SrValue be labeled in the other little square frame of r, show also label of a r;
C, from the point of label, find out not label point p of all adjacent, if L is arranged with these points
Sp=min
p{ min
r{ min{L
Ss+ d
SpL
Sr+ d
Rp, then to the p piont mark, and with L
SpValue be labeled in the other little square frame of p point;
D, repeating step C are until the t point obtains till the label.
The matrix algorithm of any point-to-point transmission beeline
If with d
IjThe distance of two consecutive points i and j supposes that i and j are non-conterminous in the expression network diagram, makes d
Ij=∞, obviously, d
Ii=0, the matrix D that obtains thus (0) shows the direct beeline of ordering from the i point to j, but the shortest path of ordering to j from the i point is not necessarily i → j, and step is following when finding the solution any point-to-point transmission shortest path with matrix algorithm:
A1, begin from D (0);
In this step, element d
Ij(0) is the direct beeline of ordering, establishes k=1 from the i point to j.
B1, utilize the new matrix D (k) of D (k-1) structure, provide any 2 beelines when being no more than (2k-1) in the network through the intermediate node numbers.
In this step, method is: make d
Ij(k+1)=min{d
Ir(k-1)+d
Rj(k-1) }, wherein, k=k+1.
D (m+1)=D (m) appears in C1, repeating step B1 when k=m, each element value of matrix D (m) is designated as beeline between each point.
In this step, suppose to have in the network n point, then generally calculate iterations m and satisfy following formula:
Second kind of situation: the network system load is heavy
When system load acquires a certain degree, launch again to obtain and plan and find the solution for the monitoring of capacity parameter.At this moment, with above-mentioned " transportation problem " further refinement, refinement obtains " minimum cost flow problem ", and the Mathematical Modeling of minimum cost flow problem is described with the derivation algorithm explanation is follow-up again, according to resource bottleneck present position difference, is divided into two kinds of situation again:
If the one of which resource bottleneck is the physical server end, then can consider to use logic pond method to increase obtaining and using for the allocation and transportation cost parameter;
If two, resource bottleneck is that network connects, can considers to use physics pond method to increase modeling analysis, and use the analysis of obtaining that physics pond method is used for capacity parameter instead for the allocation and transportation cost parameter.
Like this; Scheduling of resource decision-making device under the DSN resource convergence platform framework uses the monitoring situation according to the resource in the resource pool; Adjustment in real time is for the problem modeling of optimization scheduling of resource, and combination optimizing scheduling result adjusts employed physical planning derivation algorithm with the expectation matching degree.
In the practical application; When class of business increases, the network-type dense traffic forms on a large scale, during the multi-service resource pond; Can further take all factors into consideration concrete restriction of resource pool network environment and remote resource access cost two aspect factors, can consider to adopt " minimum cost flow problem ".For this reason, need at first accomplish the mapping of from actual schedule scene (SNA) to the fundamental plan model (scheduling scheme model), describe respectively below.
One, the mapping of fundamental plan model:
In the embodiment of the invention; In the time of can taking all factors into consideration " network bandwidth limitations " with factors such as " operation system resource consumption demand costs " through following concrete steps; In virtual resource pool, carry out the problem of resources optimization scheduling between each physical server, find the solution through " minimum cost flow " planning problem that each factor is mapped as under the network diagram model.That is to say that the actual schedule scene that will comprise each influencing factor is mapped as the fundamental plan model, this fundamental plan model is the minimum cost flow plan model.
At first; Each physical server in the resource pool is categorized as resource physical server more than needed and the nervous physical server of resource: the optimal scheduling problem of global resource is abstract at some resource output points; Be resource physical server and some resources consumptions point more than needed, promptly resource is allocated and transported problem between the physical server of resource anxiety;
In this step; When overall scheduling mechanism is triggered; DSN multi-service resource manager obtains on each physical server current available resources reserves and each local service virtual machine to the operating position statistical information of Resources allocation; Calculate, obtain the resources requirement s in the network diagram corresponding to the node i of physical server
i
Secondly, it is the flow upper limit of arc between node in the network that the network between the physical server in the cloud computing resource pool is connected bandwidth restriction abstract (mapping) in real time, obtains and definite capacity limit c
Ij
In this step, the method for specifically confirming possibly comprise but be not limited to:
Be subject under the end points server scene in communication performance, do not consider the logic pond method of group-network construction; With,
Be subject under network topology and the interconnect architecture scene in communication performance, take all factors into consideration the physics pond method of group-network construction.
Specify below.
One, bandwidth constraints is to the flow modeling conversion method of resource limit
In the embodiment of the invention, input variable such as the flow upper limit is for adopting the similar resource of unified module, and in concrete resource pool overall scheduling problem on the product resource in the minimum cost flow problem and the arc; Product is to stride the hardware resource that physical server is shared, for example, and internal memory etc.; And use arc stream amount modeling reflection be network connection bandwidth constraints, be not resource of the same race, linear module is also inconsistent; For example, the hardware resource linear module is M, and arc stream metric unit is M/s; For addressing this problem, can adopt following method to convert bandwidth constraints to restriction to long-range shared resource:
(1) utilize the network equipment to obtain any two node i, the real-time network bandwidth constraints d between the j
IjUnder real resource scheduling scene,, can directly use the end-to-end bandwidth of two ends physical server to monitor numerical value in real time as d if any hardware condition
IjInput.
(2), confirm the bandwidth m that its consumer consumes to the data access between the supplier in the unit interval for the long-range shared resource of the unit-sized of being considered.For example, for long-range shared drive, given operation system type and shared drive size can be estimated that the resource user carries out the long-distance inner visit and the network data communication amount of consumption for resource provider within a period of time, thereby estimate m.The method of estimation comprises:
Estimate the remote access frequency through the local internal storage access frequency: the local internal storage access frequency and average each visit data amount/local memory size of obtaining service logic node in the given operation system; Obtain long-distance inner and visit the clean traffic with this two number long-pending long-distance inner size that multiply by again on duty; Again divided by cost on network communication ratio (remote access protocol and IP network data packet head expense), thereby obtain estimating numerical value; Perhaps,
Directly measure under the different long-distance inner allocated size, the statistic of the network access data traffic, after carry out curve fitting, obtain the quantitative relation of the two, calculate according to it again.
(3) any two nodes, for example, physical server node i, flow restriction c between the j
IjConfirm as:
c
ij=d
ij/m
Two, do not consider the logic pond method of group-network construction
Under this situation, the unlimited abundance for current needs of the network interconnection bandwidth in the resource pool between each physical server.The bandwidth constraints of at this moment, striding the physical server shared resource is confirmed according to the smaller value of the bandwidth of self having more than needed of two end points servers.
