CN106815082A - The method for scheduling task and device of a kind of container - Google Patents
The method for scheduling task and device of a kind of container Download PDFInfo
- Publication number
- CN106815082A CN106815082A CN201710096581.7A CN201710096581A CN106815082A CN 106815082 A CN106815082 A CN 106815082A CN 201710096581 A CN201710096581 A CN 201710096581A CN 106815082 A CN106815082 A CN 106815082A
- Authority
- CN
- China
- Prior art keywords
- server
- task
- mathematical modeling
- user
- state parameter
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
- G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses the method for scheduling task and device of a kind of container, the method includes:The state parameter of server cluster is obtained, state parameter enables expense, the current computing capability of each server, the unit multiplexed transport expense of each user to each server, the mission requirements amount of user including each server;State parameter is converted into the corresponding Mathematical Modeling of linear programming, it is minimum as object function using total cost in Mathematical Modeling, using the maximum load capability of the mission requirements amount of each user and each server as constraints;Mathematical Modeling is solved.Due in the Mathematical Modeling, it is minimum as object function using total cost, using the maximum load capability of the mission requirements amount of each user and each server as constraints, therefore, ensure that all of user task is processed, next is able to ensure that and is distributed to computing capability of the task summation of each server no more than server itself, and the use cost of totality is minimum.
Description
Technical field
The present invention relates to field of cloud computer technology, the method for scheduling task and device of more particularly to a kind of container.
Background technology
Container is the set of interfaces being located in application server between component and platform.With the rise of container technique, more
Start to distribute in the form of with container and dispose come more software systems.In the task processing system based on container, hold
The scheduler of device is responsible for the responsibility that user task is distributed to server.
Due to thering are multiple users and multiple servers participating in specific implementation, how each user task to be distributed to
Which server in current system come perform be scheduler an important process task.If the task scheduling side of scheduler
Method is unreasonable, then the application cost for not only resulting in server is very high, or even in some cases it cannot be guaranteed that all of user appoints
Business is processed or beyond the disposal ability of server.
As can be seen here, it is this area how the rational distribution of substantial amounts of user task to be carried out into treatment to corresponding server
Technical staff's problem demanding prompt solution.
The content of the invention
It is an object of the invention to provide the method for scheduling task and device of a kind of container, for substantial amounts of user task to be closed
The distribution of reason to corresponding server is processed so as to ensure all of user task, and next is able to ensure that and is distributed to each
The task summation of server is no more than the computing capability of server itself, and the use cost of totality is minimum.
In order to solve the above technical problems, the present invention provides a kind of method for scheduling task of container, including:
The state parameter of server cluster is obtained, the state parameter enables expense, each server including each server
Current computing capability, the unit multiplexed transport expense of each user to each server, the mission requirements amount of user;
The state parameter is converted into the corresponding Mathematical Modeling of linear programming, it is minimum with total cost in the Mathematical Modeling
As object function, using the maximum load capability of the mission requirements amount of each user and each server as constraints;
The Mathematical Modeling is solved to obtain corresponding destination server and task sendout.
Preferably, it is described that the Mathematical Modeling is solved to obtain corresponding destination server and task distribution measurer
Body is:
The Mathematical Modeling is converted to the canonical form of linear programming;
The canonical form is solved.
Preferably, it is described solution is carried out to the canonical form to specifically include:
Proportion of the number relative to the total number in the server cluster according to the destination server, by the mark
Quasi- form is converted into the compressed sensing algorithm mathematics model based on the corresponding sparsity constraints of L1 norms;
Using compressed sensing algorithm mathematics model described in compressed sensing Algorithm for Solving.
Preferably, also include:
Store the state parameter, and the destination server and the task sendout.
In order to solve the above technical problems, the present invention also provides a kind of task scheduling apparatus of container, including:
Acquiring unit, the state parameter for obtaining server cluster, the state parameter includes enabling for each server
Expense, the current computing capability of each server, the unit multiplexed transport expense of each user to each server, the mission requirements of user
Amount;
Model construction unit, for the state parameter to be converted into the corresponding Mathematical Modeling of linear programming, the mathematics
It is minimum as object function using total cost in model, made with the maximum load capability of the mission requirements amount of each user and each server
It is constraints;
Computing unit, is distributed for being solved to the Mathematical Modeling with obtaining corresponding destination server and task
Amount.
