CN108984307A - Calculating task moving method and calculating task transference apparatus - Google Patents
Calculating task moving method and calculating task transference apparatus Download PDFInfo
- Publication number
- CN108984307A CN108984307A CN201810798448.0A CN201810798448A CN108984307A CN 108984307 A CN108984307 A CN 108984307A CN 201810798448 A CN201810798448 A CN 201810798448A CN 108984307 A CN108984307 A CN 108984307A
- Authority
- CN
- China
- Prior art keywords
- calculating task
- migration
- current
- transition matrix
- cost
- 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
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
- G06F9/505—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 considering the load
-
- 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]
- G06F9/5083—Techniques for rebalancing the load in a distributed system
- G06F9/5088—Techniques for rebalancing the load in a distributed system involving task migration
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 present invention discloses a kind of calculating task moving method and calculating task transference apparatus.The calculating task moving method includes: step S1, obtains each calculating task transition matrix, and the calculating task transition matrix includes current migration distance cost, current migration flow cost and current migration load balancing degrees；Step S2, Optimal calculation task immigration matrix is determined according to the current migration distance cost of each calculating task transition matrix, current migration flow cost and current migration load balancing degrees；Step S3, the Optimal calculation task immigration matrix is exported.Calculating task moving method and calculating task transference apparatus provided by the invention, by determining Optimal calculation task immigration matrix, so that it is determined that going out Optimal calculation task immigration strategy corresponding with Optimal calculation task immigration matrix, low time delay, low discharge cost and the load balancing of calculating task migration are realized.
Description
Technical field
The present invention relates to calculating task migrating technology field, in particular to a kind of calculating task moving method and calculating task
Transference apparatus.
Background technique
As Internet of Things is grown rapidly, the quantity of edge termination equipment increases sharply, while produced by edge termination equipment
Data volume has reached damp byte (ZB) rank.Magnanimity caused by edge termination equipment cannot be effectively treated in centralized data processing
Data, edge calculations are generally regarded as one of the main trend of Next Generation of Digital transition by industry.Mobile edge calculations
(Mobile Edge Computing, referred to as: MEC) is for moving to access for the part calculating task of traditional cloud computing platform
Domain, and traditional business and Internet service are subjected to depth integration, the end-to-end time delay that traditional business is delivered is reduced, and then to fortune
The running of battalion quotient brings new mode, and establishes completely new industrial chain and the ecosphere.In this case, the side in face of being increasingly urgent to
Edge calculates and intelligent industrial internet development demand, and the disaster relief calculating task migration optimization based on edge calculations is for edge
It calculates and the rapid sustainable development of disaster relief is of great significance.
In the prior art, with edge calculations and the rapid growth of disaster relief business, with generation the high time delay of migration,
The problems such as load imbalance, high flow capacity cost, becomes increasingly conspicuous.Existing cloud computing system have migrate high time delay, load imbalance,
The characteristics such as high flow capacity cost do not fully take into account high time delay, high flow capacity cost, load imbalance etc. problem.
Summary of the invention
The present invention is directed at least solve one of above-mentioned the technical problems existing in the prior art, a kind of calculating task is provided and is moved
Shifting method and calculating task transference apparatus, for realizing low time delay, low discharge cost and the load balancing of calculating task migration.
To achieve the above object, the present invention provides a kind of calculating task moving method, the calculating task moving method packets
It includes:
Step S1, each calculating task transition matrix is obtained, the calculating task transition matrix includes current migration distance
Cost, current migration flow cost and current migration load balancing degrees；
Step S2, according to the current migration distance cost of each calculating task transition matrix, current migration flow cost
Optimal calculation task immigration matrix is determined with current migration load balancing degrees；
Step S3, the Optimal calculation task immigration matrix is exported.
Optionally, step S2 includes:
Step S21, according to the current migration distance cost of each calculating task transition matrix, current migration flow cost
The current judgement parameter of each calculating task transition matrix is generated with current migration load balancing degrees；
Step S22, the current judgement parameter for comparing each calculating task transition matrix determines currently there is minimum judgement
The calculating task transition matrix of parameter；
Step S23, according to currently have minimum judge parameter calculating task transition matrix current migration distance at
Originally, current migration flow cost and current migration load balancing degrees generate the calculating task that parameter is currently judged with minimum
The first evaluation of estimate, the second evaluation of estimate and the third evaluation of estimate of transition matrix；
Step S24, judge whether first evaluation of estimate meets the first default evaluation function, whether is second evaluation of estimate
Meet the second default evaluation function and whether the third evaluation of estimate meets third and preset evaluation function, if being judged as YES,
Step S27 is executed, otherwise, executes step S25；
Step S25, according to the current migration distance cost of each calculating task transition matrix, current migration flow cost
Moving after the migration distance cost, iteration after generating the iteration of each calculating task transition matrix with current migration load balancing degrees
Migration load balancing degrees after moving flow cost and iteration；
Step S26, using the migration distance cost after the iteration of each calculating task transition matrix as current migration distance
Cost is made using the migration flow cost after iteration as current migration flow cost and with the migration load balancing degrees after iteration
For current migration load balancing degrees, step S21 is executed；
Step S27, determining currently there is minimum to judge the calculating task transition matrix of parameter for Optimal calculation task immigration
Matrix simultaneously executes step S3.
Optionally, before step S21 further include: setting iteration initial parameter k=0；
Step S21 includes:
Step S21a, current iteration number k is carried out plus 1 is handled；
Step S21b, by current migration distance cost, global history average mobility distance costs, current migration flow
Cost, global history average mobility flow cost, current migration load balancing degrees and global history average mobility load balancing
Degree, which substitutes into, judges parameter formula:
Calculate the current judgement parameter of each calculating task transition matrix；
Wherein,Indicate current migration distance cost,Indicate global history average mobility distance costs,Table
Show current migration flow cost,Indicate global history average mobility flow cost,Indicate that current migration load is equal
Weighing apparatus degree,Indicate global history average mobility load balancing degrees, ZkIndicate that current judgement parameter, k indicate current iteration time
Number, k ∈ [1, d], d are preset threshold, and m, n, p indicate that regulating constant, i, j and t indicate that the three-dimensional of calculating task transition matrix is sat
Mark, i indicate that the abscissa of calculating task transition matrix, i=0,1,2 ..., m, j indicate the ordinate of calculating task transition matrix,
J=0,1,2, the vertical coordinate of n, t expression calculating task transition matrix, t=0,1,2 ..., p, α, β, γ ∈ (0,1), alpha+beta+
γ=1.