Three, take all factors into consideration the physics pond method of group-network construction
Under this situation, the concrete group-network construction in comprehensive resources pond is introduced for the bandwidth constraints of inner exchanging machine and is considered.The bandwidth constraints of at this moment, striding the physical server shared resource is confirmed according to the minimum value of the current bandwidth more than needed of each related exchange equipment that the local exchange service is provided between two end points physical servers, network cable.For example, at present be furnished with the traffic monitoring function on the mainstream switches, can monitor the bandwidth in real time more than needed of each port in real time, get that the Monitoring Data minimum value on the network path gets final product between the end points physical server.
Under actual network environment, the traffic monitoring of switching equipment capable of using self mechanism realizes the real-time statistics to link bandwidth assignment d between node and utilization ratio v, thereby, can be directly with d (1-v) numerical value as d
IjInput.
Confirm allocation and transportation cost b
Ij
The 3rd step, the resource consumption cost that the global resource scheduling mechanism is introduced for example, bandwidth consumption for the influence of operation system as node in the network diagram to arc (i, the resource cost of transportation b between j)
Ij
Corresponding aforesaid flow restriction modeling method, confirm that the concrete implementation method of allocation and transportation cost includes but not limited to following two types equally:
Be subject under the end points server scene in communication performance, do not consider the logic pond method of group-network construction; With,
Be subject under network topology and the interconnect architecture scene in communication performance, take all factors into consideration the physics pond method of group-network construction.
Specify below.
One, do not consider the logic pond method of group-network construction
If the unlimited abundance of the relative demand of network interconnection bandwidth in the resource pool between each physical server; The bandwidth consumption cost of at this moment, striding the physical server shared resource is from two end points physical servers self that can supply the operation system local backup to use originally bandwidth more than needed.Concrete grammar is following:
(1) increases the efficiency gains of unit bandwidth and the benefit loss that the local service system reduces unit bandwidth, the long-range allocation and transportation cost of its local unit bandwidth resource of comprehensive estimate according to physical server node local service system.The marginal cost of the long-range allocation and transportation of used operator comprehensive resources, its actual connotation can define as required, and for example, numerical value is taken advantage of, is got higher value, gets smaller value etc.
(2) according to the long-range allocation and transportation costs of the local unit bandwidth resource of two physical server nodes, comprehensive estimate unit resource allocation and transportation cost between the two.Comprehensive two the node local coots of used operator, its actual connotation can define as required, and for example, numerical value is taken advantage of, is got higher value, gets smaller value etc.
Two, take all factors into consideration the physics pond method of group-network construction
Under this situation, the network interconnection limited bandwidth in the resource pool between each physical server.The bandwidth consumption cost of at this moment, striding the physical server shared resource is also from the bandwidth more than needed of corresponding switching equipment and network cable.Specifically, its modeling method is following:
(1) increases the efficiency gains of unit bandwidth and the benefit loss that the local service system reduces unit bandwidth, the long-range allocation and transportation cost of its local unit bandwidth resource of comprehensive estimate according to physical server node local service system.The marginal cost of the long-range allocation and transportation of used operator comprehensive resources, its actual connotation can define as required, and for example, numerical value is taken advantage of, is got higher value, gets smaller value etc.
(2) according to the long-range allocation and transportation costs of the local unit bandwidth resource of two physical server nodes, comprehensive estimate unit resource allocation and transportation cost between the two.Comprehensive two the node local coots of used operator, its actual connotation can define as required, and for example, numerical value is taken advantage of, is got higher value, gets smaller value etc.
(3) according to the local unit bandwidth resource allocation and transportation cost of corresponding physics server node, confirm the unit resource allocation and transportation cost between telephone net node and the adjacent server node.
(4) allocation and transportation of the unit resource between basis and direct neighbor physical server node cost sum comprehensively determines two unit resource allocation and transportation costs between the neighboring switch node.Comprehensive two the node local coots of wherein used operator, its actual connotation can customize as required, and for example, numerical value adds, gets higher value, gets smaller value etc.
Like this, the minimum cost flow problem in the network model that optimal scheduling the question resolves itself into of global resource is set up, related algorithm capable of using, for example, the minimum cost flow derivation algorithm formulate cost performance/the resource allocation and transportation scheme of suboptimum.
Facing minimum cost flow problem and derivation algorithm thereof down describes.
Minimum cost flow problem is described: establishing network has n node (n is a natural number), f
Ij(the expression node i offers the resource quantity of the long-range use of node j for i, the flow on j) for arc; c
IjBe the capacity (real-time bandwidth restriction) of this arc, b
IjFor at arc (i, the expense when j) going up, s through unit discharge
iThat representes i node can confession amount or demand, when i be in the network diagram send out some the time, s
i>0, i.e. resource physical server node more than needed; When i is the sink in the network diagram, s
i<0, i.e. the nervous physical server node of resource; When i is intermediate transit point (other nodes in the network diagram), s
i=0, promptly neither send out the also non-sink of point, for example, the resources requirement of local area network (LAN) InterWorking Equipments such as switch is always zero.
When network needs balance
altogether; With each some physical distribution to each sink (or press maximum stream flow be maneuvered into each sink) from each; Make the minimum problem of total allocation and transportation expense, can be summed up as following linear programming model:
When asking minimum cost flow, on the one hand, can adjust flow through seeking augmenting chain; And differentiate whether reached maximum stream flow; But then, minimum for the expense that guarantees the flow flower that every step is whole, need find out the minimum augmenting chain of each step expense; Flow or maximum stream flow to guarantee finally to provide also are that expense is minimum, and augmenting chain is to make in the network given point to flow adjustment path that sink resource allocation and transportation quantity increases.
If b (f) is the expense of feasible flow f, along the adjusted flow of augmenting chain be f ' (f '>f), corresponding expense is b (f '), then expense difference Δ b (f):
Make ratio [Δ b (f)/θ] minimum, also be about to the minimum augmenting chain of unit of adjustment's flow cost as the minimum augmenting chain of expense.
In the practical application, if when every arc occurred as forward arc or opposite arc, the expense of the unit discharge through this arc is other as the flexible strategy mark at this arc, then seeks the minimum augmenting chain of expense, can be converted into one again and ask the critical path problem of sending out point to sink.Therefore, the above-mentioned step of minimum cost flow of asking can be summed up as follows.
Fig. 6 finds the solution minimum cost flow rudimentary algorithm min_flow (G) schematic flow sheet for the embodiment of the invention.Referring to Fig. 6, this flow process comprises:
Step 601 is from zero stream f
0Beginning;
In this step, f
0Being feasible flow, also is corresponding flow minimum feasible flow of expense when being zero.