Preferably, the computing unit tool includes:
Modular converter, the canonical form for the Mathematical Modeling to be converted to linear programming;
Computing module, for being solved to the canonical form.
Preferably, the computing module specifically for the number according to the destination server relative to the server set
The proportion of the total number in group, is converted into the canonical form compressed sensing based on the corresponding sparsity constraints of L1 norms and calculates
Method Mathematical Modeling, and using compressed sensing algorithm mathematics model described in compressed sensing Algorithm for Solving.
Preferably, also include:
Memory cell, for storing the state parameter, and the destination server and the task sendout.
The method for scheduling task and device of container provided by the present invention, one is converted to linearly by the process of task scheduling
The Mathematical Modeling of planning, the result of task scheduling is can obtain by being solved to Mathematical Modeling.In due to the Mathematical Modeling,
It is minimum as object function using total cost, using the maximum load capability of the mission requirements amount of each user and each server as constraint
Condition, therefore, the method ensure that all of user task is processed, and next is able to ensure that and is distributed to each server
Computing capability of the task summation no more than server itself, and the use cost of totality is minimum.
Brief description of the drawings
In order to illustrate more clearly the embodiments of the present invention, the accompanying drawing to be used needed for embodiment will be done simply below
Introduce, it should be apparent that, drawings in the following description are only some embodiments of the present invention, for ordinary skill people
For member, on the premise of not paying creative work, other accompanying drawings can also be obtained according to these accompanying drawings.
Fig. 1 is a kind of flow chart of the method for scheduling task of container provided in an embodiment of the present invention;
Fig. 2 is the flow chart of the method for scheduling task of another container provided in an embodiment of the present invention;
Fig. 3 is a kind of structure chart of the task scheduling apparatus of container provided in an embodiment of the present invention.
Specific embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Site preparation is described, it is clear that described embodiment is only a part of embodiment of the invention, rather than whole embodiments.Based on this
Embodiment in invention, those of ordinary skill in the art are not under the premise of creative work is made, and what is obtained is every other
Embodiment, belongs to the scope of the present invention.
Core of the invention is to provide the method for scheduling task and device of a kind of container, for substantial amounts of user task to be closed
The distribution of reason to corresponding server is processed so as to ensure all of user task, and next is able to ensure that and is distributed to each
The task summation of server is no more than the computing capability of server itself, and the use cost of totality is minimum.
In order that those skilled in the art more fully understand the present invention program, with reference to the accompanying drawings and detailed description
The present invention is described in further detail.
Fig. 1 is a kind of flow chart of the method for scheduling task of container provided in an embodiment of the present invention.As shown in figure 1, container
Method for scheduling task comprise the following steps.
S10:The state parameter of server cluster is obtained, state parameter enables expense, each server including each server
Current computing capability, the unit multiplexed transport expense of each user to each server, the mission requirements amount of user.
S11:State parameter is converted into the corresponding Mathematical Modeling of linear programming, in Mathematical Modeling using total cost it is minimum as
Object function, using the maximum load capability of the mission requirements amount of each user and each server as constraints.
S12:Mathematical Modeling is solved to obtain corresponding destination server and task sendout.
In specific implementation, first it is to be understood that the state parameter of server cluster, describes, hereafter for the ease of use formula
It is middle that these parameters are indicated using corresponding letter.
If the expense that enables of the constant expense that i-th server is produced when being activated, i.e. server is ai, user j to clothes
The unit multiplexed transport expense of business device i is cij, the current computing capability of server is bi, the mission requirements amount of user j is dj.Order
Variable
Make variable xijExpression is transferred to the multiplexed transport amount of server i from user j, then the corresponding Mathematical Modeling of linear programming
For:
Wherein, first group of constraints, i.e.,Represent that the calculating demand of each user must is fulfilled for;The
Two groups of constraintss, i.e.,Represent the maximum load capability that can not exceed server i.
As can be seen here, the scheduling problem of container can be just converted to by the corresponding number of linear programming by step S11
Model is learned, can obtain destination server and task sendout by being solved to above-mentioned Mathematical Modeling.It should be noted that
Goal server is the server being selected in server cluster, i.e., the server for being obtained in solution, multiplexed transport amount
It is exactly xijSolving result.