Optionally, step S23 includes:
Step S23a, by the current migration distance cost of the current calculating task transition matrix that there is minimum to judge parameter
Substitute into the first evaluation of estimate function formula:
Acquire the first evaluation of estimate of the current calculating task transition matrix that there is minimum to judge parameter, wherein ε ∈ (0,1),Expression is worked as
The first preceding evaluation of estimate；
Step S23b, by the current migration flow cost of the current calculating task transition matrix that there is minimum to judge parameter
Substitute into the second evaluation of estimate function formula:
The second evaluation of estimate of the current calculating task transition matrix that there is minimum to judge parameter is acquired,Indicate the second current evaluation
Value；
Step S23c, by the current migration load balancing of the current calculating task transition matrix that there is minimum to judge parameter
Degree substitutes into third evaluation of estimate function formula:
Acquire the third evaluation of estimate of the current calculating task transition matrix that there is minimum to judge parameter, whereinIndicate current third
Evaluation of estimate.
Optionally, the described first default evaluation function isThe second default evaluation function isInstitute
Stating the default evaluation function of third is
Optionally, step S25 includes:
Step S25a, according to the current migration distance cost of each calculating task transition matrixCurrent migration flow
CostWith current migration load balancing degreesGenerate the current composite vector of each calculating task transition matrix
Step S25b, by the current migration distance cost of each calculating task transition matrixIt is migrated when kth time iteration
The average value of distance costsCurrent migration flow costThe average value of flow cost is migrated when kth time iterationWhen
Preceding migration load balancing degreesAnd the average value of load balancing degrees is migrated when kth time iterationSubstitute into formula:It calculates each when+1 iteration of kth
The load balancing and shortest distance enhancement factor of calculating task transition matrix
Step S25c, by the current migration distance cost of each calculating task transition matrixGlobal history migration distance
The minimum value L of costminG, current migration flow costThe minimum value C of global history migration flow costminG, it is current
Migrate load balancing degreesAnd global history average mobility load balancing degreesSubstitute into formula:Calculate each calculating task transition matrix when+1 iteration of kth
Load balancing and shortest distance direction factor
Step S25d, by the current composite vector of each calculating task transition matrixIt is respectively calculated when+1 iteration of kth
The load balancing and shortest distance enhancement factor of task immigration matrixEach calculating task transition moment when with+1 iteration of kth
The load balancing and shortest distance direction factor of battle arraySubstitute into formula:Calculate kth
The composite vector of each calculating task transition matrix when+1 iteration
Step S25e, the composite vector of each calculating task transition matrix when according to+1 iteration of kthDetermine kth+
The migration distance cost of each calculating task transition matrix, migration flow cost and migration load balancing degrees when 1 iteration.
Optionally, the migration distance cost of each calculating task transition matrix is made when step S26 includes: with+1 iteration of kth
For the current migration distance cost of each calculating task transition matrix, each calculating task transition matrix when with+1 iteration of kth
Current migration flow cost of the flow cost as each calculating task transition matrix is migrated, and in terms of when+1 iteration of kth respectively
Current migration load balancing degrees of the migration load balancing degrees of task immigration matrix as each calculating task transition matrix are calculated, are held
Row step S201.
Optionally, after step S25 further include:
Step S25e, judge whether the value of current iteration number k is greater than preset threshold d, if so, step S25f is executed, if
It is no, execute step S26；
Step S25f, the calculating task transition matrix that output currently there is minimum to judge parameter.
To achieve the above object, the present invention provides a kind of calculating task transference apparatus, which includes:
Module is obtained, for obtaining each calculating task transition matrix, the calculating task transition matrix includes current moves
Move distance costs, current migration flow cost and current migration load balancing degrees；
Determining module, for the current migration distance cost according to each calculating task transition matrix, current migration stream
Amount cost and current migration load balancing degrees determine Optimal calculation task immigration matrix；
Output module, for exporting the Optimal calculation task immigration matrix.
Beneficial effects of the present invention:
In the technical solution of calculating task moving method and calculating task transference apparatus provided by the invention, appointed according to each calculate
Current migration distance cost, current migration flow cost and the current migration load balancing degrees of business transition matrix are determined
Optimal calculation task immigration matrix, so that it is determined that going out optimal calculating task migration corresponding with Optimal calculation task immigration matrix
Strategy realizes low time delay, low discharge cost and the load balancing of calculating task migration.
Detailed description of the invention
Fig. 1 is a kind of flow chart for calculating task moving method that the embodiment of the present invention one provides；
Fig. 2 is a kind of flow chart of calculating task moving method provided by Embodiment 2 of the present invention；
Fig. 3 is the structural schematic diagram of calculating task transition matrix；
Fig. 4 is the application schematic diagram of calculating task moving method provided by Embodiment 2 of the present invention；
Fig. 5 is the application schematic diagram for the calculating task transference apparatus that the embodiment of the present invention three provides.
Specific embodiment
To make those skilled in the art more fully understand technical solution of the present invention, the present invention is mentioned with reference to the accompanying drawing
The calculating task moving method and calculating task transference apparatus of confession are described in detail.
Fig. 1 is a kind of flow chart for calculating task moving method that the embodiment of the present invention one provides, as shown in Figure 1, the meter
Calculating task immigration method includes:
Step S1, obtain each calculating task transition matrix, calculating task transition matrix include current migration distance cost,
Current migration flow cost and current migration load balancing degrees.
Step S2, according to the current migration distance cost of each calculating task transition matrix, current migration flow cost
Optimal calculation task immigration matrix is determined with current migration load balancing degrees.
Step S3, Optimal calculation task immigration matrix is exported.
In the technical solution of calculating task moving method provided by the present embodiment, according to each calculating task transition matrix
Current migration distance cost, current migration flow cost and current migration load balancing degrees determine Optimal calculation task
Transition matrix realizes so that it is determined that going out optimal calculating task migration strategy corresponding with Optimal calculation task immigration matrix
Low time delay, low discharge cost and the load balancing of calculating task migration.
Fig. 2 is a kind of flow chart of calculating task moving method provided by Embodiment 2 of the present invention, as shown in Fig. 2, the meter
Calculating task immigration method includes:
Step S201, calculating task migration request is obtained.
Specifically, in the present embodiment, calculating task migration request can be by the access terminal of sensing layer, for example, disaster
Rescue the communication in the disaster monitoring sensor, the portable edge termination of disaster relief, disaster relief communication support vehicle of sensing layer
The disaster reliefs awareness apparatus such as terminal, the communication terminal of disaster relief floor manager vehicle or rescue locator, when preset
Between active reporting obtain, or be periodically asked by access terminal mechanism inquiry and obtain, after obtaining calculating task migration request,
Calculating task migration request is summarized.
Step S202, according to calculating task migration request carry out calculating task migration, generate initial migration distance cost,
Initial migration flow cost and initial migration load balancing degrees.
Specifically, calculating task migration is carried out with different calculating task migration strategies, generates each calculating task migration plan
Migration results slightly, migration results include that initial migration distance cost, initial migration flow cost and initial migration are negative
Carry equilibrium degree.
Step S203, it is loaded according to initial migration distance cost, initial migration flow cost and initial migration equal
Weighing apparatus degree generates calculating task transition matrix.
Specifically, according to the corresponding initial migration distance cost of each calculating task migration strategy, initial migration flow
Cost and initial migration load balancing degrees generate the corresponding calculating task transition matrix of each calculating task migration strategy.
Step S204, obtain each calculating task transition matrix, calculating task transition matrix include current migration distance at
Originally, current migration flow cost and current migration load balancing degrees.
Preferably, current migration distance cost, current migration flow cost and migration load balancing degrees currently
Quantity is multiple.According to current migration distance costCurrent migration flow costIt is loaded with current migration
Equilibrium degreeProduce the current composite vector of each calculating task transition matrixWherein,
I, the three-dimensional coordinate of j and t expression calculating task transition matrix, the abscissa of i expression calculating task transition matrix, i=0,1,
The ordinate of 2 ..., m, j expression calculating task transition matrix, j=0,1,2, n, t indicate the vertical of calculating task transition matrix
Coordinate, t=0,1,2 ..., p, m, n, p indicate regulating constant.Fig. 3 is the structural schematic diagram of calculating task transition matrix, calculates and appoints
Business transition matrix can be understood as calculating task migration storage model, as shown in figure 3, including at least in calculating task transition matrix
One
Step S205, iterative model is established, the iterative model includes:
Judge parameter formula:
First evaluation of estimate function formula:
Second evaluation of estimate function formula:
Third evaluation of estimate function formula:
First default evaluation function:Second default evaluation function:Third presets evaluation function:
Wherein,K indicates the number of iterations, k
∈ [1, d], d are preset threshold, ZkIndicate the judgement parameter of calculated calculating task transition matrix when kth time iteration,Table
Show the migration distance cost of calculated calculating task transition matrix when kth time iteration,Indicate global history average mobility away from
From cost,Indicate the migration flow cost of calculated calculating task transition matrix when kth time iteration,Indicate that the overall situation is gone through
History average mobility flow cost,Indicate the migration load balancing of calculated calculating task transition matrix when kth time iteration
Degree,Indicate global history average mobility load balancing degrees,Indicate calculated calculating task transition moment when kth time iteration
First evaluation of estimate of battle array,Indicate the second evaluation of estimate of calculated calculating task transition matrix when kth time iteration,It indicates
The third evaluation of estimate of calculated calculating task transition matrix when kth time iteration,Indicate calculated meter when kth time iteration
The composite vector of task immigration matrix is calculated,Calculated calculating task transition matrix is compound when expression+1 iteration of kth
Vector, Indicate+1 iteration of kth
When calculated calculating task transition matrix load balancing and shortest distance enhancement factor, Indicate calculated calculating task transition moment when+1 iteration of kth
The load balancing and shortest distance direction factor of battle array,Indicate the migration distance of calculating task transition matrix when kth time iteration at
This average value,Indicate the average value of the migration flow cost of calculating task transition matrix when kth time iteration,Indicate the
The average value of the migration load balancing degrees of calculating task transition matrix, L when k iterationminGIndicate global history migration distance at
This minimum value, CminGIndicate the minimum value of global history migration flow cost, m, n, p indicate that regulating constant, i, j and t indicate
The three-dimensional coordinate of calculating task transition matrix, i indicate that the abscissa of calculating task transition matrix, i=0,1,2 ..., m, j indicate
The ordinate of calculating task transition matrix, j=0,1,2, the vertical coordinate of n, t expression calculating task transition matrix, t=0,1,
2 ..., p, α, β, γ ∈ (0,1), alpha+beta+γ=1, ε ∈ (0,1).
Step S206, iteration initial parameter k=0 and maximum number of iterations d is set.
Step S207, current iteration number k is carried out plus 1 is handled.
Step S208, according to the current migration distance cost of each calculating task transition matrix, current migration flow at
This generates the current judgement parameter of each calculating task transition matrix with current migration load balancing degrees.
When k=1, the initial migration distance cost of calculating task transition matrix is the current of calculating task transition matrix
Migration distance cost, initial migration flow cost is current migration flow cost, initial migration load balancing degrees
As current migration load balancing degrees.It should be noted that each calculating task transition matrix is worked as after each iteration later
Preceding migration distance cost, current migration flow cost and current migration load balancing degrees can change, for example, k
When=2, then with the 2nd iteration when calculated migration distance cost, migration flow cost and migration load balancing degrees make respectively
It carries out inputting and carrying out step for current migration distance cost, current migration flow cost and current migration load balancing degrees
Rapid S208.And so on, it will not enumerate herein.
Specifically, step S208 includes: by current migration distance cost, global history average mobility distance costs, current
Migration flow cost, global history average mobility flow cost, current migration load balancing degrees and global history average mobility
Load balancing degrees substitution judges parameter formula:
Calculate the current judgement parameter of each calculating task transition matrix.
Wherein,Indicate current migration distance cost,Indicate global history average mobility distance costs,It indicates
Current migration flow cost,Indicate global history average mobility flow cost,Indicate current migration load balancing
Degree,Indicate global history average mobility load balancing degrees, ZkIndicate that current judgement parameter, k indicate current iteration number, k
∈ [1, d], d are preset threshold, and m, n, p indicate that regulating constant, i, j and t indicate the three-dimensional coordinate of calculating task transition matrix, i
Indicate that the abscissa of calculating task transition matrix, i=0,1,2 ..., m, j indicate the ordinate of calculating task transition matrix, j=
0,1,2, the vertical coordinate of n, t expression calculating task transition matrix, t=0,1,2 ..., p, α, β, γ ∈ (0,1), alpha+beta+γ=
1。
Step S209, the current judgement parameter for comparing each calculating task transition matrix, determines currently to sentence with minimum
The calculating task transition matrix of disconnected parameter.
Specifically, according to parameter formula is judged, after the current judgement parameter for calculating each calculating task transition matrix, lead to
Cross the current judgement parameter i.e. Z for comparing each calculating task transition matrixkValue size, determine currently to have and the smallest sentence
Disconnected parameter ZkValue calculating task transition matrix.
Step S210, according to currently have minimum judge parameter calculating task transition matrix current migration distance at
Originally, current migration flow cost and current migration load balancing degrees generate the calculating task that parameter is currently judged with minimum
The first evaluation of estimate, the second evaluation of estimate and the third evaluation of estimate of transition matrix.
Specifically, step S210 includes:
Step S210a, by the current migration distance of the current calculating task transition matrix that there is minimum to judge parameter at
This substitution the first evaluation of estimate function formula:
Acquire the first evaluation of estimate of the current calculating task transition matrix that there is minimum to judge parameter, wherein ε ∈ (0,1),Expression is worked as
The first preceding evaluation of estimate.
Step S210b, by the current migration flow of the current calculating task transition matrix that there is minimum to judge parameter at
This substitution the second evaluation of estimate function formula:
The second evaluation of estimate of the current calculating task transition matrix that there is minimum to judge parameter is acquired,Indicate the second current evaluation
Value.
Step S210c, the current migration of the current calculating task transition matrix that there is minimum to judge parameter is loaded equal
Weighing apparatus degree substitutes into third evaluation of estimate function formula:
Acquire the third evaluation of estimate of the current calculating task transition matrix that there is minimum to judge parameter, whereinIndicate current third
Evaluation of estimate.
Step S211, judge whether first evaluation of estimate meets the first default evaluation function, whether the second evaluation of estimate is full
Whether the second default evaluation function of foot and third evaluation of estimate, which meet third, is preset evaluation function, if being judged as YES, executes step
Otherwise S216 executes step S212.
Specifically, the first evaluation of estimate step S210 obtainedSubstitute into the first default evaluation function:Judgement
Whether first evaluation of estimate meets the described first default evaluation function, the second evaluation of estimate that step S210 is obtainedIt substitutes into
Second default evaluation function:Judge whether second evaluation of estimate meets the described second default evaluation function, will walk
The third evaluation of estimate that rapid S210 is obtainedIt substitutes into third and presets evaluation function:Whether judge the third evaluation of estimate
Meet the third and preset evaluation function, if above-mentioned be judged as when being all satisfied executes step S216, if being judged as in the presence of at least one
It is a that step S212 is executed when being unsatisfactory for.
Step S212, according to the current migration distance cost of each calculating task transition matrix, current migration flow at
Originally after the migration distance cost, iteration after generating the iteration of each calculating task transition matrix with current migration load balancing degrees
Migration load balancing degrees after migrating flow cost and iteration.
Specifically, step S212 includes:
Step S212a, according to the current migration distance cost of each calculating task transition matrixCurrent migration stream
Measure costWith current migration load balancing degreesGenerate the current composite vector of each calculating task transition matrix
Step S212b, by the current migration distance cost of each calculating task transition matrixIt is migrated when kth time iteration
The average value of distance costsCurrent migration flow costThe average value of flow cost is migrated when kth time iterationWhen
Preceding migration load balancing degreesAnd the average value of load balancing degrees is migrated when kth time iterationSubstitute into formula:When calculating+1 iteration of kth
The load balancing and shortest distance enhancement factor of each calculating task transition matrix
Step S212c, by the current migration distance cost of each calculating task transition matrixGlobal history migration away from
Minimum value L from costminG, current migration flow costThe minimum value C of global history migration flow costminG, when
Preceding migration load balancing degreesAnd global history average mobility load balancing degreesSubstitute into formula:Calculate the negative of each calculating task transition matrix when+1 iteration of kth
Carry balanced and shortest distance direction factor
Step S212d, by the current composite vector of each calculating task transition matrixIt is respectively calculated when+1 iteration of kth
The load balancing and shortest distance enhancement factor of task immigration matrixEach calculating task transition moment when with+1 iteration of kth
The load balancing and shortest distance direction factor of battle arraySubstitute into formula:Calculate
The composite vector of each calculating task transition matrix when k+1 iteration
Step S212e, the composite vector of each calculating task transition matrix when according to+1 iteration of kthDetermine kth
The migration distance cost of each calculating task transition matrix, migration flow cost and migration load balancing degrees when+1 iteration.
Due toTherefore, each when+1 iteration of kth
The migration distance cost of calculating task transition matrixBe it is confirmable, when+1 iteration of kth each calculating task transition matrix
Migration flow costBe it is confirmable, when+1 iteration of kth each calculating task transition matrix migration load balancing degreesIt is also confirmable.Wherein, the migration distance cost of each calculating task transition matrix, migration flow when+1 iteration of kth
Cost and migration load balancing degrees be each calculating task transition matrix iteration after migration distance cost, migration flow cost
With migration load balancing degrees.
Step S213, judge whether the value of current iteration number k is greater than preset threshold d, if so, step S214 is executed, if
It is no, execute step S215.
In the present embodiment, d 50.
Step S214, output currently there is minimum to judge the calculating task transition matrix of parameter, and terminate process.
Step S215, using the migration distance cost after the iteration of each calculating task transition matrix as current migration distance
Cost is made using the migration flow cost after iteration as current migration flow cost and with the migration load balancing degrees after iteration
For current migration load balancing degrees, step 207 is executed.
Specifically, the migration distance cost of each calculating task transition matrix is made when step S215 includes: with+1 iteration of kth
For the current migration distance cost of each calculating task transition matrix, each calculating task transition matrix when with+1 iteration of kth
Current migration flow cost of the flow cost as each calculating task transition matrix is migrated, and in terms of when+1 iteration of kth respectively
Current migration load balancing degrees of the migration load balancing degrees of task immigration matrix as each calculating task transition matrix are calculated, are held
Row step S207.
Step S216, determining currently there is minimum to judge the calculating task transition matrix of parameter for Optimal calculation task immigration
Matrix.
Step S217, Optimal calculation task immigration matrix is exported.
Determine Optimal calculation task immigration matrix, so that it is determined that Optimal calculation task immigration strategy, to realize that calculating is appointed
Low time delay, low discharge cost and the load balancing of business migration.
It should be noted that " current " word that full text occurs is it can be appreciated that " kth time iteration in the present embodiment
When ", that is to say, that appeared in full text such as " current migration distance cost ", " current migration flow cost ", " when
Preceding migration load balancing ", " current judgement parameter ", " current composite vector ", " current the first evaluation of estimate ", " current
The second evaluation of estimate ", in " current third evaluation of estimate " and " the current calculating task transition matrix that there is minimum to judge parameter "
" current " be to indicate when iteration " kth time ", therefore, calculating when in this tittle of calculating " current " being iteration secondary with kth
Corresponding substitute into of amount out is acquired.
Calculating task moving method provided by the present embodiment is the calculating task moving method based on edge calculations, Fig. 4
For the application schematic diagram of calculating task migrationization method provided by Embodiment 2 of the present invention, as shown in figure 4, based on edge calculations
Calculating task moving method can be applied to disaster relief scene, the disaster relief calculating task moving method based on edge calculations
Application system mainly divides five levels: disaster relief sensing layer, unmanned plane base station edge network transport layer, disaster relief edge net
Close access layer, disaster relief edge data central core and disaster relief command centre analysis layer.
Wherein, disaster relief sensing layer includes that disaster monitoring sensor, the portable edge termination of disaster relief, disaster relief are logical
Believe the disaster reliefs awareness apparatus such as service truck, disaster relief floor manager vehicle, rescue locator, realizes data collection and control；
Unmanned plane base station edge network transport layer includes unmanned plane base station and telecommunication satellite, realizes connecing for unmanned plane base station and satellite network
Enter and data are transmitted；Disaster relief Border Gateway access layer is made of several disaster relief Border Gateway, is realized and is come from operator
The access of edge network, satellite network；Disaster relief edge data central core is made of several rescue Edge Servers, realization pair
Disaster relief calculating task migration request from disaster relief awareness apparatus carries out part processing and forwarding remainder calculates
Task immigration request；Disaster relief command centre analysis layer is made of several rescue analysis processors and disaster relief database,
Realize the processing and analysis to the calculating task migration request from disaster relief awareness apparatus, main processing and analysis are except migration
To the calculating task migration request of the remainder disaster relief awareness apparatus outside rescue Edge Server.
Wherein, disaster monitoring sensor, the portable edge termination of disaster relief, disaster relief communication support vehicle, disaster relief
The disaster reliefs awareness apparatus such as floor manager vehicle, rescue locator, for accessing the nothing of unmanned plane base station edge network transport layer
Man-machine base station, telecommunication satellite, and transmit disaster relief calculating task migration request；The nothing of unmanned plane base station edge network transport layer
Man-machine base station, telecommunication satellite are used for the disaster relief directly or indirectly through internet access disaster relief Border Gateway access layer
Border Gateway, and transmit disaster relief calculating task migration request；Disaster relief Border Gateway is for being linked into disaster relief side
The rescue Edge Server of edge data center's layer, and transmit disaster relief calculating task migration request；Disaster relief edge data
The rescue Edge Server of central core is used to obtain corresponding portion disaster according to part disaster relief calculating task migration request and rescue
Corresponding portion disaster relief service is passed through provider edge network, satellite network, unmanned plane base station, disaster relief by the service of helping
Awareness apparatus is supplied to user, and remainder disaster relief calculating task migration request is sent to disaster relief command centre
Analysis layer；The rescue analysis processor of disaster relief command centre analysis layer is analyzed in former disaster relief calculating task migration request
Remaining disaster relief Analysis Service request, and the calamity needed for extracting remaining disaster relief Analysis Service in disaster relief database
Hardly possible rescue perception data, by analysis result, that is, Optimal calculation task of required remainder disaster relief calculating task migration request
The corresponding calculating task migration strategy of transition matrix and disaster relief perception data return to rescue Edge Server, by rescuing side
Edge server is by disaster relief Border Gateway, provider edge network, unmanned plane base station, satellite network by remainder disaster
Rescue calculating task migration request the corresponding calculating task migration strategy of analysis result, that is, Optimal calculation task immigration matrix and
Disaster relief perception data return to disaster monitoring sensor, the portable edge termination of disaster relief, disaster relief communication support vehicle,
The disaster reliefs awareness apparatus such as disaster relief floor manager vehicle, rescue locator.
In the present embodiment, each calculating task migration request has different priority levels.It is asked when receiving calculating task migration
After asking, calculating task migration is carried out by different calculating task migration strategies according to calculating task migration request, including
The migration results of initial migration distance cost, initial migration flow cost and initial migration load balancing degrees, then pass through
Iterative model in the present embodiment carries out carrying out the migration results data analysis, judgement, so that it is determined that optimal calculating is appointed out
The corresponding Optimal calculation task immigration strategy of business transition matrix, finally exports the optimal calculating task transition matrix and its correspondence
Calculating task migration strategy.Iterative model in the present embodiment hides recurrent neural network load in conjunction with hyperspace multilayer
Balanced study analysis thought hides neural network, load balancing, probability theory, biology, fortune based on hyperspace, multilayer
, intelligent optimization, machine learning scheduling theory are raised, low time delay, the low migration flow cost, low migration of calculating task migration are realized
The advantage of distance costs and load balancing.
In the technical solution of calculating task moving method provided by the present embodiment, according to each calculating task transition matrix
Current migration distance cost, current migration flow cost and current migration load balancing degrees determine Optimal calculation task
Transition matrix realizes so that it is determined that going out optimal calculating task migration strategy corresponding with Optimal calculation task immigration matrix
Low time delay, low discharge cost and the load balancing of calculating task migration.
The embodiment of the present invention three provides a kind of calculating task transference apparatus, which includes: acquisition module,
For obtaining each calculating task transition matrix, the calculating task transition matrix includes current migration distance cost, current
Migrate flow cost and current migration load balancing degrees；Determining module, for according to the current of each calculating task transition matrix
Migration distance cost, current migration flow cost and current migration load balancing degrees determine Optimal calculation task immigration
Matrix；Output module, for exporting the Optimal calculation task immigration matrix.
Calculating task transference apparatus provided by the present embodiment, the calculating task migration provided for realizing above-described embodiment two
Method specifically describes and can be found in above-described embodiment two, not reinflated herein to repeat.
Fig. 5 is the application schematic diagram for the calculating task transference apparatus that the embodiment of the present invention three provides, as shown in figure 5, when calculating
After task immigration device receives m calculating task migration request, different computing tasks are used to each calculating task migration request
Migration strategy carries out calculating task and migrates optimization processing, generates migration distance cost, migration flow cost and migration load balancing
Degree generates calculating task transition matrix according to migration distance cost, migration flow cost and migration load balancing degrees, further according to moving
It moves distance costs, migration flow cost and migration load balancing degrees and determines Optimal calculation task immigration matrix, so that it is determined that going out
Optimal calculation task immigration strategy corresponding with Optimal calculation task immigration matrix, finally obtains each calculating task migration request most
Analysis afterwards is as a result, analysis result is similarly m.Wherein, each calculating task migration request is not interfere with each other independently of each other, and m is to set
Fixed number amount.
In the technical solution of calculating task transference apparatus provided by the present embodiment, according to working as each calculating task transition matrix
Preceding migration distance cost, current migration flow cost and current migration load balancing degrees determine that Optimal calculation task is moved
Matrix is moved, so that it is determined that going out optimal calculating task migration strategy corresponding with Optimal calculation task immigration matrix, realizes meter
Calculate low time delay, low discharge cost and the load balancing of task immigration.
It is understood that the principle that embodiment of above is intended to be merely illustrative of the present and the exemplary implementation that uses
Mode, however the present invention is not limited thereto.For those skilled in the art, essence of the invention is not being departed from
In the case where mind and essence, various changes and modifications can be made therein, these variations and modifications are also considered as protection scope of the present invention.
Claims (9)
1. a kind of calculating task moving method characterized by comprising
Step S1, obtain each calculating task transition matrix, the calculating task transition matrix include current migration distance cost,
Current migration flow cost and current migration load balancing degrees；
Step S2, according to the current migration distance cost of each calculating task transition matrix, current migration flow cost and work as
Preceding migration load balancing degrees determine Optimal calculation task immigration matrix；
Step S3, the Optimal calculation task immigration matrix is exported.
2. calculating task moving method according to claim 1, which is characterized in that step S2 includes:
Step S21, according to the current migration distance cost of each calculating task transition matrix, current migration flow cost and work as
Preceding migration load balancing degrees generate the current judgement parameter of each calculating task transition matrix；
Step S22, the current judgement parameter for comparing each calculating task transition matrix determines currently there is minimum to judge parameter
Calculating task transition matrix；
Step S23, according to currently have minimum judge parameter calculating task transition matrix current migration distance cost, when
Preceding migration flow cost and current migration load balancing degrees generate the calculating task migration that currently there is minimum to judge parameter
The first evaluation of estimate, the second evaluation of estimate and the third evaluation of estimate of matrix；
Step S24, judge whether first evaluation of estimate meets the first default evaluation function, whether second evaluation of estimate meets
Whether the second default evaluation function and the third evaluation of estimate meet the default evaluation function of third executes if being judged as YES
Otherwise step S27 executes step S25；
Step S25, according to the current migration distance cost of each calculating task transition matrix, current migration flow cost and work as
Preceding migration load balancing degrees generate the migration distance cost after the iteration of each calculating task transition matrix, the migration stream after iteration
Migration load balancing degrees after measuring cost and iteration；
Step S26, using the migration distance cost after the iteration of each calculating task transition matrix as current migration distance cost,
Using the migration flow cost after iteration as current migration flow cost and using the migration load balancing degrees after iteration as working as
Preceding migration load balancing degrees execute step S21；
Step S27, determining currently there is minimum to judge the calculating task transition matrix of parameter for Optimal calculation task immigration matrix
And execute step S3.
3. calculating task moving method according to claim 2, which is characterized in that
Before step S21 further include: setting iteration initial parameter k=0；
Step S21 includes:
Step S21a, current iteration number k is carried out plus 1 is handled；
Step S21b, by current migration distance cost, global history average mobility distance costs, current migration flow at
Sheet, global history average mobility flow cost, current migration load balancing degrees and global history average mobility load balancing degrees
Substitution judges parameter formula:
Calculate the current judgement parameter of each calculating task transition matrix；
Wherein,Indicate current migration distance cost,Indicate global history average mobility distance costs,Indicate current
Migration flow cost,Indicate global history average mobility flow cost,Indicate current migration load balancing degrees,Indicate global history average mobility load balancing degrees, ZkIndicate that current judgement parameter, k indicate current iteration number, k ∈
[1, d], d are preset threshold, and m, n, p indicate that regulating constant, i, j and t indicate the three-dimensional coordinate of calculating task transition matrix, i table
Showing the abscissa of calculating task transition matrix, i=0,1,2 ..., m, j indicate the ordinate of calculating task transition matrix, j=0,
1,2, the vertical coordinate of n, t expression calculating task transition matrix, t=0,1,2 ..., p, α, β, γ ∈ (0,1), alpha+beta+γ=
1。
4. calculating task moving method according to claim 3, which is characterized in that step S23 includes:
Step S23a, the current migration distance cost of the current calculating task transition matrix that there is minimum to judge parameter is substituted into
First evaluation of estimate function formula:
Acquire the first evaluation of estimate of the current calculating task transition matrix that there is minimum to judge parameter, wherein ε ∈ (0,1),It indicates
The first current evaluation of estimate；
Step S23b, the current migration flow cost of the current calculating task transition matrix that there is minimum to judge parameter is substituted into
Second evaluation of estimate function formula:
The second evaluation of estimate of the current calculating task transition matrix that there is minimum to judge parameter is acquired,Indicate the second current evaluation
Value；
Step S23c, by the current migration load balancing degrees generation of the current calculating task transition matrix that there is minimum to judge parameter
Enter third evaluation of estimate function formula:
Acquire the third evaluation of estimate of the current calculating task transition matrix that there is minimum to judge parameter, whereinIndicate current third
Evaluation of estimate.
5. calculating task moving method according to claim 4, which is characterized in that the first default evaluation function isThe second default evaluation function isThe third presets evaluation function
6. calculating task moving method according to claim 5, which is characterized in that step S25 includes:
Step S25a, according to the current migration distance cost of each calculating task transition matrixCurrent migration flow costWith current migration load balancing degreesGenerate the current composite vector of each calculating task transition matrix
Step S25b, by the current migration distance cost of each calculating task transition matrixWhen kth time iteration migration distance at
This average valueCurrent migration flow costThe average value of flow cost is migrated when kth time iterationCurrent moves
Move load balancing degreesAnd the average value of load balancing degrees is migrated when kth time iterationSubstitute into formula:It is respectively counted when calculating+1 iteration of kth
Calculate the load balancing and shortest distance enhancement factor of task immigration matrix
Step S25c, by the current migration distance cost of each calculating task transition matrixGlobal history migration distance cost
Minimum value LminG, current migration flow costThe minimum value C of global history migration flow costminG, current migration
Load balancing degreesAnd global history average mobility load balancing degreesSubstitute into formula:Calculate each calculating task transition matrix when+1 iteration of kth
Load balancing and shortest distance direction factor
Step S25d, by the current composite vector of each calculating task transition matrixEach calculating task when+1 iteration of kth
The load balancing and shortest distance enhancement factor of transition matrixWith calculating task transition matrix each when+1 iteration of kth
Load balancing and shortest distance direction factorSubstitute into formula:It calculates kth+1 time
The composite vector of each calculating task transition matrix when iteration
Step S25e, the composite vector of each calculating task transition matrix when according to+1 iteration of kthIt determines kth+1 time
The migration distance cost of each calculating task transition matrix, migration flow cost and migration load balancing degrees when iteration.
7. calculating task moving method according to claim 6, which is characterized in that step S26 includes: to be changed with kth+1 time
For when each calculating task transition matrix migration distance cost as each calculating task transition matrix current migration distance at
This, migration flow cost the working as each calculating task transition matrix of each calculating task transition matrix when using+1 iteration of kth
Preceding migration flow cost, and the migration load balancing degrees of using+1 iteration of kth when each calculating task transition matrix are as in terms of each
The current migration load balancing degrees of task immigration matrix are calculated, step S201 is executed.
8. calculating task moving method according to claim 7, which is characterized in that after step S25 further include:
Step S25e, judge whether the value of current iteration number k is greater than preset threshold d, if so, step S25f is executed, if it is not, holding
Row step S26；
Step S25f, the calculating task transition matrix that output currently there is minimum to judge parameter.
9. a kind of calculating task transference apparatus characterized by comprising
Obtain module, for obtaining each calculating task transition matrix, the calculating task transition matrix include current migration away from
From cost, current migration flow cost and current migration load balancing degrees；
Determining module, for according to the current migration distance cost of each calculating task transition matrix, current migration flow at
This and current migration load balancing degrees determine Optimal calculation task immigration matrix；
Output module, for exporting the Optimal calculation task immigration matrix.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810798448.0A CN108984307B (en) | 2018-07-19 | 2018-07-19 | Computing task migration method and computing task migration device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810798448.0A CN108984307B (en) | 2018-07-19 | 2018-07-19 | Computing task migration method and computing task migration device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108984307A true CN108984307A (en) | 2018-12-11 |
| CN108984307B CN108984307B (en) | 2020-11-10 |
Family
ID=64549506
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810798448.0A Active CN108984307B (en) | 2018-07-19 | 2018-07-19 | Computing task migration method and computing task migration device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108984307B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111787060A (en) * | 2020-05-28 | 2020-10-16 | 网宿科技股份有限公司 | Traffic scheduling method, system and device |
| CN111857976A (en) * | 2020-08-07 | 2020-10-30 | 华侨大学 | Multi-objective optimization calculation migration method based on decomposition |
| CN112261120A (en) * | 2020-10-19 | 2021-01-22 | 广东电网有限责任公司电力调度控制中心 | Cloud-side cooperative task unloading method and device for power distribution internet of things |
| CN112887349A (en) * | 2019-11-29 | 2021-06-01 | 北京百度网讯科技有限公司 | Method and device for distributing files |
| CN112929427A (en) * | 2021-01-27 | 2021-06-08 | 北京邮电大学 | Service node determination method and device for low-earth-orbit satellite edge calculation |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8554918B1 (en) * | 2011-06-08 | 2013-10-08 | Emc Corporation | Data migration with load balancing and optimization |
| CN106155785A (en) * | 2016-06-23 | 2016-11-23 | 中国人民解放军国防科学技术大学 | A kind of data migration method across data center's cloud computing system |
| CN107734558A (en) * | 2017-10-26 | 2018-02-23 | 北京邮电大学 | A kind of control of mobile edge calculations and resource regulating method based on multiserver |
| CN107919986A (en) * | 2017-11-09 | 2018-04-17 | 北京邮电大学 | VM migrates optimization method between MEC nodes in super-intensive network |
-
2018
- 2018-07-19 CN CN201810798448.0A patent/CN108984307B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8554918B1 (en) * | 2011-06-08 | 2013-10-08 | Emc Corporation | Data migration with load balancing and optimization |
| CN106155785A (en) * | 2016-06-23 | 2016-11-23 | 中国人民解放军国防科学技术大学 | A kind of data migration method across data center's cloud computing system |
| CN107734558A (en) * | 2017-10-26 | 2018-02-23 | 北京邮电大学 | A kind of control of mobile edge calculations and resource regulating method based on multiserver |
| CN107919986A (en) * | 2017-11-09 | 2018-04-17 | 北京邮电大学 | VM migrates optimization method between MEC nodes in super-intensive network |
Non-Patent Citations (3)
| Title |
|---|
| DAN ZHAO ET.AL.: ""A Service Migration Strategy Based on Multiple Attribute Decision in Mobile Edge Computing"", 《17TH IEEE INTERNATIONAL CONFERENCE ON COMMUNICATION TECHNOLOGY》 * |
| 徐雷等: ""云计算环境下的网络技术研究"", 《通信学报》 * |
| 王德文等: ""基于虚拟机动态迁移的电力仿真云计算平台资源调度策略"", 《电力系统自动化》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112887349A (en) * | 2019-11-29 | 2021-06-01 | 北京百度网讯科技有限公司 | Method and device for distributing files |
| CN112887349B (en) * | 2019-11-29 | 2023-05-12 | 北京百度网讯科技有限公司 | Method and device for distributing files |
| CN111787060A (en) * | 2020-05-28 | 2020-10-16 | 网宿科技股份有限公司 | Traffic scheduling method, system and device |
| CN111857976A (en) * | 2020-08-07 | 2020-10-30 | 华侨大学 | Multi-objective optimization calculation migration method based on decomposition |
| CN111857976B (en) * | 2020-08-07 | 2023-06-02 | 华侨大学 | Multi-objective optimization computing migration method based on decomposition |
| CN112261120A (en) * | 2020-10-19 | 2021-01-22 | 广东电网有限责任公司电力调度控制中心 | Cloud-side cooperative task unloading method and device for power distribution internet of things |
| CN112929427A (en) * | 2021-01-27 | 2021-06-08 | 北京邮电大学 | Service node determination method and device for low-earth-orbit satellite edge calculation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108984307B (en) | 2020-11-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108984307A (en) | Calculating task moving method and calculating task transference apparatus | |
| CN108924198B (en) | Data scheduling method, device and system based on edge calculation | |
| CN105550323B (en) | Load balance prediction method and prediction analyzer for distributed database | |
| US10924535B2 (en) | Resource load balancing control method and cluster scheduler | |
| CN106776005B (en) | Resource management system and method for containerized application | |
| CN108089921B (en) | Server for cloud big data operation architecture and operation resource optimization method thereof | |
| CN109561148A (en) | Distributed task dispatching method in edge calculations network based on directed acyclic graph | |
| CN109190821B (en) | Disaster rescue scheduling method, device and system based on edge calculation | |
| Javadpour | Improving resources management in network virtualization by utilizing a software-based network | |
| WO2014157748A1 (en) | Apparatus and method for predicting rate of flow and concentration of constituents flowing into sewage treatment plant using nearest neighbor technique | |
| CN113342510B (en) | Water and power basin emergency command cloud-side computing resource cooperative processing method | |
| CN106534318A (en) | OpenStack cloud platform resource dynamic scheduling system and method based on flow affinity | |
| CN101873224A (en) | Cloud computing load balancing method and equipment | |
| CN106708989A (en) | Spatial time sequence data stream application-based Skyline query method | |
| CN109709985B (en) | Unmanned aerial vehicle task optimization method, device and system | |
| CN111836284B (en) | Energy consumption optimization calculation and unloading method and system based on mobile edge calculation | |
| CN116134448A (en) | Joint machine learning using locality sensitive hashing | |
| CN109388480A (en) | A kind of method and device handling cloud resource | |
| CN103279505A (en) | Mass data processing method based on semantic meaning | |
| CN108595255A (en) | Workflow task dispatching method based on shortest path first in geographically distributed cloud | |
| CN114936708A (en) | Fault diagnosis optimization method based on edge cloud collaborative task unloading and electronic equipment | |
| CN108768857B (en) | Virtual route forwarding method, device and system | |
| CN103984737A (en) | Optimization method for data layout of multi-data centres based on calculating relevancy | |
| CN112187894B (en) | Container dynamic scheduling method based on load correlation prediction | |
| CN112543481B (en) | Method, device and system for balancing computing force load of edge node |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |