CN106961467A - Based on affairs and QoS the service quality performance Forecasting Methodology combined and device - Google Patents
Based on affairs and QoS the service quality performance Forecasting Methodology combined and device Download PDFInfo
- Publication number
- CN106961467A CN106961467A CN201611194711.2A CN201611194711A CN106961467A CN 106961467 A CN106961467 A CN 106961467A CN 201611194711 A CN201611194711 A CN 201611194711A CN 106961467 A CN106961467 A CN 106961467A
- Authority
- CN
- China
- Prior art keywords
- service
- transactional
- transactions
- selection
- combinations
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/02—Protocols based on web technology, e.g. hypertext transfer protocol [HTTP]
- H04L67/025—Protocols based on web technology, e.g. hypertext transfer protocol [HTTP] for remote control or remote monitoring of applications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L65/00—Network arrangements, protocols or services for supporting real-time applications in data packet communication
- H04L65/80—Responding to QoS
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Multimedia (AREA)
- Debugging And Monitoring (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The present invention relates to technical field of the computer network, affairs and the service quality performance Forecasting Methodology and device of QoS combinations are based particularly on.This method includes:Transactional is assigned to Web service first, then analysis requires the combinatorial problem of service with transactional;Secondly analysis has the performance issue of transaction attributes composite services;Finally, according to the Performance comparision of transactional composite services, the method for service selection combined based on affairs and QoS is proposed;Present invention additionally comprises a kind of service quality performance prediction meanss combined based on affairs and QoS of use above method, the device also includes extraction module, processing module, extraction module, determining module.
Description
Technical field
The present invention relates to technical field of the computer network, the service quality performance for being based particularly on affairs and QoS combinations is pre-
Survey method and device.
Background technology
Web service is the technology of automatic interaction between a business, and enterprise can issue their interior business flow and make
For a service, their service is then provided on Web.With going deep into for application, operation flow may include one or many
Individual business function, because these business may call the Web service from different tissues, so needing to combine according to certain way
To cooperate, and ensure the uniformity of its reliability of operation and result.It is following due to facing in complicated network environment
Problem, Services Composition may become sufficiently complex:Component service may long-play, service environment be loosely match and from
The asynchronism of body, because anabolic process may be cancelled caused by machine failure, service may be removed or cancel, and handle potential
Failure be necessary etc..Therefore, Web service needs affair mechanism to ensure the uniformity of output and holding properly and securely
OK.
In traditional composite services system of selection analysis, the different structure that researcher is typically based on workflow is combined
Service assembly QoS (Quality of Service) is analyzed, and the selection of component is then carried out according to QoS.But, these researchs are all
The requirement of affairs could not be considered.Although the introducing of affairs ensure that the execution properly and securely of composite services, it have impact on
The QoS attributes of composite services.It is therefore desirable to propose a kind of method to select the service based on affairs and QoS combinations.
The content of the invention
For problem of the prior art, the present invention proposes the service quality performance Forecasting Methodology combined based on affairs and QoS
And device.
The service quality performance Forecasting Methodology combined based on affairs and QoS, this method is included:
Transactional is assigned to Web service first, then analysis requires the combinatorial problem of service with transactional;
Secondly analysis has the performance issue of transaction attributes composite services;
Finally, according to the Performance comparision of transactional composite services, the services selection side combined based on affairs and QoS is proposed
Method;This method specifically includes following steps:
The first step, when carrying out the selection of current component service, according to the affairs that above chosen suitable assembly is serviced
Feature, and the rule serviced according to combinations of transactions requires that acquisition meets the candidate component service of transactional requirement;
Second step, to each candidate component service carry out Performance Evaluation, assess consider Web service QoS and transactional will
Ask;
3rd step, repetition above procedure, desired composite services are met until obtaining one.
The first step, second step particular content it is as follows:
In common method for service selection, the QoS that Performance Evaluation is all based on all component service is analyzed, this method
Based on transactional and QoS, both are combined the composite services system of selection got based on affairs;
In the composite services selection course based on affairs, when the similar candidate component service of some features can be obtained
When obtaining, their non-functional requirements QoS can reflect the demand of user, and this method proposes that three QoS refer to below Web service ws
Mark:
(1) strike price (Execution price, EP):Requestor calls the cost required for service ws, is expressed as qep
(ws);
(2) time (Execution time, ET) is performed:Service ws and perform once the required time, be expressed as qet
(ws);This method thinks that the time of service execution once is identical, and within the identical time, service execution once may
Failure, it is also possible to successfully complete;
(3) probability (Probability of Success, PS) is successfully completed:Service ws successfully completes the general of user's request
Rate, is expressed as qps(ws);
Wherein, the execution time needs the change that the transactional for considering composite services is brought, and analysis process is as follows:
(1) service has compensability, then it can provide compensation policy to cancel the influence of the service, use symbol
" c " is represented;
(2) services have can continuous retrying property, then it can be retried, and be reached by sufficient number of retry
Final success, is represented with symbol " r ";
(3) services have center dot characteristics, then it is once run succeeded, and its influence is permanently present, and not
It can be revoked, if it performs failure, without any influence, be represented with symbol " p ";
Dynamic, uncertain and opening due to network environment, the Component service from different tissues are often adjusted
With failure, it is assumed that the probability of a Web service successful execution is ps, the time of service execution once is t, then in time t
One Web service either successful execution or performs failure, and service all fails in first n-1 times execution, what n-th ran succeeded
Probability meets geometry distribution, and probability function is:
P (X=n)=(1-ps)n-1ps
Χ represents which time of service is performed, then mathematical expectation is:
The mathematic expectaion reaction of geometry distribution is exactly that the average time for successfully averagely performing number of times, i.e. successful execution is
1/ps, the time of service execution once is t, then, the average time st of successful execution is:
St=t/ps.
Composite services are made up of basic integrated mode, for the time performance of analysis combinations of transactions service, it is proposed that basic group
The time performance analysis method of syntype;Four kinds of basic integrated modes are respectively:Sequence pattern, parallel schema, selection mode
And circulation pattern;And Services Composition is described by Work flow model, TP represents transactional, and TO represents once to perform time, ST
The successful execution time is represented, PS represents successful execution probability;Tp, to, st and ps represent the occurrence of composite services corresponding concepts;
Each only one of which Component service performs failure in pattern.
The particular content of 3rd step is as follows:
Method for service selection of this method based on affairs and QoS, proposes two QoS indexes:Successfully complete time and execution
Price, the time that successfully completes obtains according to time performance analysis method, and strike price is according to each Component service of composite services
Strike price sum is obtained;Then processing is weighted to successfully completing time and strike price, to obtain combinations of transactions service
QoS fractional values, and be used as the standard that selection is carried out to combinations of transactions service.
The QoS index also includes prestige, availability.
This method also includes the method for (1) preference pattern;
In Web environment, composite services are typically made up of other services or composite services;Including three kinds of integrated modes:Sequence
Row, parallel, selection;As the similar service of more and more features is obtained, optimal Service Instance is therefrom selected;In order to true
The reliable and correct execution of composite services is protected, Services Composition needs the support of affairs;This method proposes that two kinds are based on affairs and QoS
Services selection model;
The first combinations of transactions services selection model determines services selection model for transactional;Composite services are retouched by workflow
State, suitable transactional has been assigned to all activities in workflow;Therefore, candidate service is via feature and affairs
Property require determine;
Second of combinations of transactions services selection model is the unknown preference pattern of transactional;Composite services are still retouched by workflow
State;But, movable transactional is that unknown, movable transactional performs movable candidate's clothes by being selected in workflow
The transactional of business is determined;Based on the two preference patterns, and the system of selection based on QoS is combined, this method also proposed accordingly
Global optimum's selection algorithm.
This method also includes (2) global optimum selection algorithm;
In the composite services preference pattern based on affairs and QoS, the QoS time performances of the introducings of affairs to composite services
Generate influence;Therefore, when selecting optimal candidate service not in view of affairs pair according to local optimal time index
Between performance influence;So, this method is selected service using global optimum's method;
First, based on selected model, the transactional of workflow is it has been determined that each active transaction of i.e. workflow is
It is known, workflow known to transactional is referred to as Transactional Workflows (Transactional Workflow, TWF) here;
Assuming that activity number is n, TWF={ a in Transactional Workflows TWF1,a2,…,an};To each movable a in Transactional Workflowsj, j
=1,2 ..., n have qjIndividual candidate's Transaction ServiceCan execution activity aj, the candidate service collection
Close satisfaction activity ajFeature and transactional requirement;It is assigned to each movable candidate service and can obtain one to can perform
Combinations of transactions service;So, an executable candidate's combinations of transactions of set can be obtained according to all candidate services to service
TCWS (Transactional Composite Web Service), TCWS={ tcws1,tcws2,…,tcwsm, m=Π
qj, here, j=1,2 ..., n;Then, optimal executable combinations of transactions is selected from set TCWS using global selection algorithm
Service;Here, object function is Score (tcwsi), it is that each executable combinations of transactions service uses simple method of weighting
To calculate the QoS fractional values of whole executable combinations of transactions service;Score(tcwsi)=∑jwjqij, wherein wj∈ [0,1],
∑jwj=1, i=1,2 ..., m, qijRepresent two QoS attributes of each candidate service;It is executable with minimum Score values
Combinations of transactions service is as optimal selection, if several executable combinations of transactions services are with minimum Score values, then
One is randomly choosed to perform combinations of transactions service;Global optimum selection algorithm (Global based on Transactional Workflows
Selection Based on TWF, GST) input to include the Transactional Workflows of n activity, be output as having minimum
The executable combinations of transactions service BestTCWS of Score values;Arthmetic statement is as follows:
The first step:According to each movable candidate service set, all executable things for meeting composite services requirement are obtained
Business composite services set,
Second step, according to time performance analysis algorithm, what the executable combinations of transactions of acquisition every was serviced successfully completes the time,
ST [i] is assigned to, while obtaining strike price, EP [i] is assigned to, is handled according to weighting, tcws [i] Score values are obtained;
3rd step, contrasts the Score values of all executable combinations of transactions services, execution transaction set minimum selection Score
Conjunction service is assigned to BestTCWS;Score [i] is the Score values of executable combinations of transactions;
Algorithm GST false codes are as follows:
Input:Transactional Workflows TWF
Output:Combinations of transactions service tcws with global minima Score values
Start
Terminate
This method also provides another option algorithm;According to workflow transactions assignment method, first, to workflow mould
Type carries out transactional assignment, i.e., only assign suitable transactional to workflow activities, and does not temporarily select suitable Service Instance
To perform activity;It is possible to obtain the Transactional Workflows TWF={ twf of a set1,twf2,…,twfm, both meet group
The Functional Requirement of service is closed, transactional requirement is met again;To each Transactional Workflows, it is complete to obtain to use GST algorithms
The optimal executable combinations of transactions service of office, then from these optimal executable combinations of transactions set of services, reselection one
Individual optimal combinations of transactions service meets user's request;Global optimum selection algorithm (Global based on workflow
Selection Based on WF, GSW) input be workflow, be output as the combinations of transactions with minimum Score values
Service BestTCWS;Arthmetic statement is as follows:
The first step, according to transactional assignment method, then they be assigned to collection by all feasible Transactional Workflows of activity
Close twf;
Each Transactional Workflows in twf are selected optimal combinations of transactions service, obtained simultaneously by second step using algorithm GST
Their Score values are obtained, and are assigned to BestTcws and MiniScore respectively;Finally, selected most from set BestTcws
Excellent combinations of transactions service BestTCWS;TempMiniScore is the variable that a temporary combinations of transactions services Score values;
Pseudo-code of the algorithm is as follows:
Input:Workflow WF
Output:Combinations of transactions service TCWS with global minima QoS
Start
Terminate.
This method also includes (3) heuristic selection algorithm;In global optimum selection algorithm GSW, when workflow activities
When number becomes very big, algorithm needs the time spent to become to be difficult to receive;So, it would be desirable in acceptable algorithm
In, the combinations of transactions service of a near-optimization is obtained to meet user's request;The algorithm GSW times are mainly expended two sides
Face:First, transactional assignment is carried out to workflow, all feasible Transactional Workflows are obtained;Second, to each Transactional Workflows
Optimal executable combinations of transactions service is obtained by algorithm GST;So, reduce the algorithms selection time from these two aspects, it is proposed that
One heuristic selection algorithm;
Heuristic evaluation function is used for assessing and finding the suitable path from current node to destination node;Calculated heuristic
In method, heuristic evaluation function is used for assessing the Score values for including the different businesslike whole combinations of transactions services of current node;
Based on heuristic evaluation function, the transactional of current active is selected first, then, is selected from the businesslike candidate service is met
Optimal service is selected to perform current active;
In sequence and parallel schema, if the transactional of current active is entered as p or pr, then all movable below
Transactional is all necessary for pr or cr;In parallel schema, if the transactional of current active is entered as c, then, comprising current living
The activity of other all branches of dynamic parallel schema is only c or cr;In selection mode, the transactional and the mould of branch's activity
The transactional of the branch of other in formula is identical;Which reduces the time of workflow transactions assignment;Then affairs are determined
Property movable all candidate services QoS average values as reference information, acquisition determines businesslike thing comprising current active
The successful execution time of business composite services and strike price, and processing acquisition Score values are weighted as heuristic evaluation function
Value;So we just reduce the time of GST selection algorithms;
Here, heuristic evaluation function HEF is divided into three parts, be defined as follows;
Define 1:Heuristic evaluation functionAnd:
(1)G1(x) it is the information of current active;
(2)G2(x) candidate of combinations of transactions service is had to enter into for historical act and knowing according to current active transactional
Information on services;
(3)G3(x) it is unknown transactional action message;
In order to obtain heuristic evaluation function, it should be understood that road comprising current active determine the service of businesslike combinations of transactions into
Work(deadline and strike price;Here, all activities of workflow are divided into four parts by we:
Part I is the activity for having selected for suitable Transaction Service, is referred to as historical information;So, this part is lived
Dynamic successfully completes time and strike price just as G2(x) partial information in;
Part II is current active;If current active determines transactional, then meet the institute of transactional requirement
There is the average qos value of candidate service as G1(x) information;
Part III is that businesslike follow-up activities are determined according to current active transactional;According to these movable affairs
Property, then the average qos value of all candidate services of transactional requirement is met as G2(x) another part information;
Part IV is the unknown activity of transactional;First, all possible transactional assignment, then identical movement are enumerated
The average qos value of all candidate services;It is used as G3(x) information included of information;
According to this four partial information, we can pass throughObtain and include current active
Combinations of transactions service path the successful execution time and strike price, then weighting processing gains enlightenment formula valuation functions HEF;
Heuristic selection algorithm HSW (Heuristic Selection based on WF) process is:Consider current work
It is dynamic, according to the above affairs feature of chosen appropriate services, the affairs feature set for meeting current active requirement can be obtained
tpset;If current active selection transactional is p or pr service, the candidate service for meeting the business requirements is G1(x) believe
Breath;If activity is the branch of selection mode, remaining branch's active transaction of the selection mode is also p or pr, remaining activity thing
Business property is pr or cr;The service of above active candidate is G2(x) information, and without G3(x) information;Current active selection transactional is c
Service when, meet the business requirements candidate service be G1(x) information;If the branch of selection mode, the selection mode
Remaining branch's active transaction is also c, if in parallel organization, remaining branch's active transaction of the parallel schema is c or cr,
The service of above active candidate is G2(x) information;The unknown activity of remaining transactional is G3(x) information;Current active selects transactional
For cr service when, meet the business requirements candidate service be G1(x) information;If the branch of selection mode, the selection
Remaining branch's active transaction of pattern is also cr, is G2(x) information;Otherwise activity below is all unknown, is G3(x) information;Than
The HEF of more each affairs feature, is that current active selects a transactional, and selected from the businesslike candidate service is met
Optimal service performs current active;In following heuritic approach, TP represents above the thing of chosen appropriate services
Business feature, TCWS represents the combinations of transactions service of selection;
Algorithm HSW false codes are as follows:
Algorithm HSW
Input:Include the workflow WF of n element
Output:Executable combinations of transactions service TCWS with global minima QoS
Start
Terminate.
A kind of service quality performance prediction meanss combined based on affairs and QoS of use above method, the device is also wrapped
Include:
Extraction module, for traveling through the composite services according to the workflow sequence of composite services, extracts the combination clothes
Component service in business;
Processing module, for judging whether the Component service is atomic service, if so, then calculating the atom clothes
The performance parameter of business, and assign corresponding transactional for the atomic service;Otherwise, set algorithm is called by the Component service
It is integrated into atomic service, and calculates the atomic service performance parameter, and corresponding affairs is assigned for the atomic service
Property;
The extraction module, is additionally operable to extract next Component service in the composite services, and return to continuation judgement
Whether next Component service is atomic service, untill all component service is traversed in the composite services;
Determining module, for the performance parameter according to each atomic service calculated and the affairs being each endowed
Property, determine the service quality performance of the composite services.
The technical scheme prediction accuracy that the present invention is provided is good, and with higher feasibility and validity.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
The accompanying drawing to be used needed for having technology description is briefly described, it should be apparent that, drawings in the following description are this hairs
Some bright embodiments, for those of ordinary skill in the art, without having to pay creative labor, can be with
Other accompanying drawings are obtained according to these accompanying drawings.
Fig. 1 is the schematic diagram that integrated mode of the present invention is sequence pattern;
Fig. 2 is the schematic diagram that integrated mode of the present invention is constellation;
Fig. 3 is the schematic diagram that integrated mode of the present invention is selection mode;
Fig. 4 is that the transactional in the embodiment of the present invention determines preference pattern;
Fig. 5 is the unknown preference pattern of transactional in the embodiment of the present invention.
Embodiment
Realize the technical scheme is that:Transactional is assigned to Web service first, then analysis has transactional requirement
The combinatorial problem of service, secondly analysis has the performance issue of transaction attributes composite services, finally, according to transactional composite services
Performance comparision, propose the method for service selection combined based on affairs and QoS.Specifically include following steps:
The first step, when carrying out the selection of current component service, according to the affairs that above chosen suitable assembly is serviced
Feature, and the rule serviced according to combinations of transactions requires that acquisition meets the candidate component service of transactional requirement.
Second step, to each candidate component service carry out Performance Evaluation, assess consider Web service QoS and transactional will
Ask.
3rd step, repetition above procedure, desired composite services are met until obtaining one.
The particular content of combinations of transactions method of servicing and Performance Evaluation is as follows in the first step and second step:
In common method for service selection, the QoS that Performance Evaluation is all based on all component service is analyzed, and this hair
It is bright to be based on transactional and QoS, both are combined the composite services system of selection got based on affairs.
In the composite services selection course based on affairs, when the similar candidate component service of some features can be obtained
When obtaining, their non-functional requirements QoS just becomes extremely important.Because QoS can reflect the demand of user well, so
It is non-functional index the most frequently used at present, and the present invention considers three QoS indexes below Web service ws:
(1) strike price (Execution price, EP):Requestor calls the cost required for service ws, is expressed as qep
(ws)。
(2) time (Execution time, ET) is performed:Service ws and perform once the required time, be expressed as qet
(ws).It is recognized herein that the time of service execution once is identical, and within the identical time, service execution may once be lost
Lose, it is also possible to successfully complete.
(3) probability (Probability of Success, PS) is successfully completed:Service ws successfully completes the general of user's request
Rate, is expressed as qps(ws)。
Wherein, the execution time needs the change that the transactional for considering composite services is brought, and analysis process is as follows:
(1) service has compensability, then it can provide compensation policy to cancel the influence of the service, use symbol
" c " is represented.
(2) services have can continuous retrying property, then it can be retried, and be reached by sufficient number of retry
Final success, is represented with symbol " r ".
(3) services have center dot characteristics, then it is once run succeeded, and its influence is permanently present, and not
It can be revoked, if it performs failure, without any influence, be represented with symbol " p ".
Due to the dynamic of network environment, uncertain and open, the Component service from different tissues is often adjusted
With failure, it is assumed that the probability of a Web service successful execution is ps.The time of service execution once is t, then in time t
One Web service either successful execution or performs failure.Service all fails in first n-1 times execution, what n-th ran succeeded
Probability meets geometry distribution, and probability function is:
P (X=n)=(1-ps)n-1ps
Χ represents which time of service is performed, then mathematical expectation is:
The mathematic expectaion reaction of geometry distribution is exactly that the average time for successfully averagely performing number of times, i.e. successful execution is
1/ps, the time of service execution once is t, then, the average time st of successful execution is:
St=t/ps
Generally, composite services are made up of basic integrated mode, therefore in order to consider the time performance of combinations of transactions service, I
Need to analyze the time performance of basic integrated mode.Here, we discuss four kinds of basic integrated modes, are respectively:Sequence,
Parallel, selection and circulation pattern.And Services Composition is described by Work flow model.TP represents transactional, and TO represents once to hold
Row time, ST represents the successful execution time, and PS represents successful execution probability.Tp, to, st and ps represent composite services corresponding concepts
Occurrence.In order to simply consider, each only one of which Component service performs failure in pattern.
(1) sequence pattern time performance analysis
Sequence pattern is used for combinations of definitions service and taken by a series of components performed by permanent order, by n Component service
The sequence pattern of composition is as shown in Figure 1.In sequence pattern, Component service is from left to right performed, the component before and if only if
Service after the completion of successful execution, Component service below could start to perform.If Component service performs failure, thing will be utilized
Business mechanism processing failure.The preceding i Component service successful execution time is sti, Component service WSiThe time performed once is ti, clothes
Be engaged in WSiThe probability of successful execution is psi。
The composite sequence pattern of two service compositions is considered first, and the integrated mode of n service composition is considered as preceding n-1
The composite services of service and the combination of n-th of service.
In sequence pattern, if first service WS1The transactional attribute being endowed is p or pr, according to 4.3 Hes of definition
4.6, WS1Once successfully completing, it cannot be revoked, therefore second service WS2It must assure that and necessarily successfully complete, be i.e. WS2Must
R transactional must be included, so WS2Imparting transactional is pr or cr.{tp1,tp2In intermediate scheme from left to right or from upper
The transactional of two services under.
If sequence pattern is { p, pr } or { p, cr }, the once pattern is performed, if WS1Perform failure, then pattern
Failure is performed, the time is t1.If WS1Successful execution is completed, due to WS2The completion that is sure of success is retried by finite number of time, so
Pattern successful execution is completed, and the time is t1+t2/ps2.So in t1+t2/ps2In time, we may determine that the pattern whether into
Work(performs completion, therefore to=t1+t2/ps2。WS1It is ps that successful execution, which completes probability,1, due to WS2Being retried by finite number of time must
Successfully complete, in pattern implementation procedure, it is believed that it is 1, i.e. WS that service execution of the transactional comprising r, which completes probability,2Success
It is 1 to perform and complete probability.So the successful execution probability of the composite services is ps=ps1, the transactional of composite services is p.
If WS1Failure is performed, then pattern performs failure, in order to analyze the time that successfully completes of the pattern, WS1Will weight
It is multiple to perform, until WS1Successful execution is completed, then pattern is successfully completed.Assuming that WS1The successful execution deadline be st1(such as
Fruit WS1For atomic service, st1=t1/ps1If, WS1For composite services, st1The time is successfully completed for composite services), due to
WS2The completion that is sure of success is retried by finite number of time, then st=st1+t2/ps2。
If sequence pattern is { pr, pr } or { pr, cr }, a pattern, WS are performed1And WS2All retried by finite number of time
Be sure of success completion, then to=t1/ps1+t2/ps2, it is also the successful execution deadline to perform a time, therefore st=st1+
t2/ps。WS1And WS2It is all 1 that successful execution, which completes probability, then the probability of pattern successful execution is ps=1, the thing of composite services
Business property is pr.
In sequence pattern, if first service WS1The transactional of imparting is c or cr, then if second service
WS2Failure is performed, first service execution can be revoked, so second service WS2Any transactional can be assigned.
In sequence pattern, if pattern is { p, pr } or { p, cr }, then, tp=p, to=t1+t2/ps2, st=st1
+t2/ps2, ps=ps1.If sequence pattern is { pr, pr } or { pr, cr }, then, tp=pr, to=t1/ps1+t2/ps2, st
=st1+t2/ps2, ps=1.
In sequence pattern, if pattern is { c, p } or { c, c }, then, tp=p or c, to=t1+t2, st=(st1+
t2)/ps2, ps=ps1ps2.If sequence pattern is { c, pr } or { c, cr }, tp=p or c, to=t1+t2/ps2, st=st1+
t2/ps2, ps=ps1.If sequence pattern is { cr, p } or { cr, c }, tp=p or c, to=t1/ps1+t2, st=(st1+t2)/
ps2, ps=ps2.If sequence pattern is { cr, pr } or { cr, cr }, tp=pr or cr, to=t1/ps1+t2/ps2, st=st1
+t2/ps2, ps=1.
If sequence pattern is { c, p } or { c, c }, the once pattern is performed, if WS1Perform failure, then pattern is held
Row failure, the time is t1.If WS1Successful execution is completed, WS2Perform once, if WS2Perform failure, pattern failure, such as
Fruit WS2Successful execution complete, pattern successful execution complete, the time is t1+t2.So in t1+t2In time, we may determine that
Whether successful execution is completed pattern, then to=t1+t2。
If WS1Perform failure, then pattern performs failure, equally, repeats WS1, until WS1Successful execution is completed.
Assuming that WS1Successful execution complete time be st1, WS2Execution can unsuccessfully cause WS1Compensated, and roll back to pattern
Start position, then the pattern of also resulting in performs failure.Equally, WS is repeated2, until WS2Run succeeded completion.WS2Every time
Failure can all cause WS1And WS2Re-execute once, until WS1And WS2It is carried out success.So service WS2Held by k times
Successfully completed after row (i.e. by k-1 times perform failure after success) time be:k(st1+t2), probability is:
P (X=k)=(1-ps2)k-1ps2
So successfully completing time i.e. mathematic expectaion is
WS1And WS2Successful execution complete probability be ps respectively1And ps2, then pattern successful execution completes probability and is
Ps=ps1ps2.If pattern is { c, p }, tp=p.If pattern is { c, c }, tp=c.
If sequence pattern is { c, pr } or { c, cr }, then WS2Retry one by finite number of time and surely successfully complete and (determine
Justice 4.3), WS1And WS2The successful execution deadline be st respectively1And t2/ps2, so st=st1+t2/ps2.Similar theorem
Sequence pattern is the proof of { p, pr } and { p, cr }, to=t in 4.11+t2/ps2, ps=1.
If sequence pattern is { cr, p } or { cr, c }, the once pattern, WS are performed1Retry and be sure of success by limited number of time
Complete and (define 4.3), WS1The successful execution time be st1=t1/ps1Although (with the WS that transactional described above is c1Successfully hold
The row time is identical, but reason is different).WS2Perform once, according to WS2Implementing result, pattern or perform failure or into
Work(performs completion, and the time is t1/ps1+t2.So in t1/ps1+t2In time, we may determine that pattern, and whether successful execution is complete
Into, then to=t1/ps1+t2.Similar sequence pattern is the proof procedure of { c, p } or { c, c }, st=(st1+t2)/ps2。WS1
And WS2Successful execution complete probability be 1 and ps respectively2, then, ps=ps2.If pattern is { cr, p }, tp=p.If
Pattern is { cr, c }, tp=c.
If sequence pattern is { cr, pr/cr }, the explanation of similar sequence pattern { pr, pr/cr }.
The sequence pattern transactional assignment of table 1 and time performance analysis
(2) parallel schema time performance computational methods
Parallel schema, which is used for Component service in combinations of definitions service, does not have strict implement order, must be while the combination carried out
Service.Its pattern is as shown in Figure 2.When only all Parallel Services all successful executions are completed, this composite services could successfully be held
Row is completed.If some service successful executions are completed, other service executions failure, then need using affair mechanism to performing
The service of failure is handled.Here, max { x, y } represents both x and y maximum.Min { x, y } represents both x and y most
Small value.Assuming that when performing parallel schema service, priority is to perform from top to bottom.We consider what two services were constituted first
Parallel schema.
In parallel schema, if service WS1Transactional be p, once WS1Successful execution is completed, and the influence that it is produced is not
It can be revoked, therefore service WS2It must assure that certain successful execution is completed, i.e. WS2Transactional comprising r.On the other hand, once
WS1Perform failure, WS2It must be able to provide compensation policy, so WS2Transactional comprising c, then WS2Transactional be only
cr。
Work as WS1And WS2When performing simultaneously, due to WS2Transactional be cr, WS2It can ensure into by calling for finite number of time
Work(performs completion, and the time is t2/ps2。WS1The time performed once is t1.So in time max { t1,t2/ps2In, parallel mould
Formula may determine whether that successful execution is completed.So, to=max { t1,t2/ps2}。WS1Failure can cause WS2Compensated, made
Obtain parallel schema and perform failure, so we repeat the pattern until successful execution is completed.WS1By k times perform after into
Work(complete probability be:
P (X=k)=(1-ps1)k-1ps1
So, the successful execution time is:
The probability of the completion that runs succeeded of parallel schema is WS1And WS2Run succeeded completion probability product, and WS2Execution
It is 1 to successfully complete probability, then, ps=ps1, tp=p.
In parallel schema, if a service WS1Transactional be pr, WS1By calling for finite number of time, one surely
Successful execution is completed, and after successfully completing, its influence cannot be revoked (definition 4.3).So another service WS2It is necessary
Assure success and perform completion, i.e., transactional must include r, then service WS2Transactional be pr or cr.
In parallel schema, if pattern is { pr, pr } or { pr, cr }, then, tp=pr, to=max { t1/ps1,t2/
ps2, st=max { t1/ps1,t2/ps2, ps=1.
Due to WS1And WS2Transactional all include r, performed so they can retry to assure success by limited number of time
Into, and the time performed once is respectively t1/ps1And t2/ps2, and perform time once be exactly successful execution complete when
Between, the probability that successful execution is completed all is 1.So in parallel schema, to=max { t1/ps1,t2/ps2, st=max { t1/
ps1,t2/ps2, ps=1, tp=pr.
In parallel schema, if a service WS1Transactional be c, then WS1May failure.So another is serviced
WS2Must be compensable.That is WS2Be transactional include c, then, WS2Transactional can be c or cr[64]。
In parallel schema, if pattern is { c, cr }, tp=c, to=max { t1,t2/ps2, st=max { t1,t2/
ps2}/ps1, ps=ps1.If parallel schema is { c, c }, tp=c, to=max { t1,t2, st=max { t1,t2}/min
{ps1,ps2, ps=ps1ps2。
If parallel schema is { c, cr }, it was demonstrated that similar, no longer repeat here.If parallel schema is { c, c }, due to
WS1And WS2The time performed respectively once is t1And t2, therefore to=max { t1,t2, due to WS1And WS2Compensation plan can be provided
Slightly, after compensation policy is provided for some service (service execution failure), a parallel schema is re-executed every time, it is actual
Upper two services all compensate for once.So compensation number of times depends on needing more to be serviced by compensation number of times, that is, into
Work(, which is performed, completes the smaller service of probability.Therefore st=max { t1,t2}/min{ps1,ps2}。WS1And WS2It is general that successful execution is completed
Rate is respectively ps1And ps2, therefore ps=ps1ps2.According to definition 4.4, tp=c.
In parallel schema, if a service WS1Transactional be cr, according to define 4.2, call WS by limited1Energy
Enough assure success completion.Also, work as WS2Perform failure, WS1It is compensable.So the WS of another service2Can be any
Businesslike service.
If parallel schema is { cr, cr }, then, tp=cr, to=max { t1/ps1,t2/ps2, st=max { t1/
ps1,t2/ps2, ps=1.
Because service is symmetrical in parallel schema, parallel schema { cr, p }, the formula of { cr, c } and { cr, pr } can join
The respective formula provided in theorem 4.3,4.4,4.5 is examined, is no longer repeated here.The formula of parallel schema { cr, cr } may be referred to
The proof of pattern { pr, pr } in theorem 4.3.
It is as shown in the table for the transactional assignment and time performance analysis of parallel schema based on two services.
The parallel schema transactional assignment of table 2 and time performance analysis
(3) selection mode time performance analysis
In this section, it is contemplated that two kinds of selection modes:One kind is that mutually restriction is closed with repelling for defining each other
Branch's service of system, this kind of branch's service is often performed according to specific implementation status from multiple one branch of branching selection, choosing
Selecting the service of pattern only one of which can be performed.So the discussion of sequence pattern can be attributed to.Another option pattern is
The selection mode of optional one, branch's service can be substituted, and after some branch, which performs, to fail, select another branch
Substitute, until the branch succeeded, or all branches all fail, then compensate, as shown in Figure 3.We mainly examine
Consider the time performance analysis of second of selection mode.
Assuming that when performing selection mode service, priority is to perform from top to bottom.We consider two service groups first
Into selection mode.
In selection mode, two services all assign the service of same transaction characteristic, then the transactional of pattern is component
Service corresponding transactional, to=(t1+t2), st=(t1+t2)/(ps1+ps2-2ps1ps2), ps=ps1+ps2-2ps1ps2。
In selection mode, due to being chosen the service that service is all completion said function, so, herein, in order to
For the sake of simple, it is believed that the transactional that service assigns service in selection mode is reciprocity, that is to say, that in selection mode
In, the service that service can only assign same transaction could meet the physical meaning of selection mode.So, two services assign phase
With businesslike service.In selection mode, if the transactional of Component service is p or c, then if WS1Successful execution is complete
Into the time is t1If, WS1Perform failure, WS2Successful execution is completed, and the time is t1+t2If, WS1And WS2All fail, parallel
Pattern fails, and the time is t1+t2.So in t1+t2, it inside may determine that whether successful execution is completed parallel schema, therefore to=(t1+
t2).Because the service of selection mode only one of which enters in combinations of transactions service, therefore do not consider the ST that individually services.First clothes
It is engaged in successfully, pattern success, probability is, first serv-fail, and second successful probability of service is (1-ps1)ps2, transaction set
It is ps to close the successful probability of service1+(1-ps1)ps2=ps1+ps2-ps1ps2, TCWS transactional is p or c.Composite services are selected
The transactional assignment and time performance analysis of pattern are as shown in table 3.Such as in selection mode, if the transactional of Component service is
Pr or cr, pattern only needs to first Component service of execution can be make it that pattern successful execution is completed, st=to=t1/ps1,
Ps=1.
The selection mode transactional assignment of table 3 and time performance analysis
The particular content of combinations of transactions method for service selection is as follows in 3rd step:
This section mainly studies the method for service selection based on affairs and QoS, and considers two QoS indexes:Successfully complete
Time and strike price, the former can obtain according to time performance analysis algorithm above, and the latter can be every according to composite services
The strike price sum of individual Component service is obtained.Then both are weighted with processing, to obtain QoS points of combinations of transactions service
Numerical value, and it is used as the standard that selection is carried out to combinations of transactions service.Due to it is used herein be the QoS attributes of standard, so its
Its QoS index, such as:Prestige, availability etc., consideration can be extended by same mode.
(1) preference pattern
In Web environment, composite services are typically made up of other services or composite services, including three kinds of integrated modes:Sequence
Row, parallel, selection.As the similar service of more and more features is obtained, optimal Service Instance should be selected therefrom
Come.At present, most of work is all based on QoS service is selected.But, in order to ensure composite services are reliable and correct
Perform, Services Composition needs the support of affairs.Then, it is proposed that two kinds of services selection models based on affairs and QoS.
The first combinations of transactions services selection model is as shown in figure 4, referred to as transactional determines services selection model.Combination clothes
Business is by work stream description, and suitable transactional has been assigned to all activities in workflow.Therefore, candidate service via
Feature and transactional require to determine.Such as, according to Functional Requirement, movable A1 have three candidate's Transaction Services S11, S12 and
S13.Required according to transactional, only S12 and S13 transactional are c.It therefore meets the requirement of both feature and transactional
Candidate service is S12 and S13, then selects optimal service from both to perform movable A1 according to qos requirement.
However, may have different affairs there is provided the candidate service of identity function in the Services Composition based on affairs
Property.So, second of combinations of transactions services selection model is as shown in figure 5, the referred to as unknown preference pattern of transactional.Composite services are still
So by work stream description.But, movable transactional is that unknown, movable transactional is somebody's turn to do by being selected to perform in workflow
The transactional of the candidate service of activity is determined.Such as, according to Functional Requirement, movable A1There are three candidate's Transaction Service S11, S12
And S13.If activity A1It has selected candidate service S11, then movable transactional is p, if activity A1It has selected candidate service
S12Or S13, then movable transactional is c.According to work(transactional and qos requirement, optimal service is selected from three
Execution activity A1。
Based on the two preference patterns, and the system of selection based on QoS more commonly used at present is combined, this paper presents phase
The global optimum's selection algorithm answered.
(2) global optimum's selection algorithm
In the composite services preference pattern based on affairs and QoS, the QoS time performances of the introducings of affairs to composite services
Generate influence.Therefore, when selecting optimal candidate service not in view of affairs pair according to local optimal time index
Between performance influence.So, service is selected using global optimum's method herein.
First, based on the preference pattern in Fig. 4, the transactional of workflow is it has been determined that i.e. each activity thing of workflow
Business property be it is known, here workflow known to transactional be referred to as Transactional Workflows (Transactional Workflow,
TWF).Assuming that activity number is n, TWF={ a in Transactional Workflows TWF1,a2,…,an}.To each work in Transactional Workflows
Dynamic aj, j=1,2 ..., n have qjIndividual candidate's Transaction ServiceCan execution activity aj, the candidate
Set of service satisfaction activity ajFeature and transactional requirement.Can be obtained by being assigned to each movable candidate service by one
Executable combinations of transactions service.So, an executable candidate's transaction set of set can be obtained according to all candidate services
Close service TCWS (Transactional Composite Web Service), TCWS={ tcws1,tcws2,…,tcwsm, m
=∏ qj, here, j=1,2 ..., n.Then, optimal executable affairs are selected from set TCWS using global selection algorithm
Composite services.Here, object function is Score (tcwsi), it is that each executable combinations of transactions service uses simple weighting
Method entirely can perform the QoS fractional values of combinations of transactions service to calculate.Score(tcwsi)=∑jwjqij, wherein wj∈[0,
1], ∑jwj=1, i=1,2 ..., m, qijRepresent two QoS attributes of each candidate service.With holding for minimum Score values
Business composite services are acted as optimal selection, if several executable combinations of transactions services have minimum Score values, that
One is randomly choosed to perform combinations of transactions service.Global optimum selection algorithm (Global based on Transactional Workflows
Selection Based on TWF, GST) input to include the Transactional Workflows of n activity, be output as having minimum
The executable combinations of transactions service BestTCWS of Score values.Arthmetic statement is as follows:
The first step:According to each movable candidate service set, all executable things for meeting composite services requirement are obtained
Business composite services set (see row 2 to 3),
Second step, according to time performance analysis algorithm, what the executable combinations of transactions of acquisition every was serviced successfully completes the time,
ST [i] (see row 5) is assigned to, while obtaining strike price, EP [i] (see row 6) is assigned to, is handled according to weighting, tcws is obtained
The Score values (see false code row 7) of [i].
3rd step, contrasts the Score values of all executable combinations of transactions services, execution transaction set minimum selection Score
Conjunction service is assigned to BestTCWS (see false code row 9-11).Score [i] is the Score values of executable combinations of transactions.
In actual combinations of transactions service selection process, the preference pattern in Fig. 5 can more meet ISP's kimonos
Make sure the demand of user.The selection algorithm based on model in Fig. 5 is considered below.
Preference pattern based on Fig. 5, has and has researched and proposed one based on local affairs and QoS selection algorithms, due to right
Activity carries out affairs feature of the Transaction Service selection dependent on above chosen suitable Transaction Service, so, global selecting party
Method can not be used, and only local optimal selection algorithm is presented.As discussed above, influence of the affairs to QoS is simultaneously
It is not considered.Also, local optimal selection be by not considering other activities under the influence of, single service is selected, this
The local optimal service of sample selection is probably suboptimum in the overall situation.Therefore, in order to consider the influence of affairs, and local choosing is overcome
The limitation selected, global selection algorithm is provided based on the preference pattern in Fig. 5.
According to the workflow transactions assignment method above provided, first, we can carry out affairs to Work flow model
Property assignment, i.e., only give workflow activities to assign suitable transactional, and temporarily do not select suitable Service Instance to perform activity.
So, we can obtain the Transactional Workflows TWF={ twf of a set1,twf2,…,twfm, both meet composite services
Functional Requirement, meets transactional requirement again.To each Transactional Workflows, GST algorithms given above can be used to obtain
The executable combinations of transactions service of global optimum, then from these optimal executable combinations of transactions set of services, reselection
One optimal combinations of transactions service meets user's request.Global optimum selection algorithm (Global based on workflow
Selection Based on WF, GSW) input be workflow, be output as the combinations of transactions with minimum Score values
Service BestTCWS.Arthmetic statement is as follows:
The first step, according to transactional assignment method, then they be assigned to collection by all feasible Transactional Workflows of activity
Close twf (see row 1-3).
Each Transactional Workflows in twf are selected optimal combinations of transactions service, obtained simultaneously by second step using algorithm GST
Their Score values are obtained, and are assigned to BestTcws and MiniScore respectively.Finally, selected most from set BestTcws
Excellent combinations of transactions service BestTCWS (see row 6-13).TempMiniScore is a temporary combinations of transactions service Score value
Variable.
(3) heuristic selection algorithm
In global optimum selection algorithm GSW, when workflow activities number becomes very big, algorithm needs the time spent
It may become to be difficult to receive.So, it would be desirable within the acceptable algorithm time, obtain the combinations of transactions of a near-optimization
Service meets user's request.The algorithm GSW times are mainly expended at two aspects:First, transactional assignment is carried out to workflow,
Obtain all feasible Transactional Workflows.Second, optimal executable transaction set is obtained by algorithm GST to each Transactional Workflows
Close service.So, it is contemplated that reducing the algorithms selection time from these two aspects, it is proposed that a heuristic selection algorithm.
Generally, heuristic evaluation function is used for assessing and finding the suitable path from current node to destination node.At me
Heuritic approach in, heuristic evaluation function is used for assessing comprising the different businesslike whole combinations of transactions clothes of current node
The Score values of business.Based on heuristic evaluation function, we select the transactional of current active first, then, from meeting the affairs
Optimal service is selected to perform current active in the candidate service of property.
In order to design more preferable heuristic evaluation function, we must obtain Given information as much as possible.According to above work
Making stream transactional assignment procedure, we understand, in sequence and parallel schema, if the transactional of current active be entered as p or
Pr, then all movable transactionals are all necessary for pr or cr below.In parallel schema, if the transactional of current active is assigned
It is worth for c, then, the activity of other all branches of the parallel schema comprising current active is only c or cr.In selection mode,
The transactional of branch's activity is identical with the transactional of other branches in the pattern.So we just reduce workflow transactions tax
The time of value.Then the QoS average values that businesslike movable all candidate services are determined are wrapped as reference information
Successful execution time and the strike price of businesslike combinations of transactions service are determined containing current active, and weights processing acquisition
Score values as heuristic evaluation function value.So we just reduce the time of GST selection algorithms.
Here, heuristic evaluation function HEF is divided into three parts by we, is defined as follows.
Define 1:Heuristic evaluation functionAnd:
(1)G1(x) it is the information of current active.
(2)G2(x) candidate of combinations of transactions service is had to enter into for historical act and knowing according to current active transactional
Information on services.
(3)G3(x) it is unknown transactional action message.
In order to obtain heuristic evaluation function, we must be known by determining businesslike combinations of transactions clothes comprising current active
That is engaged in successfully completes time and strike price.Here, all activities of workflow are divided into four parts by we:
Part I is the activity for having selected for suitable Transaction Service, is referred to as historical information.So, this part is lived
Dynamic successfully completes time and strike price just as G2(x) partial information in.
Part II is current active.If current active determines transactional, then meet the institute of transactional requirement
There is the average qos value of candidate service as G1(x) information.
Part III is that businesslike follow-up activities are determined according to current active transactional.According to these movable affairs
Property, then the average qos value of all candidate services of transactional requirement is met as G2(x) another part information.
Part IV is the unknown activity of transactional.First, all possible transactional assignment, then identical movement are enumerated
The average qos value of all candidate services.It is used as G3(x) information included of information.
According to this four partial information, we can pass throughObtain and include current active
Combinations of transactions service path the successful execution time and strike price, then weighting processing gains enlightenment formula valuation functions HEF.
Heuristic selection algorithm HSW (Heuristic Selection based on WF) process is:Consider current work
It is dynamic, according to the above affairs feature of chosen appropriate services, the affairs feature set for meeting current active requirement can be obtained
tpset.If current active selection transactional is p or pr service, the candidate service for meeting the business requirements is G1(x) believe
Breath.If activity is the branch of selection mode, remaining branch's active transaction of the selection mode is also p or pr, remaining activity thing
Business property is pr or cr.The service of above active candidate is G2(x) information, and without G3(x) information.Current active selection transactional is c
Service when, meet the business requirements candidate service be G1(x) information.If the branch of selection mode, the selection mode
Remaining branch's active transaction is also c, if in parallel organization, remaining branch's active transaction of the parallel schema is c or cr,
The service of above active candidate is G2(x) information.The unknown activity of remaining transactional is G3(x) information.Current active selects transactional
For cr service when, meet the business requirements candidate service be G1(x) information.If the branch of selection mode, the selection
Remaining branch's active transaction of pattern is also cr, is G2(x) information.Otherwise activity below is all unknown, is G3(x) information.Than
The HEF of more each affairs feature, is that current active selects a transactional, and selected from the businesslike candidate service is met
Optimal service performs current active.In following heuritic approach, TP represents above the thing of chosen appropriate services
Business feature, TCWS represents the combinations of transactions service of selection.
Claims (9)
1. the service quality performance Forecasting Methodology combined based on affairs and QoS, it is characterised in that this method includes:
Transactional is assigned to Web service first, then analysis requires the combinatorial problem of service with transactional;
Secondly analysis has the performance issue of transaction attributes composite services;
Finally, according to the Performance comparision of transactional composite services, the method for service selection combined based on affairs and QoS is proposed;Should
Method specifically includes following steps:
The first step, when carrying out the selection of current component service, according to the affairs feature that above chosen suitable assembly is serviced,
And the rule serviced according to combinations of transactions is required, obtains the candidate component service for meeting transactional requirement;
Second step, to each candidate component service carry out Performance Evaluation, assess consider Web service QoS and transactional requirement;
3rd step, repetition above procedure, desired composite services are met until obtaining one.
2. according to the method described in claim 1, it is characterised in that:
The first step, second step particular content it is as follows:
In common method for service selection, the QoS that Performance Evaluation is all based on all component service is analyzed, and this method is based on
Transactional and QoS, are combined both the composite services system of selection got based on affairs;
In the composite services selection course based on affairs, when the similar candidate component service of some features can be obtained
When, their non-functional requirements QoS can reflect the demand of user, and this method proposes that three QoS refer to below Web service ws
Mark:
(1) strike price (Execution price, EP):Requestor calls the cost required for service ws, is expressed as qep
(ws);
(2) time (Execution time, ET) is performed:Service ws and perform once the required time, be expressed as qet(ws);This
Method thinks that the time of service execution once is identical, and within the identical time, service execution may once fail,
It may successfully complete;
(3) probability (Probability of Success, PS) is successfully completed:Service ws successfully completes the probability of user's request,
It is expressed as qps(ws);
Wherein, the execution time needs the change that the transactional for considering composite services is brought, and analysis process is as follows:
(1) service has compensability, then it can provide compensation policy to cancel the influence of the service, with symbol " c "
Represent;
(2) services have can continuous retrying property, then it can be retried, and by it is sufficient number of retry reach it is final
Success, is represented with symbol " r ";
(3) services have center dot characteristics, then it is once run succeeded, and its influence is permanently present, and can not be by
Revocation, if it performs failure, without any influence, is represented with symbol " p ";
Dynamic, uncertain and opening due to network environment, the Component service from different tissues often call mistake
Lose, it is assumed that the probability of a Web service successful execution is ps, the time of service execution once is t, then one in time t
Web service either successful execution or performs failure, and service all fails in first n-1 times execution, the probability that n-th runs succeeded
Meet geometry distribution, probability function is:
P (X=n)=(1-ps)n-1ps
Χ represents which time of service is performed, then mathematical expectation is:
The mathematic expectaion reaction of geometry distribution is exactly that the successful average time for averagely performing number of times, i.e. successful execution is 1/
Ps, the time of service execution once is t, then, the average time st of successful execution is:
St=t/ps.
3. according to the method described in claim 1, it is characterised in that:
Composite services are made up of basic integrated mode, for the time performance of analysis combinations of transactions service, it is proposed that basic combination die
The time performance analysis method of formula;Four kinds of basic integrated modes are respectively:Sequence pattern, parallel schema, selection mode and follow
Ring mode;And Services Composition is described by Work flow model, TP represents transactional, and TO is represented once to perform the time, and ST is represented
Successful execution time, PS represents successful execution probability;Tp, to, st and ps represent the occurrence of composite services corresponding concepts;Pattern
In every time only one of which Component service perform failure.
4. according to the method described in claim 1, it is characterised in that:
The particular content of 3rd step is as follows:
Method for service selection of this method based on affairs and QoS, proposes two QoS indexes:Time and strike price are successfully completed,
Successfully complete the time obtains according to time performance analysis method, and strike price is according to the strike price of each Component service of composite services
Lattice sum is obtained;Then processing is weighted to successfully completing time and strike price, to obtain QoS points of combinations of transactions service
Numerical value, and it is used as the standard that selection is carried out to combinations of transactions service.
5. according to the method described in claim 1, it is characterised in that:The QoS index also includes prestige, availability.
6. according to the method described in claim 1, it is characterised in that:This method also includes the method for (1) preference pattern;
In Web environment, composite services are typically made up of other services or composite services;Including three kinds of integrated modes:Sequence, and
OK, select;As the similar service of more and more features is obtained, optimal Service Instance is therefrom selected;In order to ensure group
The reliable and correct execution of service is closed, Services Composition needs the support of affairs;This method proposes two kinds of clothes based on affairs and QoS
Business preference pattern;
The first combinations of transactions services selection model determines services selection model for transactional;Composite services by work stream description,
Suitable transactional has been assigned to all activities in workflow;Therefore, candidate service is via feature and transactional
It is required that determining;
Second of combinations of transactions services selection model is the unknown preference pattern of transactional;Composite services are still by work stream description;
But, movable transactional is that unknown, movable transactional performs the movable candidate service by being selected in workflow
Transactional determine;Based on the two preference patterns, and the system of selection based on QoS is combined, this method also proposed accordingly
Global optimum's selection algorithm.
7. method according to claim 6, it is characterised in that:This method also includes (2) global optimum selection algorithm;
In the composite services preference pattern based on affairs and QoS, the introducing of affairs is produced to the QoS time performances of composite services
Influence;Therefore, optimal candidate service is selected not consider affairs to timeliness according to local optimal time index
The influence of energy;So, this method is selected service using global optimum's method;
First, based on selected model, the transactional of workflow is it has been determined that each active transaction of i.e. workflow is known
, workflow known to transactional is referred to as Transactional Workflows (Transactional Workflow, TWF) here;Assuming that
Activity number is n, TWF={ a in Transactional Workflows TWF1,a2,…,an};To each movable a in Transactional Workflowsj, j=1,
2 ..., n, there is qjIndividual candidate's Transaction ServiceCan execution activity aj, the candidate service set expire
Foot activity ajFeature and transactional requirement;An executable thing can be obtained by being assigned to each movable candidate service
Business composite services;So, an executable candidate's combinations of transactions of set can be obtained according to all candidate services and service TCWS
(Transactional Composite Web Service), TCWS={ tcws1,tcws2,…,tcwsm, m=Π qj, this
In, j=1,2 ..., n;Then, optimal executable combinations of transactions service is selected from set TCWS using global selection algorithm;
Here, object function is Score (tcwsi), it is that each executable combinations of transactions service is counted using simple method of weighting
Calculate the QoS fractional values of whole executable combinations of transactions service;Score(tcwsi)=∑jwjqij, wherein wj∈ [0,1], ∑jwj=
1, i=1,2 ..., m, qijRepresent two QoS attributes of each candidate service;Executable combinations of transactions with minimum Score values
Service is as optimal selection, if several executable combinations of transactions services are with minimum Score values, then random selection
One performs combinations of transactions service;Global optimum's selection algorithm (Global Selection based on Transactional Workflows
Based on TWF, GST) input to include the Transactional Workflows of n activity, be output as that there is minimum Score values to hold
Capable combinations of transactions service BestTCWS;Arthmetic statement is as follows:
The first step:According to each movable candidate service set, all executable transaction sets for meeting composite services requirement are obtained
Close set of service,
Second step, according to time performance analysis algorithm, what the executable combinations of transactions of acquisition every was serviced successfully completes time, assignment
ST [i] is given, while obtaining strike price, EP [i] is assigned to, is handled according to weighting, tcws [i] Score values are obtained;
3rd step, contrasts the Score values of all executable combinations of transactions services, execution combinations of transactions clothes minimum selection Score
Business is assigned to BestTCWS;Score [i] is the Score values of executable combinations of transactions;
Algorithm GST false codes are as follows:
Input:Transactional Workflows TWF
Output:Combinations of transactions service tcws with global minima Score values
Start
Terminate
This method also provides another option algorithm;According to workflow transactions assignment method, first, Work flow model is entered
Row transactional assignment, i.e., only assign suitable transactional to workflow activities, and does not temporarily select suitable Service Instance to hold
Row activity;It is possible to obtain the Transactional Workflows TWF={ twf of a set1,twf2,…,twfm, both meet combination clothes
The Functional Requirement of business, meets transactional requirement again;To each Transactional Workflows, GST algorithms can be used to obtain the overall situation most
Excellent executable combinations of transactions service, then from these optimal executable combinations of transactions set of services, reselection one is most
Excellent combinations of transactions service meets user's request;Global optimum's selection algorithm (Global Selection based on workflow
Based on WF, GSW) input be workflow, be output as the combinations of transactions service with minimum Score values
BestTCWS;Arthmetic statement is as follows:
The first step, according to transactional assignment method, then they be assigned to set by all feasible Transactional Workflows of activity
twf;
Each Transactional Workflows in twf are selected optimal combinations of transactions service, while obtaining it by second step using algorithm GST
Score values, and be assigned to BestTcws and MiniScore respectively;Finally, select optimal from set BestTcws
Combinations of transactions services BestTCWS;TempMiniScore is the variable that a temporary combinations of transactions services Score values;
Pseudo-code of the algorithm is as follows:
Input:Workflow WF
Output:Combinations of transactions service TCWS with global minima QoS
Start
Terminate.
8. method according to claim 7, it is characterised in that:This method also includes (3) heuristic selection algorithm;In the overall situation
In optimal selection algorithm GSW, when workflow activities number becomes very big, algorithm needs the time spent to become to be difficult to connect
By;So, it would be desirable within the acceptable algorithm time, obtain the combinations of transactions service of a near-optimization to meet user
Demand;The algorithm GSW times are mainly expended at two aspects:First, transactional assignment is carried out to workflow, obtained all feasible
Transactional Workflows;Second, optimal executable combinations of transactions service is obtained by algorithm GST to each Transactional Workflows;So, from
These two aspects reduces the algorithms selection time, it is proposed that a heuristic selection algorithm;
Heuristic evaluation function is used for assessing and finding the suitable path from current node to destination node;In heuritic approach
In, heuristic evaluation function is used for assessing the Score values for including the different businesslike whole combinations of transactions services of current node;Base
In heuristic evaluation function, the transactional of current active is selected first, then, is selected from the businesslike candidate service is met
Optimal service performs current active;
In sequence and parallel schema, if the transactional of current active is entered as p or pr, then all movable affairs below
Property is all necessary for pr or cr;In parallel schema, if the transactional of current active is entered as c, then, include current active
The activity of other all branches of parallel schema is only c or cr;In selection mode, in the transactional and the pattern of branch's activity
The transactional of other branches is identical;Which reduces the time of workflow transactions assignment;Then it is businesslike determining
The QoS average values of all candidate services of activity determine businesslike transaction set as reference information, acquisition comprising current active
Successful execution time and the strike price of service are closed, and weights processing and obtains value of the Score values as heuristic evaluation function;This
Sample we just reduce the time of GST selection algorithms;
Here, heuristic evaluation function HEF is divided into three parts, be defined as follows;
Define 1:Heuristic evaluation functionAnd:
(1)G1(x) it is the information of current active;
(2)G2(x) the candidate service letter of combinations of transactions service is had to enter into for historical act and knowing according to current active transactional
Breath;
(3)G3(x) it is unknown transactional action message;
In order to obtain heuristic evaluation function, it should be understood that road determines that the success of businesslike combinations of transactions service is complete comprising current active
Into time and strike price;Here, all activities of workflow are divided into four parts by we:
Part I is the activity for having selected for suitable Transaction Service, is referred to as historical information;So, this piecemeal activity
Time and strike price are successfully completed just as G2(x) partial information in;
Part II is current active;If current active determines transactional, then meet all times of transactional requirement
The average qos value of choosing service is used as G1(x) information;
Part III is that businesslike follow-up activities are determined according to current active transactional;According to these movable transactionals,
The average qos value of all candidate services of transactional requirement is so met as G2(x) another part information;
Part IV is the unknown activity of transactional;First, all possible transactional assignment is enumerated, then identical movement owns
The average qos value of candidate service;It is used as G3(x) information included of information;
According to this four partial information, we can pass throughObtain the thing for including current active
The successful execution time in business composite services path and strike price, then weighting handle the formula valuation functions HEF that gains enlightenment;
Heuristic selection algorithm HSW (Heuristic Selection based on WF) process is:Consider current activity,
According to the above affairs feature of chosen appropriate services, the affairs feature set for meeting current active requirement can be obtained
tpset;If current active selection transactional is p or pr service, the candidate service for meeting the business requirements is G1(x) believe
Breath;If activity is the branch of selection mode, remaining branch's active transaction of the selection mode is also p or pr, remaining activity thing
Business property is pr or cr;The service of above active candidate is G2(x) information, and without G3(x) information;Current active selection transactional is c
Service when, meet the business requirements candidate service be G1(x) information;If the branch of selection mode, the selection mode
Remaining branch's active transaction is also c, if in parallel organization, remaining branch's active transaction of the parallel schema is c or cr,
The service of above active candidate is G2(x) information;The unknown activity of remaining transactional is G3(x) information;Current active selects transactional
For cr service when, meet the business requirements candidate service be G1(x) information;If the branch of selection mode, the selection
Remaining branch's active transaction of pattern is also cr, is G2(x) information;Otherwise activity below is all unknown, is G3(x) information;Than
The HEF of more each affairs feature, is that current active selects a transactional, and selected from the businesslike candidate service is met
Optimal service performs current active;In following heuritic approach, TP represents above the thing of chosen appropriate services
Business feature, TCWS represents the combinations of transactions service of selection;
Algorithm HSW false codes are as follows:
Algorithm HSW
Input:Include the workflow WF of n element
Output:Executable combinations of transactions service TCWS with global minima QoS
Start
Terminate.
9. a kind of usage right requires the service quality performance prediction meanss combined based on affairs and QoS of 1 methods described, it is special
Levy and be, the device also includes:
Extraction module, for traveling through the composite services according to the workflow sequence of composite services, is extracted in the composite services
Component service;
Processing module, for judging whether the Component service is atomic service, if so, then calculating the atomic service
Performance parameter, and assign corresponding transactional for the atomic service;Otherwise, set algorithm is called to integrate the Component service
Into atomic service, and the atomic service performance parameter is calculated, and corresponding transactional is assigned for the atomic service;
The extraction module, is additionally operable to extract next Component service in the composite services, and it is described to return to continuation judgement
Whether next Component service is atomic service, untill all component service is traversed in the composite services;
Determining module, the performance parameter of each atomic service calculated for basis and the transactional being each endowed, really
The service quality performance of the fixed composite services.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611194711.2A CN106961467B (en) | 2016-12-21 | 2016-12-21 | Service quality performance prediction method and device based on transaction and QoS combination |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611194711.2A CN106961467B (en) | 2016-12-21 | 2016-12-21 | Service quality performance prediction method and device based on transaction and QoS combination |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106961467A true CN106961467A (en) | 2017-07-18 |
CN106961467B CN106961467B (en) | 2020-06-05 |
Family
ID=59481615
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611194711.2A Expired - Fee Related CN106961467B (en) | 2016-12-21 | 2016-12-21 | Service quality performance prediction method and device based on transaction and QoS combination |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106961467B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108076143A (en) * | 2017-12-01 | 2018-05-25 | 河南理工大学 | A kind of logistics Web service dynamic optimized combination method |
CN109451037A (en) * | 2018-12-05 | 2019-03-08 | 大连海事大学 | The selection method of errors repair strategy in CWS Fault-Tolerant Problems |
CN113364621A (en) * | 2021-06-04 | 2021-09-07 | 浙江大学 | Service quality prediction method under service network environment |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103268523A (en) * | 2013-05-28 | 2013-08-28 | 北京邮电大学 | Service combination method for achieving meeting multiple performance index requirements simultaneously |
CN104320358A (en) * | 2014-09-28 | 2015-01-28 | 国家电网公司 | QoS (Quality of Service) business control method in power telecommunication net |
-
2016
- 2016-12-21 CN CN201611194711.2A patent/CN106961467B/en not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103268523A (en) * | 2013-05-28 | 2013-08-28 | 北京邮电大学 | Service combination method for achieving meeting multiple performance index requirements simultaneously |
CN104320358A (en) * | 2014-09-28 | 2015-01-28 | 国家电网公司 | QoS (Quality of Service) business control method in power telecommunication net |
Non-Patent Citations (1)
Title |
---|
JOYCE EL HADDAD,ETC.: "TQoS:Transactional and QoS-Aware Selection Algorithm for Automatic Web Service Composition", 《IEEE TRANSACTIONS ON SERVICES COMPUTING》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108076143A (en) * | 2017-12-01 | 2018-05-25 | 河南理工大学 | A kind of logistics Web service dynamic optimized combination method |
CN109451037A (en) * | 2018-12-05 | 2019-03-08 | 大连海事大学 | The selection method of errors repair strategy in CWS Fault-Tolerant Problems |
CN109451037B (en) * | 2018-12-05 | 2021-06-08 | 大连海事大学 | Method for selecting error repair strategy in CWS fault tolerance problem |
CN113364621A (en) * | 2021-06-04 | 2021-09-07 | 浙江大学 | Service quality prediction method under service network environment |
CN113364621B (en) * | 2021-06-04 | 2022-07-26 | 浙江大学 | Service quality prediction method under service network environment |
Also Published As
Publication number | Publication date |
---|---|
CN106961467B (en) | 2020-06-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Satyal et al. | Business process improvement with the AB-BPM methodology | |
López-Ibáñez et al. | Beam-ACO for the travelling salesman problem with time windows | |
Law et al. | Economic globalization and financial development in East Asia: a panel cointegration and causality analysis | |
Ferreira et al. | Using logical decision trees to discover the cause of process delays from event logs | |
CN106961467A (en) | Based on affairs and QoS the service quality performance Forecasting Methodology combined and device | |
Landa-Torres et al. | Evaluating the internationalization success of companies through a hybrid grouping harmony search—extreme learning machine approach | |
Bolsinger et al. | Process improvement through economically driven routing of instances | |
CN108647960A (en) | A kind of donation fund management method and system based on block chain technology | |
CN111444213B (en) | Ledger clearing system and method based on credit business | |
CN106790536B (en) | Combined Web service selection method based on transaction and QoS | |
KR102153533B1 (en) | Customized value suggestion system for Enterprise | |
CN110245985B (en) | Information processing method and device | |
Ashraf et al. | Empirical analysis of foreign direct investment and economic growth in Pakistan using VECM model | |
Hinkelmann et al. | Combining process modelling and case modeling | |
CN102880784A (en) | Method for calculating creditability of composite services based on data dependence | |
Corradini et al. | BProVe: tool support for business process verification | |
Ding et al. | Performance evaluation of transactional composite web services | |
Nowak et al. | An application of the multiple criteria decision tree in project planning | |
CN113159935B (en) | To-be-handled business processing method and device based on block chain | |
Hakro et al. | Foreign direct investment, determinants and policy analysis: case study of Pakistan | |
Falk | Evaluation of a pattern-based approach for business process improvement | |
Leal et al. | Exploiting mde for platform-independent testing of service orchestrations | |
Duan et al. | Towards value-driven business modelling based on service brokerage | |
Thiagarasu et al. | Multi-agent coordination in project scheduling: priority rules based resource allocation | |
Xu et al. | Optimized composite service transactions through execution results prediction |
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 | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200605 Termination date: 20201221 |
|
CF01 | Termination of patent right due to non-payment of annual fee |