In addition, the solution to above-mentioned Mathematical Modeling has diversified forms, it is contemplated that the problem of computational efficiency, as preferred
Implementation method, step S12 is specially:
Mathematical Modeling is converted to the canonical form of linear programming;
Canonical form is solved.
After the canonical form of linear programming being converted to above-mentioned Mathematical Modeling, it is possible to according to the solution side to canonical form
Formula is entered to solve.It is understood that being prior art for the solution procedure of canonical form, the present embodiment is repeated no more.Linearly
The canonical form of planning is:
Wherein, x0It is the f in above formula, x is xi,jAnd yiThe vector of composition, A is cijAnd aiThe matrix of composition, b is biComposition
Vector.
In specific implementation, because the number of servers in server cluster is larger, and user mission requirements amount
More, operand is also larger so in canonical form calculating process, undoubtedly increased the time of calculating.Therefore above-mentioned
On the basis of embodiment, solution is carried out to canonical form and is specifically included:
Proportion of the number relative to the total number in server cluster according to destination server, canonical form is converted into
Compressed sensing algorithm mathematics model based on the corresponding sparsity constraints of L1 norms;
Using compressed sensing Algorithm for Solving compressed sensing algorithm mathematics model.
On the basis of above formula, it is considered in real production environment, to the mission requirements amount of each user, clothes are only distributed to
A small amount of node operation in business device cluster, i.e., for whole server cluster, only a small amount of server in operation, that
X exists openness, is represented using L1 norms, that is, need to meet:
min|x|1
Thus, above linear programming problem can be converted into the compressed sensing problem based on L1 sparsity constraints:
Wherein, ∈ be substantially equal to 0 number.
So far, container Mission Scheduling just can use compressed sensing Algorithm for Solving.It is understood that compressed sensing is calculated
Method is prior art, therefore, the present embodiment is repeated no more.
The method for scheduling task of the container that the present embodiment is provided, linear programming is converted to by the process of task scheduling
Mathematical Modeling, the result of task scheduling is can obtain by being solved to Mathematical Modeling.In due to the Mathematical Modeling, taken with total
With minimum as object function, the maximum load capability using the mission requirements amount of each user and each server as constraints,
Therefore, the method ensure that all of user task is processed, and next is able to ensure that and is distributed to appointing for each server
Business summation is no more than the computing capability of server itself, and the use cost of totality is minimum.
Fig. 2 is the flow chart of the method for scheduling task of another container provided in an embodiment of the present invention.As shown in Fig. 2 making
Preferred embodiment, on the basis of above-described embodiment, also to include:
S20:Storage state parameter, and destination server and task sendout.
For the ease of the corresponding relation that the later stage is checked between each user and each server, and statistical server is appointed
Business execution amount, in the present embodiment, by state parameter, i.e., each server enable expense, the current computing capability of each server,
Each user is to the unit multiplexed transport expense of each server, the mission requirements amount of user and destination server and task sendout
Store, so can at any time transfer examination in the later stage.
Method for scheduling task with said vesse is corresponding, invention additionally discloses a kind of task scheduling of container.Due to
The embodiment of device part is mutually corresponding with the embodiment of method part, therefore the embodiment of device part refers to method part
Embodiment description, wouldn't repeat here.Fig. 3 is a kind of knot of the task scheduling apparatus of container provided in an embodiment of the present invention
Composition.As shown in figure 3, the task scheduling apparatus of container include:
Acquiring unit 10, the state parameter for obtaining server cluster, state parameter enables expense including each server
With, the current computing capability of each server, the unit multiplexed transport expense of each user to each server, the mission requirements of user
Amount;
Model construction unit 11, for state parameter to be converted into the corresponding Mathematical Modeling of linear programming, in Mathematical Modeling
It is minimum as object function using total cost, using the maximum load capability of the mission requirements amount of each user and each server as constraint
Condition;
Computing unit 12, for being solved to obtain corresponding destination server and task sendout to Mathematical Modeling.
Preferably implementation method, computing unit tool includes:
Modular converter, the canonical form for Mathematical Modeling to be converted to linear programming;
Computing module, for being solved to canonical form.
Preferably implementation method, computing module specifically for according to the number of destination server relative to server set
The proportion of the total number in group, the compressed sensing algorithm number based on the corresponding sparsity constraints of L1 norms is converted into by canonical form
Model is learned, and uses compressed sensing Algorithm for Solving compressed sensing algorithm mathematics model.
Preferably implementation method, also includes:
Memory cell, for storage state parameter, and destination server and task sendout.
The task scheduling apparatus of the container that the present embodiment is provided, linear programming is converted to by the process of task scheduling
Mathematical Modeling, the result of task scheduling is can obtain by being solved to Mathematical Modeling.In due to the Mathematical Modeling, taken with total
With minimum as object function, the maximum load capability using the mission requirements amount of each user and each server as constraints,
Therefore, the device ensure that all of user task is processed, and next is able to ensure that and is distributed to appointing for each server
Business summation is no more than the computing capability of server itself, and the use cost of totality is minimum.
The method for scheduling task and device to container provided by the present invention are described in detail above.It is each in specification
Individual embodiment is described by the way of progressive, and what each embodiment was stressed is the difference with other embodiment, respectively
Between individual embodiment identical similar portion mutually referring to.For device disclosed in embodiment, due to itself and embodiment
Disclosed method is corresponding, so description is fairly simple, related part is referring to method part illustration.It should be pointed out that right
For those skilled in the art, under the premise without departing from the principles of the invention, the present invention can also be carried out
Some improvement and modification, these are improved and modification is also fallen into the protection domain of the claims in the present invention.
Claims (8)
1. a kind of method for scheduling task of container, it is characterised in that including:
Obtain server cluster state parameter, the state parameter include each server enable expense, each server work as
Preceding computing capability, the unit multiplexed transport expense of each user to each server, the mission requirements amount of user;
The state parameter is converted into the corresponding Mathematical Modeling of linear programming, in the Mathematical Modeling using total cost it is minimum as
Object function, using the maximum load capability of the mission requirements amount of each user and each server as constraints;
The Mathematical Modeling is solved to obtain corresponding destination server and task sendout.
2. the method for scheduling task of container according to claim 1, it is characterised in that described to be carried out to the Mathematical Modeling
Solution is specially with obtaining corresponding destination server and task sendout:
The Mathematical Modeling is converted to the canonical form of linear programming;
The canonical form is solved.
3. the method for scheduling task of container according to claim 2, it is characterised in that described to be carried out to the canonical form
Solution is specifically included:
Proportion of the number relative to the total number in the server cluster according to the destination server, by the canonical form
Formula is converted into the compressed sensing algorithm mathematics model based on the corresponding sparsity constraints of L1 norms;
Using compressed sensing algorithm mathematics model described in compressed sensing Algorithm for Solving.
4. the method for scheduling task of container according to claim 1, it is characterised in that also include:
Store the state parameter, and the destination server and the task sendout.
5. a kind of task scheduling apparatus of container, it is characterised in that including:
Acquiring unit, the state parameter for obtaining server cluster, the state parameter include each server enable expense,
The current computing capability of each server, the unit multiplexed transport expense of each user to each server, the mission requirements amount of user;
Model construction unit, for the state parameter to be converted into the corresponding Mathematical Modeling of linear programming, the Mathematical Modeling
In it is minimum as object function using total cost, the maximum load capability using the mission requirements amount of each user and each server is used as about
Beam condition;
Computing unit, for being solved to obtain corresponding destination server and task sendout to the Mathematical Modeling.
6. task scheduling apparatus of container according to claim 5, it is characterised in that the computing unit tool includes:
Modular converter, the canonical form for the Mathematical Modeling to be converted to linear programming;
Computing module, for being solved to the canonical form.
7. task scheduling apparatus of container according to claim 6, it is characterised in that the computing module is specifically for root
According to the proportion of the number relative to the total number in the server cluster of the destination server, by canonical form conversion
It is the compressed sensing algorithm mathematics model based on the corresponding sparsity constraints of L1 norms, and using described in compressed sensing Algorithm for Solving
Compressed sensing algorithm mathematics model.
8. task scheduling apparatus of container according to claim 5, it is characterised in that also include:
Memory cell, for storing the state parameter, and the destination server and the task sendout.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710096581.7A CN106815082A (en) | 2017-02-22 | 2017-02-22 | The method for scheduling task and device of a kind of container |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710096581.7A CN106815082A (en) | 2017-02-22 | 2017-02-22 | The method for scheduling task and device of a kind of container |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106815082A true CN106815082A (en) | 2017-06-09 |
Family
ID=59111468
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710096581.7A Pending CN106815082A (en) | 2017-02-22 | 2017-02-22 | The method for scheduling task and device of a kind of container |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106815082A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110049130A (en) * | 2019-04-22 | 2019-07-23 | 北京邮电大学 | A kind of service arrangement and method for scheduling task and device based on edge calculations |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103699446A (en) * | 2013-12-31 | 2014-04-02 | 南京信息工程大学 | Quantum-behaved particle swarm optimization (QPSO) algorithm based multi-objective dynamic workflow scheduling method |
CN105045656A (en) * | 2015-06-30 | 2015-11-11 | 深圳清华大学研究院 | Virtual container based big data storage and management method |
CN105072182A (en) * | 2015-08-10 | 2015-11-18 | 北京佳讯飞鸿电气股份有限公司 | Load balancing method, load balancer and user terminal |
-
2017
- 2017-02-22 CN CN201710096581.7A patent/CN106815082A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103699446A (en) * | 2013-12-31 | 2014-04-02 | 南京信息工程大学 | Quantum-behaved particle swarm optimization (QPSO) algorithm based multi-objective dynamic workflow scheduling method |
CN105045656A (en) * | 2015-06-30 | 2015-11-11 | 深圳清华大学研究院 | Virtual container based big data storage and management method |
CN105072182A (en) * | 2015-08-10 | 2015-11-18 | 北京佳讯飞鸿电气股份有限公司 | Load balancing method, load balancer and user terminal |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110049130A (en) * | 2019-04-22 | 2019-07-23 | 北京邮电大学 | A kind of service arrangement and method for scheduling task and device based on edge calculations |
CN110049130B (en) * | 2019-04-22 | 2020-07-24 | 北京邮电大学 | Service deployment and task scheduling method and device based on edge computing |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109947567B (en) | Multi-agent reinforcement learning scheduling method and system and electronic equipment | |
CN104317658B (en) | A kind of loaded self-adaptive method for scheduling task based on MapReduce | |
US9099866B2 (en) | Apparatus, methods and systems for parallel power flow calculation and power system simulation | |
CN103049330B (en) | A kind of trustship type distributed task dispatching method and system | |
CN107621973A (en) | A kind of method for scheduling task and device across cluster | |
CN105718364A (en) | Dynamic assessment method for ability of computation resource in cloud computing platform | |
CN101685480A (en) | Parallel computing method for security and stability analysis of large power grid and computing platform | |
Zimmerman et al. | Matpower optimal scheduling tool most 1.0 user’s manual | |
Makhloufi et al. | Three powerful nature-inspired algorithms to optimize power flow in Algeria's Adrar power system | |
CN109240814A (en) | A kind of deep learning intelligent dispatching method and system based on TensorFlow | |
Scholz-Reiter et al. | Robust capacity allocation in dynamic production networks | |
Kaliaev | Multiagent approach for building distributed adaptive computing system | |
CN110311965A (en) | Method for scheduling task and system under a kind of cloud computing environment | |
CN116991558B (en) | Computing power resource scheduling method, multi-architecture cluster, device and storage medium | |
Ding et al. | A bi-objective DC-optimal power flow model using linear relaxation-based second order cone programming and its Pareto Frontier | |
CN115277692B (en) | Automatic operation and maintenance method, device and system for edge network computing terminal equipment | |
Jin et al. | Power grid simulation applications developed using the GridPACK™ high performance computing framework | |
CN106815082A (en) | The method for scheduling task and device of a kind of container | |
CN107729218A (en) | A kind of system and method for monitoring processing computing resource equipment | |
CN113807593A (en) | Data quantity and data quality prediction method and system based on power dispatching | |
Bilek et al. | Development of an agent-based workbench supporting collaborative structural design | |
Papaefthymiou et al. | Impact of stochastic generation in power systems contingency analysis | |
CN108683185B (en) | Disconnection impedance calculation apparatus, disconnection impedance calculation system, disconnection impedance calculation method, computer device, and storage medium | |
CN113285823A (en) | Business function chain arranging method based on container | |
Lee et al. | Relocatable service composition based on microservice architecture for cloud-native IoT-cloud services |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170609 |
|
RJ01 | Rejection of invention patent application after publication |