Step 602 is to feasible flow f
kStructure weighted network W (f
k);
In this step, structure weighted network W (f
k) specifically comprise:
To o<f
Ij<c
IjArc (i, j), when it was the forward arc, the expense of establishing through unit stream was b
Ij, during for opposite arc, corresponding expense is b
Ji=-b
IjLike this, at weighted network W (f
k) in node i and node j point between, (i is j) with (j, i), (i, j) corresponding flexible strategy are b to arc can to provide arc respectively
Ij, promptly ((j, i) corresponding flexible strategy are b to arc for i, the expense when j) going up through unit discharge at arc
Ji, promptly arc (j, the expense when i) going up through unit discharge, wherein, i≤k, j≤k.
To 0<f
Ij=c
IjArc (i, j) because this arc stream amount has been saturated, can only be in augmenting chain as opposite arc, just at weighted network W (f
k) in, (j, i), promptly (j, i) corresponding flexible strategy are b to arc only to provide arc
Ji
To f
Ij(i j), can only be the forward arc, just at weighted network W (f to=0 arc in augmenting chain
k) in, (i, j), promptly (i, j) corresponding flexible strategy are b to arc only to provide arc
Ij
Step 603 is at the weighted network W (f of structure
k) in, obtain the minimum augmenting chain of expense;
In this step, obtain the minimum augmenting chain of expense, promptly obtain from sending out the shortest path of point to sink.
Step 604, flow is adjusted to the maximum of permission on the augmenting chain that the expense of obtaining is minimum, obtains new flow f
K+1
In this step, flow is adjusted to the maximum of permission on the augmenting chain that the shortest path that obtains is corresponding, obtains a new flow f
K+1, wherein, f
K+1>f
k
Step 605 is returned execution in step 602, until at weighted network W (f
K+m) in do not have augmenting chain, this f
K+mBe required least cost max-flow.
About other explanations of minimum cost flow rudimentary algorithm, specifically can repeat no more at this referring to the correlation technique document.
Fig. 7 is case study of global resource optimized scheduling and the solving model structural representation in the embodiment of the invention resource pool life cycle.Referring to Fig. 7, through adopting adaptive approach, can be used for instructing whole resource pool construction and the global resource optimized scheduling case study within the operation cycle and find the solution, specifically:
When small-scale, single service resources pond, can abstractly be resource matched problem;
When scale enlarges, when the networking complexity increases extensive, the single service resources of formation pond, can abstractly be the transportation optimization problem;
And when class of business increases, the network-intensive type is professional when forming extensive, multi-service resource pond, can abstractly be minimum cost flow problem.
By above-mentioned visible, in the embodiment of the invention, in the virtual resource dispatching patcher, take all factors into consideration under the prerequisite of concrete restriction of resource pool network environment and remote resource access cost, utilize optimization planning thinking to solve global resource optimized scheduling problem;
Different phase to resource pool construction and operation; Through different state being set for scheduling entity; In conjunction with composite factors such as resource pool resource distribution and service deployment, load pressures; The dynamic case study granularity and the stipulations derivation algorithm that are adopted of adjustment, thus balance obtained in dispatching effect and response time, between assessing the cost;
Different to the present position of resource bottleneck, logic pond parameter acquiring method and the physics pond parameter acquiring method under the Internet resources limitation scene under the resource-constrained scene of server end have been proposed respectively.
Have following technological merit:
One of which to the characteristics and the demand of global resource scheduling mechanism in the virtual resource dispatching patcher, is taken all factors into consideration concrete restriction of resource pool network environment and remote resource access cost, sets up the case study that is the basis with " minimum cost flow " problem and finds the solution thinking.
They are two years old; Different phase to resource pool construction and operation; Design management person artificially triggers scene; Implement the self adaptation adjustment of scheduling problem modeling analysis and derivation algorithm, thereby between algorithm complex and dispatching effect, obtain balance: at first select for use and separate annual reporting law with the planning that the actual match degree can be accepted and complexity is minimum and make every effort to shorten scheduling scheme computing time, in case dispatching effect is during above the tolerable bottom line; Utilize again embodiment of the invention method progressively refinement for the scheduling problem of resource pool environment portrayal model, thereby come the optimized dispatching effect through the higher planning derivation algorithm of the pairing complexity of problem model after switching to refinement.
Its three, logic pond parameter acquiring method and the physics pond parameter acquiring method under the Internet resources limitation scene under the resource-constrained scene of server end are proposed.Based on this, different for monitoring modular is provided with state to the present position of resource pool resource bottleneck under the reality operation scene, dynamically adjust employed parameter acquiring method, thus model and find the solution complexity and dispatching effect between obtain Quan Heng.
The above is merely preferred embodiment of the present invention, is not to be used to limit protection scope of the present invention.All within spirit of the present invention and principle, any modification of being done, be equal to replacement and improvement etc., all should be included within protection scope of the present invention.
Claims (20)
1. the method for a virtual scheduling of resource is characterized in that, this method comprises:
Reception is asked from the external management user, or, according to the dispatch request of Provisioning Policy triggering in advance,, calculate and obtain virtual resource scheduling scheme according to the resource requirement information of virtual machine and physical server in the current system;
According to the physical server resource flow information that comprises in the virtual resource scheduling scheme that receives, the resource of scheduling related physical server;
The capacity information that comprises according to virtual resource scheduling scheme is set up the chnnels of resources of this capacity with the respective physical server;
Obtain scheduling resource from the physical server of mapping, the virtual resource of scheduling is provided for the outside service entity.
2. the method for claim 1 is characterized in that, what the resource requirement information of physical server comprised the physical server node can confession amount or demand information, adopts greedy matching algorithm to calculate and obtain virtual resource scheduling scheme.
3. method as claimed in claim 2 is characterized in that, the greedy matching algorithm of said employing calculates and obtains virtual resource scheduling scheme and specifically comprises:
Obtain resource pool physical server node set and resource production and marketing relation information s
i, wherein, s
iThat representes i physical server node can confession amount or demand;
All local resources physical server node that supply exceed demand is rearranged formation O from big to small by output;
All local resources physical server node that supply falls short of demand is rearranged formation I from big to small by demand;
From formation O and formation I, take out physics server node i and physical server node j respectively:
If | s
i|>| s
j|, with { i → j:s
jAdd virtual resource scheduling scheme, upgrade s
i=s
i+ s
j, i is inserted formation I again;
If | s
i|<| s
j|, with { i → j:s
iAdd virtual resource scheduling scheme, upgrade s
j=s
i+ s
j, j is inserted formation O again;
Export virtual resource scheduling scheme.
4. method as claimed in claim 3 is characterized in that, further comprises the long-range allocation and transportation cost information of resource in the resource requirement information of said physical server, adopts the transportation optimized Algorithm to calculate and obtain virtual resource scheduling scheme.
5. method as claimed in claim 4 is characterized in that, said employing transportation optimized Algorithm is calculated and obtained virtual resource scheduling scheme and specifically comprises:
But obtain the resource physical server node funding source a that supply exceed demand
i
Obtain the resource physical server node b that supply falls short of demand
j
Obtain the long-range allocation and transportation cost of the resource c of i resource physical server node that supply exceed demand and j the resource physical server node that supply falls short of demand
Ij
M is the resource physical server node number that supply exceed demand, and n is the resource physical server node number that supply falls short of demand.
6. method as claimed in claim 3 is characterized in that, further comprises the long-range allocation and transportation cost information of resource in the resource requirement information of said physical server, adopts shortest path algorithm to calculate and obtain virtual resource scheduling scheme.
7. method as claimed in claim 3; It is characterized in that; Further comprise long-range allocation and transportation cost information of resource and network system load information in the resource requirement information of said physical server, adopt the minimum cost flow algorithm computation and obtain virtual resource scheduling scheme.
8. method as claimed in claim 7 is characterized in that, when resource bottleneck is the physical server end, obtains the long-range allocation and transportation cost information of resource through logic pond method.
9. method as claimed in claim 7 is characterized in that, when resource bottleneck is that network connects, obtains the long-range allocation and transportation cost information of resource through physics pond method.
10. method as claimed in claim 7 is characterized in that, further comprises network bandwidth limitations information in the resource requirement information of said physical server.
11., it is characterized in that said employing minimum cost flow algorithm computation is also obtained virtual resource scheduling scheme and specifically comprised like each described method of claim 7 to 10:
Each physical server in the resource pool is categorized as resource physical server more than needed and the nervous physical server of resource;
Network between the physical server in the resource pool is connected bandwidth restriction in real time be mapped as in the network flow upper limit of arc between node, obtain and definite capacity limit c
Ij
With bandwidth consumption for the influence of operation system as node in the network diagram to arc (i, the resource cost of transportation b between j)
Ij
Obtain the minimum value of
.
12. method as claimed in claim 11 is characterized in that, when communication performance is subject to physical server, adopts logic pond method to obtain resource cost of transportation information.
13. method as claimed in claim 11 is characterized in that, when communication performance is subject to network topology and interconnect architecture, adopts physics pond method to obtain resource cost of transportation information.
14. method as claimed in claim 11 is characterized in that, obtains said minimum value through the augmenting chain that unit of adjustment's flow cost is minimum as the minimum augmenting chain of expense.
15. a virtual resource scheduling system is characterized in that, this system comprises: distributed business network DSN multi-service resource manager, scheduling of resource decision-making device, distributed virtual machine Resource Scheduler, a plurality of virtual machine and a plurality of physical server, wherein,
DSN multi-service resource manager; Be used for business interface being provided according to the Business Entity of type of service to the outside; The resource mapping of maintenance service type corresponding virtual machine and physical server receives the request from the external management user, or; According to the dispatch request of Provisioning Policy triggering in advance, export the scheduling of resource decision-making device to;
The scheduling of resource decision-making device is used to provide user interface, and virtual resource scheduling scheme is calculated and obtained to the receiving scheduling request according to the resource requirement information of virtual machine and physical server in the current system,, is sent to the distributed virtual machine Resource Scheduler;
The distributed virtual machine Resource Scheduler is used for the physical server resource flow information that comprises according to the virtual resource scheduling scheme that receives, the resource of scheduling related physical server;
Virtual machine is used for obtaining scheduling resource from the physical server of mapping, and the virtual resource of scheduling is provided for the outside service entity through DSN multi-service resource manager;
Physical server is used for being virtualized into a plurality of separate virtual machines through the operation virtualization software, and the capacity information that comprises according to virtual resource scheduling scheme is set up the chnnels of resources of this capacity with the respective physical server.
16. system as claimed in claim 15 is characterized in that, said DSN multi-service resource manager is further used for obtaining the Business Entity load information, implements the user and asks shunting and adjustment.
17. system as claimed in claim 16 is characterized in that, said virtual resource scheduling scheme comprises: do not shut down virtual machine (vm) migration, local resource and flow and the strange land resource flow.
18., it is characterized in that said scheduling of resource decision-making device comprises like each described system of claim 15 to 17: mapping block and computing module, wherein,
Mapping block; Be used to provide user interface, the receiving scheduling request is mapped as the fundamental plan model with the actual schedule scene; According to the resource requirement information of virtual machine and physical server in the current system, confirm optimizing scheduling target and corresponding network description parameter sets;
Computing module is used for the network description parameter sets definite according to mapping block, calculates and obtain the virtual resource scheduling scheme of resource flow, is sent to the distributed virtual machine Resource Scheduler.
19. system as claimed in claim 18 is characterized in that, said distributed virtual machine Resource Scheduler comprises: monitoring modular and enforcement module, wherein,
Monitoring modular is used for extracting the required parameter of fundamental plan model and being used to determine that mapping block adopts the state parameter of fundamental plan problem model from the real network environment;
Implement module, be used for the physical server resource flow information that comprises according to the virtual resource scheduling scheme that receives, the resource of scheduling related physical server.
20. system as claimed in claim 19 is characterized in that, said monitoring modular comprises parameter acquiring submodule and status monitoring submodule, wherein,
The parameter acquiring submodule is used for extracting required limiting parameter and the cost parameter of plan model from the real network environment;
The status monitoring submodule is used for accomplishing from the real network environment and extracts the state parameter that is used to determine mapping block employing fundamental plan problem model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201010621842.0A CN102546379B (en) | 2010-12-27 | 2010-12-27 | Virtualized resource scheduling method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201010621842.0A CN102546379B (en) | 2010-12-27 | 2010-12-27 | Virtualized resource scheduling method and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN102546379A true CN102546379A (en) | 2012-07-04 |
CN102546379B CN102546379B (en) | 2014-12-31 |
Family
ID=46352371
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201010621842.0A Active CN102546379B (en) | 2010-12-27 | 2010-12-27 | Virtualized resource scheduling method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102546379B (en) |
Cited By (52)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102938790A (en) * | 2012-11-20 | 2013-02-20 | 无锡城市云计算中心有限公司 | Resource allocation method of cloud computing system |
CN103607459A (en) * | 2013-11-21 | 2014-02-26 | 东北大学 | Dynamic resource monitoring and scheduling method of cloud computing platform IaaS layer |
WO2014036717A1 (en) * | 2012-09-07 | 2014-03-13 | 运软网络科技(上海)有限公司 | Virtual resource object component |
CN103812789A (en) * | 2013-09-18 | 2014-05-21 | 广东电网公司佛山供电局 | Cloud service resource automatic allocating method and system |
CN103870314A (en) * | 2014-03-06 | 2014-06-18 | 中国科学院信息工程研究所 | Method and system for simultaneously operating different types of virtual machines by single node |
CN103944997A (en) * | 2014-04-29 | 2014-07-23 | 上海交通大学 | Load balancing method with combination of random sampling and virtualization technology |
CN103973465A (en) * | 2013-01-25 | 2014-08-06 | 中国电信股份有限公司 | Distributed cross-platform managing method and system for virtualization capacity |
WO2015018035A1 (en) * | 2013-08-08 | 2015-02-12 | 华为技术有限公司 | Virtualized network allocation method, controller, device and system |
CN104394198A (en) * | 2014-11-10 | 2015-03-04 | 中国电子科技集团公司第二十八研究所 | A global scheduling method based on an ESB |
WO2015027935A1 (en) * | 2013-08-28 | 2015-03-05 | Tencent Technology (Shenzhen) Company Limited | Method and device for allocating computational resources |
CN104407926A (en) * | 2014-12-15 | 2015-03-11 | 成都英力拓信息技术有限公司 | Scheduling method of cloud computing resources |
WO2015035920A1 (en) * | 2013-09-11 | 2015-03-19 | International Business Machines Corporation | Workload deployment with real-time consideration of global network congestion |
CN104572298A (en) * | 2014-12-31 | 2015-04-29 | 四达时代通讯网络技术有限公司 | Video cloud platform resource dispatching method and device |
CN104657194A (en) * | 2013-11-25 | 2015-05-27 | 国际商业机器公司 | Calculating The Effect Of An Action In A Network |
CN105282199A (en) * | 2014-07-14 | 2016-01-27 | 中兴通讯股份有限公司 | Cloud resource optimizing method, cloud platform and cloud user client |
CN105407056A (en) * | 2014-09-16 | 2016-03-16 | 中国电信股份有限公司 | Method and system for establishing service chain in software-defined network |
WO2016058118A1 (en) * | 2014-10-13 | 2016-04-21 | 华为技术有限公司 | Service optimization method, transmission network controller, customer controller and system |
CN105630598A (en) * | 2015-12-29 | 2016-06-01 | 宇龙计算机通信科技(深圳)有限公司 | Data processing method, data processing device and wearable intelligent device |
CN105847428A (en) * | 2016-05-20 | 2016-08-10 | 北京首都在线科技股份有限公司 | Mobile cloud platform |
CN105940377A (en) * | 2014-01-28 | 2016-09-14 | 甲骨文国际公司 | Methods, systems, and computer readable media for a cloud-based virtualization orchestrator |
CN106095581A (en) * | 2016-06-18 | 2016-11-09 | 南京采薇且歌信息科技有限公司 | A kind of network storage virtualization dispatching method under the conditions of privately owned cloud |
CN106227600A (en) * | 2016-07-25 | 2016-12-14 | 湖南大学 | A kind of multidimensional virtual resource allocation method based on Energy-aware |
CN106663021A (en) * | 2014-06-26 | 2017-05-10 | 英特尔公司 | Intelligent gpu scheduling in a virtualization environment |
CN106844058A (en) * | 2017-02-20 | 2017-06-13 | 郑州云海信息技术有限公司 | The management method and device of a kind of virtual resources |
CN106909597A (en) * | 2016-06-30 | 2017-06-30 | 阿里巴巴集团控股有限公司 | A kind of database migration method and apparatus |
CN106970831A (en) * | 2017-05-15 | 2017-07-21 | 金航数码科技有限责任公司 | The resources of virtual machine dynamic scheduling system and method for a kind of facing cloud platform |
CN106982255A (en) * | 2017-03-31 | 2017-07-25 | 联想(北京)有限公司 | Physical server, cloud scheduler, cloud platform and the method for it |
CN107133083A (en) * | 2017-04-13 | 2017-09-05 | 大连理工大学 | A kind of virtual data center resource provider method based on virtualization technology |
CN107203256A (en) * | 2016-03-20 | 2017-09-26 | 田文洪 | Energy-conservation distribution method and device under a kind of network function virtualization scene |
CN107305505A (en) * | 2016-04-20 | 2017-10-31 | 中兴通讯股份有限公司 | The operation method and virtual platform of virtual platform |
CN107423085A (en) * | 2017-04-24 | 2017-12-01 | 北京百度网讯科技有限公司 | Method and apparatus for application deployment |
CN107908459A (en) * | 2017-11-10 | 2018-04-13 | 山东劳动职业技术学院 | System is dispatched in a kind of cloud computing |
CN107967179A (en) * | 2017-12-12 | 2018-04-27 | 山东省计算中心(国家超级计算济南中心) | A kind of cloud computing resources distribution method for supporting emergency |
CN108021441A (en) * | 2016-11-03 | 2018-05-11 | 中国移动通信集团广东有限公司 | A kind of resources of virtual machine collocation method and device based on cloud computing |
CN108574645A (en) * | 2017-03-14 | 2018-09-25 | 华为技术有限公司 | A kind of array dispatching method and device |
CN108848151A (en) * | 2018-06-07 | 2018-11-20 | 广州敏视数码科技有限公司 | A kind of distributed speech recognition and storage management system and method |
CN109189548A (en) * | 2018-08-01 | 2019-01-11 | 平安科技(深圳)有限公司 | Resource regulating method, device, computer equipment and computer readable storage medium |
CN109564528A (en) * | 2017-07-06 | 2019-04-02 | 华为技术有限公司 | The system and method for computational resource allocation in distributed computing |
CN110226166A (en) * | 2017-01-31 | 2019-09-10 | 索尼公司 | For providing the device and method of virtual unit |
CN110333939A (en) * | 2019-06-17 | 2019-10-15 | 腾讯科技(成都)有限公司 | Task mixed scheduling method, device, dispatch server and Resource Server |
CN110502338A (en) * | 2019-07-16 | 2019-11-26 | 阿里巴巴集团控股有限公司 | A kind of performance dilatation prediction technique, calculates equipment and storage medium at device |
CN110532086A (en) * | 2018-05-23 | 2019-12-03 | 阿里巴巴集团控股有限公司 | Resource multiplexing method, equipment, system and storage medium |
CN111158850A (en) * | 2019-12-09 | 2020-05-15 | 国网辽宁省电力有限公司大连供电公司 | Optimization algorithm based on virtual machine iaas layer resource allocation |
CN111182006A (en) * | 2018-11-09 | 2020-05-19 | 阿里巴巴集团控股有限公司 | Method and device for mapping physical cluster into cloud computing resource |
CN111193676A (en) * | 2014-02-05 | 2020-05-22 | 华为技术有限公司 | Method and device for mapping resources in network |
CN111950919A (en) * | 2020-08-19 | 2020-11-17 | 重庆邮电大学 | Industrial virtualization production and service architecture based on network virtualization technology |
US10970129B2 (en) | 2015-09-22 | 2021-04-06 | Intel Corporation | Intelligent GPU scheduling in a virtualization environment |
CN113079062A (en) * | 2021-03-24 | 2021-07-06 | 广州虎牙科技有限公司 | Resource adjusting method and device, computer equipment and storage medium |
TWI744068B (en) * | 2020-11-03 | 2021-10-21 | 啟碁科技股份有限公司 | Resource management system and resource management method |
CN113886185A (en) * | 2021-10-16 | 2022-01-04 | 深圳市瑞云科技有限公司 | Method for automatically judging busy state of platform |
CN114301987A (en) * | 2022-03-07 | 2022-04-08 | 天津市城市规划设计研究总院有限公司 | Dynamic scheduling method and system for virtualized network resources |
US11388082B2 (en) | 2013-11-27 | 2022-07-12 | Oracle International Corporation | Methods, systems, and computer readable media for diameter routing using software defined network (SDN) functionality |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107273179A (en) * | 2016-04-07 | 2017-10-20 | 中国移动通信有限公司研究院 | The dispatching method and device of a kind of hardware resource |
CN107590000B (en) * | 2017-09-06 | 2020-11-10 | 中国科学院上海高等研究院 | Secondary random resource management method/system, computer storage medium and device |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101765120A (en) * | 2010-01-21 | 2010-06-30 | 华为技术有限公司 | Method and a device allocating distribution type spectrum resource |
CN101859317A (en) * | 2010-05-10 | 2010-10-13 | 浪潮电子信息产业股份有限公司 | Method for establishing database cluster by utilizing virtualization |
-
2010
- 2010-12-27 CN CN201010621842.0A patent/CN102546379B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101765120A (en) * | 2010-01-21 | 2010-06-30 | 华为技术有限公司 | Method and a device allocating distribution type spectrum resource |
CN101859317A (en) * | 2010-05-10 | 2010-10-13 | 浪潮电子信息产业股份有限公司 | Method for establishing database cluster by utilizing virtualization |
Cited By (93)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2014036717A1 (en) * | 2012-09-07 | 2014-03-13 | 运软网络科技(上海)有限公司 | Virtual resource object component |
CN103827825A (en) * | 2012-09-07 | 2014-05-28 | 运软网络科技(上海)有限公司 | Virtual resource object component |
CN102938790A (en) * | 2012-11-20 | 2013-02-20 | 无锡城市云计算中心有限公司 | Resource allocation method of cloud computing system |
CN102938790B (en) * | 2012-11-20 | 2015-10-21 | 无锡城市云计算中心有限公司 | Resource allocation methods in cloud computing system |
CN103973465A (en) * | 2013-01-25 | 2014-08-06 | 中国电信股份有限公司 | Distributed cross-platform managing method and system for virtualization capacity |
WO2015018035A1 (en) * | 2013-08-08 | 2015-02-12 | 华为技术有限公司 | Virtualized network allocation method, controller, device and system |
WO2015027935A1 (en) * | 2013-08-28 | 2015-03-05 | Tencent Technology (Shenzhen) Company Limited | Method and device for allocating computational resources |
CN105531970A (en) * | 2013-09-11 | 2016-04-27 | 国际商业机器公司 | Workload deployment with real-time consideration of global network congestion |
US9307018B2 (en) | 2013-09-11 | 2016-04-05 | International Business Machines Corporation | Workload deployment with real-time consideration of global network congestion |
CN105531970B (en) * | 2013-09-11 | 2018-12-14 | 国际商业机器公司 | Method and system for the load that maps out the work in a network |
WO2015035920A1 (en) * | 2013-09-11 | 2015-03-19 | International Business Machines Corporation | Workload deployment with real-time consideration of global network congestion |
US9749402B2 (en) | 2013-09-11 | 2017-08-29 | International Business Machines Corporation | Workload deployment with real-time consideration of global network congestion |
CN103812789A (en) * | 2013-09-18 | 2014-05-21 | 广东电网公司佛山供电局 | Cloud service resource automatic allocating method and system |
CN103607459A (en) * | 2013-11-21 | 2014-02-26 | 东北大学 | Dynamic resource monitoring and scheduling method of cloud computing platform IaaS layer |
CN104657194B (en) * | 2013-11-25 | 2018-08-28 | 国际商业机器公司 | Method and system for calculating the influence acted in network |
CN104657194A (en) * | 2013-11-25 | 2015-05-27 | 国际商业机器公司 | Calculating The Effect Of An Action In A Network |
US11388082B2 (en) | 2013-11-27 | 2022-07-12 | Oracle International Corporation | Methods, systems, and computer readable media for diameter routing using software defined network (SDN) functionality |
CN105940377B (en) * | 2014-01-28 | 2020-01-10 | 甲骨文国际公司 | Methods, systems, and computer readable media for cloud-based virtualization orchestrators |
CN105940377A (en) * | 2014-01-28 | 2016-09-14 | 甲骨文国际公司 | Methods, systems, and computer readable media for a cloud-based virtualization orchestrator |
CN111193676B (en) * | 2014-02-05 | 2023-03-24 | 华为技术有限公司 | Method and device for mapping resources in network |
CN111193676A (en) * | 2014-02-05 | 2020-05-22 | 华为技术有限公司 | Method and device for mapping resources in network |
CN103870314B (en) * | 2014-03-06 | 2017-01-25 | 中国科学院信息工程研究所 | Method and system for simultaneously operating different types of virtual machines by single node |
CN103870314A (en) * | 2014-03-06 | 2014-06-18 | 中国科学院信息工程研究所 | Method and system for simultaneously operating different types of virtual machines by single node |
CN103944997B (en) * | 2014-04-29 | 2015-10-07 | 上海交通大学 | In conjunction with the load-balancing method of random sampling and Intel Virtualization Technology |
CN103944997A (en) * | 2014-04-29 | 2014-07-23 | 上海交通大学 | Load balancing method with combination of random sampling and virtualization technology |
CN106663021B (en) * | 2014-06-26 | 2021-01-19 | 英特尔公司 | Intelligent GPU scheduling in virtualized environments |
CN106663021A (en) * | 2014-06-26 | 2017-05-10 | 英特尔公司 | Intelligent gpu scheduling in a virtualization environment |
CN105282199A (en) * | 2014-07-14 | 2016-01-27 | 中兴通讯股份有限公司 | Cloud resource optimizing method, cloud platform and cloud user client |
CN105407056A (en) * | 2014-09-16 | 2016-03-16 | 中国电信股份有限公司 | Method and system for establishing service chain in software-defined network |
CN105407056B (en) * | 2014-09-16 | 2019-04-26 | 中国电信股份有限公司 | Business chain method for building up and system in a kind of software defined network |
WO2016058118A1 (en) * | 2014-10-13 | 2016-04-21 | 华为技术有限公司 | Service optimization method, transmission network controller, customer controller and system |
US10715390B2 (en) | 2014-10-13 | 2020-07-14 | Huawei Technologies Co., Ltd. | Service optimization method, transport controller, client controller, and system |
CN104394198B (en) * | 2014-11-10 | 2017-08-25 | 中国电子科技集团公司第二十八研究所 | A kind of overall scheduling method based on ESB |
CN104394198A (en) * | 2014-11-10 | 2015-03-04 | 中国电子科技集团公司第二十八研究所 | A global scheduling method based on an ESB |
CN104407926B (en) * | 2014-12-15 | 2017-08-29 | 成都英力拓信息技术有限公司 | A kind of dispatching method of cloud computing resources |
CN104407926A (en) * | 2014-12-15 | 2015-03-11 | 成都英力拓信息技术有限公司 | Scheduling method of cloud computing resources |
CN104572298A (en) * | 2014-12-31 | 2015-04-29 | 四达时代通讯网络技术有限公司 | Video cloud platform resource dispatching method and device |
CN104572298B (en) * | 2014-12-31 | 2017-11-24 | 北京四达时代软件技术股份有限公司 | The resource regulating method and device of video cloud platform |
US10970129B2 (en) | 2015-09-22 | 2021-04-06 | Intel Corporation | Intelligent GPU scheduling in a virtualization environment |
CN105630598B (en) * | 2015-12-29 | 2019-06-11 | 宇龙计算机通信科技(深圳)有限公司 | Data processing method, data processing equipment and wearable smart machine |
CN105630598A (en) * | 2015-12-29 | 2016-06-01 | 宇龙计算机通信科技(深圳)有限公司 | Data processing method, data processing device and wearable intelligent device |
CN107203256B (en) * | 2016-03-20 | 2021-07-30 | 田文洪 | Energy-saving distribution method and device under network function virtualization scene |
CN107203256A (en) * | 2016-03-20 | 2017-09-26 | 田文洪 | Energy-conservation distribution method and device under a kind of network function virtualization scene |
CN107305505A (en) * | 2016-04-20 | 2017-10-31 | 中兴通讯股份有限公司 | The operation method and virtual platform of virtual platform |
CN105847428A (en) * | 2016-05-20 | 2016-08-10 | 北京首都在线科技股份有限公司 | Mobile cloud platform |
CN106095581A (en) * | 2016-06-18 | 2016-11-09 | 南京采薇且歌信息科技有限公司 | A kind of network storage virtualization dispatching method under the conditions of privately owned cloud |
CN106909597B (en) * | 2016-06-30 | 2020-07-03 | 阿里巴巴集团控股有限公司 | Database migration method and device |
CN106909597A (en) * | 2016-06-30 | 2017-06-30 | 阿里巴巴集团控股有限公司 | A kind of database migration method and apparatus |
CN106227600B (en) * | 2016-07-25 | 2019-05-17 | 湖南大学 | A kind of multidimensional virtual resource allocation method based on Energy-aware |
CN106227600A (en) * | 2016-07-25 | 2016-12-14 | 湖南大学 | A kind of multidimensional virtual resource allocation method based on Energy-aware |
CN108021441A (en) * | 2016-11-03 | 2018-05-11 | 中国移动通信集团广东有限公司 | A kind of resources of virtual machine collocation method and device based on cloud computing |
CN110226166B (en) * | 2017-01-31 | 2023-06-16 | 索尼公司 | Apparatus and method for providing virtual devices |
CN110226166A (en) * | 2017-01-31 | 2019-09-10 | 索尼公司 | For providing the device and method of virtual unit |
CN106844058B (en) * | 2017-02-20 | 2020-05-29 | 郑州云海信息技术有限公司 | Management method and device for virtualized resources |
CN106844058A (en) * | 2017-02-20 | 2017-06-13 | 郑州云海信息技术有限公司 | The management method and device of a kind of virtual resources |
CN108574645A (en) * | 2017-03-14 | 2018-09-25 | 华为技术有限公司 | A kind of array dispatching method and device |
CN108574645B (en) * | 2017-03-14 | 2020-08-25 | 华为技术有限公司 | Queue scheduling method and device |
CN106982255A (en) * | 2017-03-31 | 2017-07-25 | 联想(北京)有限公司 | Physical server, cloud scheduler, cloud platform and the method for it |
CN106982255B (en) * | 2017-03-31 | 2020-08-25 | 联想(北京)有限公司 | Physical server, cloud scheduler, cloud platform, and method therefor |
CN107133083B (en) * | 2017-04-13 | 2020-01-17 | 大连理工大学 | Virtual data center resource providing method based on virtualization technology |
CN107133083A (en) * | 2017-04-13 | 2017-09-05 | 大连理工大学 | A kind of virtual data center resource provider method based on virtualization technology |
CN107423085B (en) * | 2017-04-24 | 2020-07-28 | 北京百度网讯科技有限公司 | Method and apparatus for deploying applications |
CN107423085A (en) * | 2017-04-24 | 2017-12-01 | 北京百度网讯科技有限公司 | Method and apparatus for application deployment |
CN106970831A (en) * | 2017-05-15 | 2017-07-21 | 金航数码科技有限责任公司 | The resources of virtual machine dynamic scheduling system and method for a kind of facing cloud platform |
CN106970831B (en) * | 2017-05-15 | 2019-06-11 | 金航数码科技有限责任公司 | A kind of the resources of virtual machine dynamic scheduling system and method for facing cloud platform |
CN109564528B (en) * | 2017-07-06 | 2021-11-19 | 华为技术有限公司 | System and method for computing resource allocation in distributed computing |
CN109564528A (en) * | 2017-07-06 | 2019-04-02 | 华为技术有限公司 | The system and method for computational resource allocation in distributed computing |
US10686728B2 (en) | 2017-07-06 | 2020-06-16 | Huawei Technologies Co., Ltd. | Systems and methods for allocating computing resources in distributed computing |
CN107908459A (en) * | 2017-11-10 | 2018-04-13 | 山东劳动职业技术学院 | System is dispatched in a kind of cloud computing |
CN107908459B (en) * | 2017-11-10 | 2020-06-09 | 山东劳动职业技术学院 | Cloud computing scheduling system |
CN107967179B (en) * | 2017-12-12 | 2021-08-06 | 山东省计算中心(国家超级计算济南中心) | Cloud computing resource allocation method supporting emergency |
CN107967179A (en) * | 2017-12-12 | 2018-04-27 | 山东省计算中心(国家超级计算济南中心) | A kind of cloud computing resources distribution method for supporting emergency |
CN110532086B (en) * | 2018-05-23 | 2022-11-15 | 阿里巴巴集团控股有限公司 | Resource multiplexing method, device, system and storage medium |
CN110532086A (en) * | 2018-05-23 | 2019-12-03 | 阿里巴巴集团控股有限公司 | Resource multiplexing method, equipment, system and storage medium |
CN108848151A (en) * | 2018-06-07 | 2018-11-20 | 广州敏视数码科技有限公司 | A kind of distributed speech recognition and storage management system and method |
CN109189548B (en) * | 2018-08-01 | 2023-05-26 | 平安科技(深圳)有限公司 | Resource scheduling method, device, computer equipment and computer readable storage medium |
WO2020024443A1 (en) * | 2018-08-01 | 2020-02-06 | 平安科技(深圳)有限公司 | Resource scheduling method and apparatus, computer device and computer-readable storage medium |
CN109189548A (en) * | 2018-08-01 | 2019-01-11 | 平安科技(深圳)有限公司 | Resource regulating method, device, computer equipment and computer readable storage medium |
CN111182006A (en) * | 2018-11-09 | 2020-05-19 | 阿里巴巴集团控股有限公司 | Method and device for mapping physical cluster into cloud computing resource |
CN111182006B (en) * | 2018-11-09 | 2022-11-29 | 阿里巴巴集团控股有限公司 | Method and device for mapping physical cluster into cloud computing resource |
CN110333939A (en) * | 2019-06-17 | 2019-10-15 | 腾讯科技(成都)有限公司 | Task mixed scheduling method, device, dispatch server and Resource Server |
CN110333939B (en) * | 2019-06-17 | 2023-11-14 | 腾讯科技(成都)有限公司 | Task mixed scheduling method and device, scheduling server and resource server |
CN110502338A (en) * | 2019-07-16 | 2019-11-26 | 阿里巴巴集团控股有限公司 | A kind of performance dilatation prediction technique, calculates equipment and storage medium at device |
CN110502338B (en) * | 2019-07-16 | 2023-04-11 | 蚂蚁金服(杭州)网络技术有限公司 | Performance capacity expansion prediction method and device, computing equipment and storage medium |
CN111158850A (en) * | 2019-12-09 | 2020-05-15 | 国网辽宁省电力有限公司大连供电公司 | Optimization algorithm based on virtual machine iaas layer resource allocation |
CN111158850B (en) * | 2019-12-09 | 2023-05-02 | 国网辽宁省电力有限公司大连供电公司 | Optimization algorithm based on resource allocation of iaas layer of virtual machine |
CN111950919B (en) * | 2020-08-19 | 2023-12-12 | 重庆邮电大学 | Industrial virtualization production and service system based on network virtualization technology |
CN111950919A (en) * | 2020-08-19 | 2020-11-17 | 重庆邮电大学 | Industrial virtualization production and service architecture based on network virtualization technology |
TWI744068B (en) * | 2020-11-03 | 2021-10-21 | 啟碁科技股份有限公司 | Resource management system and resource management method |
CN113079062A (en) * | 2021-03-24 | 2021-07-06 | 广州虎牙科技有限公司 | Resource adjusting method and device, computer equipment and storage medium |
CN113886185B (en) * | 2021-10-16 | 2023-07-11 | 深圳市瑞云科技有限公司 | Method for automatically judging busy state of platform |
CN113886185A (en) * | 2021-10-16 | 2022-01-04 | 深圳市瑞云科技有限公司 | Method for automatically judging busy state of platform |
CN114301987A (en) * | 2022-03-07 | 2022-04-08 | 天津市城市规划设计研究总院有限公司 | Dynamic scheduling method and system for virtualized network resources |
Also Published As
Publication number | Publication date |
---|---|
CN102546379B (en) | 2014-12-31 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102546379A (en) | Virtualized resource scheduling method and system | |
Islam et al. | A survey on task offloading in multi-access edge computing | |
JP7162385B2 (en) | Multi-User Multi-MEC Task Unload Resource Scheduling Method Based on Edge-Terminal Collaboration | |
Niewiadomska-Szynkiewicz et al. | Dynamic power management in energy-aware computer networks and data intensive computing systems | |
Ghosh et al. | Scalable multi-class traffic management in data center backbone networks | |
Kim et al. | Multi-agent reinforcement learning-based resource management for end-to-end network slicing | |
CN101977242A (en) | Layered distributed cloud computing architecture and service delivery method | |
Amokrane et al. | Greenslater: On satisfying green SLAs in distributed clouds | |
Liu et al. | ERP: Edge resource pooling for data stream mobile computing | |
Qu et al. | Study QoS optimization and energy saving techniques in cloud, fog, edge, and IoT | |
Logeshwaran et al. | The Smart Performance Analysis of Network Scheduling Framework for Mobile Systems in Cloud Communication Networks | |
Tran-Dang et al. | A survey on matching theory for distributed computation offloading in iot-fog-cloud systems: Perspectives and open issues | |
CN103929379A (en) | SDN resource distribution method based on two-sided market multihoming structure | |
CN111010295A (en) | SDN-MEC-based power distribution and utilization communication network task migration method | |
Gąsior et al. | Pareto-optimal Nash equilibrium in capacity allocation game for self-managed networks | |
Zhang et al. | Rate-adaptive fog service platform for heterogeneous IoT applications | |
Xu et al. | Quality of service in manufacturing networks: a service framework and its implementation | |
Tzanakaki et al. | A converged network architecture for energy efficient mobile cloud computing | |
Khoshkholghi et al. | Edge intelligence for service function chain deployment in NFV-enabled networks | |
He et al. | Markov approximation method for optimal service orchestration in IoT network | |
Zheng et al. | Minimizing the latency of embedding dependence-aware SFCs into MEC network via graph theory | |
Nisha et al. | A bilevel programming framework for joint edge resource management and pricing | |
CN103200113B (en) | Take into account the implementation method of operation cost and the two inter-domain flux engineering optimized of transmission performance | |
Abouelela et al. | Multidomain hierarchical resource allocation for grid applications | |
Yu et al. | Network slice mobility for 6g networks by exploiting user and network prediction |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant |