CN102739469A - Web service response time predicting method based on time sequence - Google Patents

Abstract

The invention discloses a web service response time predicting method based on increment steady-state analysis, belonging to the field of software performance prediction. The method comprises the following steps: obtaining a sequence of a series of measurement values of network service response time according to a fixed unit time interval; subsequently defining and calculating observation window response time increment types; calculating a transition probability matrix among the types; then calculating a steady-state distribution probability of the response time increment based on the transition probability matrix; and finally adding a calculated increment expected value to a response time measurement value at the t moment so as to obtain the prediction value of the response time at the t+1 moment. With the adoption of the method, the defects that traditional Web prediction mold and method depend too much on SLA (Service-Level-Agreement) specifications and hypothesis parameters are constant and unchanged are avoided; and mold support and analysis means are provided for WEB service credibility research.

Description

A kind of based on seasonal effect in time series Web service response time Forecasting Methodology

Technical field

The invention belongs to the field of software performance prediction, especially relate to a kind of based on seasonal effect in time series Web service response time Forecasting Methodology.

Background technology

More and more important role is being played the part of in the field in that ecommerce, enterprise application be integrated etc. as a kind of new distributed component model in Web service; It is considered to one of technology most crucial in the service-oriented computing framework (SOA); Through the Web service combination technique, single Web service combined becomes complete application.Yet Web service operates in the network environment of dynamic change; Its response time value is also in frequent variations; The network environment of this dynamic change will influence the run time behaviour of Web service, and then causes the variation of whole Web service composite behaviour, therefore; To web service carrying out performance prediction is the key technology that guarantees based on the application system quality of service, also is to optimize service combination important method.Most important two indexs are response time and throughput in the Web service performance, and the former representes that system accomplishes the time of given required by task, and the latter representes that system accomplishes the number of times of given task in the unit interval.

Though academia and industrial quarters have proposed a series of model and methods that are used to analyze and predict the Web service performance; But most of method is many with based on SLA (Service-level-agreement; Service-level agreement) static analysis is main means; Service-level agreement is about a contract between Internet service provider and client, has wherein defined terms such as COS, service quality and client's payment, and it has retrained the minimum of serving the serviced component that the provider provides in quantity and has connected bandwidth, maximum packet loss, linkage fault rate and equal performance-relevant parameter; Static properties analysis based on SLA; The direct exactly limit value with the above-mentioned parameter that service-level agreement was retrained is the model input, and the hypothesis performance parameter is invariable, analyzes and predict the actual performance of Web service assembly; There is very big defective in the method for this static analysis: the network and the system environments that are relied on during owing to the Web service operation are fast changing; Various parameter amounts that influence performance such as bandwidth, packet loss, linkage fault rate, message field length etc. can not be kept invariablely, and there are great deviation in this and the invariable hypothesis of performance parameter; Because the parameter that provides in the service-level agreement is the upper and lower bound value, corresponding value possibly be higher or lower than limit value during the Web service operation, thereby causes having over-evaluated or underestimated actual performance based on the performance prediction method of service-level agreement.

In order to remedy the deficiency of said method; Take into full account the dynamic of Web service performance under the true environment; The present invention analyzes on the basis of its trend variation in further investigation performance test value time series data, proposes a kind of based on seasonal effect in time series Web service response time Forecasting Methodology.

Summary of the invention

Because the above-mentioned defective of prior art, technical problem to be solved by this invention provides a kind of response time Forecasting Methodology of Web service more accurately.

Be to realize above-mentioned purpose, the invention provides a kind ofly, carry out according to the following steps based on seasonal effect in time series Web service response time Forecasting Methodology:

Step 1: the response time data sequence of obtaining the web service according to fixing unit interval;

Setting the response time data sequence has the surveying record of t response time, and said response time data sequence is r _t(i), { r _t(i) | 1≤i≤t, 1≤t＜∞ }; The response time data sequence of the said web of obtaining service is obtained by the url address test of SOAP UI test platform to the web service;

Step 2: the average increment that calculates the continuous response time;

It is one group that k continuous response time compiled, and k is a positive integer, and whole response time sequence is divided into

Individual group, be regarded as one with every group and investigate window; X y the response time of investigating in the window is r _t((x-1) * k+y); Setting x the response time average increment of investigating in the window is AINC (x);

$\mathrm{AINC}\left(x\right)=\frac{\underset{s=1}{\overset{k-1}{\mathrm{\Σ}}}{r}_t((x-1)\×k+s+1)-r_t((x-1)\×k+s)}{k},$

Step 3: transition probability between definition response time average increment class and calculated response time average increment class;

The value of minimum response time average increment is MIN in

individual response time average increment,

The value of maximum response time average increment is MAX in

individual response time average increment,

MIN is divided into p type to the interval of MX, p is a positive integer; Setting x the response time average increment of investigating in the window is map (x) to l mapping function of classifying, 1≤l≤p; And if only if $\mathrm{MIN}+\frac{\mathrm{MAX}-\mathrm{MIN}}{p}\&times;(l-1)\&le;\mathrm{AINC}\left(i\right)<\mathrm{MIN}+\frac{\mathrm{MAX}-\mathrm{MIN}}{p}\&times;l$ The time, map (x)=l;

Set MIN matrix of transition probabilities between m type and n the class in the interval of MAX and be TRAN (m, n), m, n are positive integer;

$\mathrm{ICOUNT}(m,l)=\left\{\begin{array}{cc}1& \mathrm{ifmap}\left(l\right)=\mathrm{<figure-callout\; id="0"\; label="m"\; filenames="HDA00001691607000031.png"\; state="\{\{state\}\}">m</figure-callout>}\\ 0& \mathrm{else}\end{array}\right.;$

$\mathrm{IJCOUNT}(m,n,l)=\left\{\begin{array}{cc}1& \mathrm{ifmap}(l+1)=\mathrm{nandmap}\left(l\right)=m\\ 0& \mathrm{else}\end{array}\right.;$

Step 4: under the computing system balance, the response time average increment is in the probability of each response time average increment class;

π is under the system balancing, and the response time average increment is in the probability vector of h response time average increment class; H the component of π is π (h), 1≤h≤p; π (h)=π (h) * TRAN (m, n);

Step 5: prediction t+1 response time constantly;

The desired value of setting the response time average increment under the poised state situation is EAINC;

$\mathrm{EAINC}=\underset{h=1}{\overset{p}{\mathrm{\&Sigma;}}}\mathrm{\&pi;}\left(h\right)\&times;(\mathrm{MIN}+\frac{(\mathrm{MAX}-\mathrm{MIN})\&times;(h-1)}{p}+\frac{\mathrm{MAX}-\mathrm{MIN}}{2p});$ Setting t+1 response time predicted value constantly is PRT; Calculate $\mathrm{EAINC}=\frac{\underset{w=1}{\overset{k-2}{\mathrm{\&Sigma;}}}{r}_t(t+1-w)-r_t(t-w)+(\mathrm{PRT}-r_t\left(t\right))}{k},$ Obtain the response time value that PRT needing just can obtain prediction accurately.

The invention has the beneficial effects as follows: the present invention is in further investigation performance test value time series data; Analyze on the basis that its trend changes and realize; SLA standard and the invariable deficiency of hypothesis performance parameter have been avoided too relying in traditional web performance prediction model and the method; Can support and analysis means for the credible research of web service supplies a model, can greatly improve predictablity rate and precision.

Description of drawings

Fig. 1 is a schematic flow sheet of the present invention.

Fig. 2 is a response time measured data sketch map.

Fig. 3 is the comparison sketch map of response time predicted value and actual value.

Embodiment

Below in conjunction with accompanying drawing and embodiment the present invention is described further:

As shown in Figure 1, a kind of based on seasonal effect in time series Web service response time Forecasting Methodology, carry out according to the following steps:

Step 1: the response time data sequence of obtaining the web service according to fixing unit interval;

Setting the response time data sequence has the surveying record of t response time, and said response time data sequence is r _t(i), { r _t(i) | 1≤i≤t, 1≤t＜∞ }, the response time data sequence of the said web of obtaining service is obtained by the url address test of SOAP UI test platform to the web service.

Step 2: the average increment that calculates the continuous response time;

It is one group that k continuous response time compiled, and k is a positive integer, and whole response time sequence is divided into

Individual group, be regarded as one with every group and investigate window; X y the response time of investigating in the window is r _t((x-1) * k+y); Setting x the response time average increment of investigating in the window is AINC (x); $\mathrm{AINC}\left(x\right)=\frac{\underset{s=1}{\overset{k-1}{\mathrm{\&Sigma;}}}{r}_t((x-1)\&times;k+s+1)-r_t((x-1)\&times;k+s)}{k},$

Step 3: transition probability between definition response time average increment class and calculated response time average increment class;

The value of minimum response time average increment is MIN in individual response time average increment,

The value of maximum response time average increment is MAX in

individual response time average increment,

MIN is divided into p type to the interval of MAX, p is a positive integer; Setting x the response time average increment of investigating in the window is map (x) to l mapping function of classifying, 1≤l≤p; And if only if $\mathrm{MIN}+\frac{\mathrm{MAX}-\mathrm{MIN}}{p}\&times;(l-1)\&le;\mathrm{AINC}\left(i\right)<\mathrm{MIN}+\frac{\mathrm{MAX}-\mathrm{MIN}}{p}\&times;l$ The time, map (x)=l;

Set MIN matrix of transition probabilities between m type and n the class in the interval of MAX and be TRAN (m, n), m, n are positive integer;

$\mathrm{ICOUNT}(m,l)=\left\{\begin{array}{cc}1& \mathrm{ifmap}\left(l\right)=\mathrm{<figure-callout\; id="0"\; label="m"\; filenames="HDA00001691607000031.png"\; state="\{\{state\}\}">m</figure-callout>}\\ 0& \mathrm{else}\end{array}\right.;$

$\mathrm{IJCOUNT}(m,n,l)=\left\{\begin{array}{cc}1& \mathrm{ifmap}(l+1)=\mathrm{nandmap}\left(l\right)=m\\ 0& \mathrm{else}\end{array}\right..$

Step 4: under the computing system balance, the response time average increment is in the probability of each response time average increment class;

π is under the system balancing, and the response time average increment is in the probability vector of h response time average increment class; H the component of π is π (h), 1≤h≤p; π (h)=π (h) * TRAN (m, n), concrete

$\mathrm{\&pi;}\left(1\right)=\mathrm{\&pi;}\left(1\right)\&times;\left[\begin{array}{cccc}\mathrm{TRAN}\left(\mathrm{1,1}\right)& \mathrm{TRAN}\left(\mathrm{1,2}\right)& ....& \mathrm{TRAN}(p,1)\\ \mathrm{TRAN}(2,1)& \mathrm{TRAN}\left(\mathrm{2,2}\right)& ....& \mathrm{TRAN}(p,2)\\ \&CenterDot;& \&CenterDot;& & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& \mathrm{TRAN}(p-1,p-1)& \&CenterDot;\\ \mathrm{TRAN}(p,1)& \mathrm{TRAN}(p,2)& ....& \mathrm{TRAN}(p,p)\end{array}\right],$

$\mathrm{\&pi;}\left(2\right)=\mathrm{\&pi;}\left(2\right)\&times;\left[\begin{array}{cccc}\mathrm{TRAN}\left(\mathrm{1,1}\right)& \mathrm{TRAN}\left(\mathrm{1,2}\right)& ....& \mathrm{TRAN}(p,1)\\ \mathrm{TRAN}(2,1)& \mathrm{TRAN}\left(\mathrm{2,2}\right)& ....& \mathrm{TRAN}(p,2)\\ \&CenterDot;& \&CenterDot;& & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& \mathrm{TRAN}(p-1,p-1)& \&CenterDot;\\ \mathrm{TRAN}(p,1)& \mathrm{TRAN}(p,2)& ....& \mathrm{TRAN}(p,p)\end{array}\right],......,$

$\mathrm{\&pi;}\left(p\right)=\mathrm{\&pi;}\left(p\right)\&times;\left[\begin{array}{cccc}\mathrm{TRAN}\left(\mathrm{1,1}\right)& \mathrm{TRAN}\left(\mathrm{1,2}\right)& ....& \mathrm{TRAN}(p,1)\\ \mathrm{TRAN}(2,1)& \mathrm{TRAN}\left(\mathrm{2,2}\right)& ....& \mathrm{TRAN}(p,2)\\ \&CenterDot;& \&CenterDot;& & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& \mathrm{TRAN}(p-1,p-1)& \&CenterDot;\\ \mathrm{TRAN}(p,1)& \mathrm{TRAN}(p,2)& ....& \mathrm{TRAN}(p,p)\end{array}\right].$

Step 5: prediction t+1 response time constantly;

The desired value of setting the response time average increment under the poised state situation is EAINC;

$\mathrm{EAINC}=\underset{h=1}{\overset{p}{\mathrm{\&Sigma;}}}\mathrm{\&pi;}\left(h\right)\&times;(\mathrm{MIN}+\frac{(\mathrm{MAX}-\mathrm{MIN})\&times;(h-1)}{p}+\frac{\mathrm{MAX}-\mathrm{MIN}}{2p});$ Setting t+1 response time predicted value constantly is PRT; $\mathrm{EAINC}=\frac{\underset{w=1}{\overset{k-2}{\mathrm{\&Sigma;}}}{r}_t(t+1-w)-r_t(t-w)+(\mathrm{PRT}-r_t\left(t\right))}{k},$ Can be through calculating

$\underset{i=1}{\overset{p}{\mathrm{\&Sigma;}}}\mathrm{\&pi;}\left(i\right)\&times;(\mathrm{MIN}+\frac{(\mathrm{MAX}-\mathrm{MIN})\&times;(i-1)}{p}+\frac{\mathrm{MAX}-\mathrm{MIN}}{2p})=\frac{\underset{w=1}{\overset{k-2}{\mathrm{\&Sigma;}}}{r}_t(t+1-w)-r_t(t-w)+(\mathrm{PRT}-r_t\left(t\right))}{k}$

Draw and obtain PRT, extrapolate t+1 response time predicted value constantly thus.

Like Fig. 1, shown in Figure 2; For correctness and accuracy to method are verified; Be configured to the INTEL i5-760 processor of 2.8G, in save as on the PC of 4G operation and adopt the SOAPUI test platform to provide the instance of the Web service of Weather information to test one; The URL address of said web Service Instance is http://www.webservicex.net/globalweather.asmx WSDL; The time started of test is 10: 30: 0 morning of on October 17th, 2011, and test interval is 250 milliseconds, obtains 128 response time test values continuously; Wherein preceding 100 data are used for modelling, and remaining 28 response time test values that recorded in 30 minutes 25 seconds since at 10 o'clock in the morning on October 17th, 2011 are used for modelling verification.

Model parameter is set to: k=5; Classification quantity p=8, according to the method that the present invention provides, 10: 30: 25 morning of on October 17th, 2011, later response time predicted value was illustrated among Fig. 3 with the result of other three model predication values; The time series forecasting value of lines 1 for adopting method of the present invention to obtain among Fig. 3; The time series forecasting value of lines 2 for adopting the ARMA time series models to obtain, the time series forecasting value that lines 3 obtain for the ARMA-O model that adopts the elimination outlier, the time series forecasting value of lines 4 for adopting two quantitative prediction model DQ to obtain; Lines 5 are actual value; Time series forecasting value that the inventive method obtains and actual value average error rate are 24.4%, and the time series forecasting value and the actual value average error rate that adopt the ARMA time series models to obtain are 30.1%, and time series forecasting value and actual value mean error that the ARMA-O model of employing elimination outlier obtains are 33.6%; The time series forecasting value that the two quantitative prediction model DQ of employing obtain and the error of actual value are 42.8%; It is thus clear that the method that adopts the present invention to propose has obtained better precision in the Web service performance prediction.

More than describe preferred embodiment of the present invention in detail.Should be appreciated that those of ordinary skill in the art need not creative work and just can design according to the present invention make many modifications and variation.Therefore, all technical staff in the art all should be in the determined protection range by claims under this invention's idea on the basis of existing technology through the available technical scheme of logical analysis, reasoning, or a limited experiment.

Claims (2)

1. Web service response time Forecasting Methodology based on the increment steady-state analysis is characterized in that carrying out according to the following steps:

Step 1: the response time data sequence of obtaining the web service according to fixing unit interval;

Setting the response time data sequence has the surveying record of t response time, and said response time data sequence is r _t(i); { r _t(i) | 1≤i≤t, 1≤t＜∞ };

Step 2: the average increment that calculates the continuous response time;

It is one group that k continuous response time compiled, and k is a positive integer, and whole response time sequence is divided into Individual group, be regarded as one with every group and investigate window; X y the response time of investigating in the window is r _t((x-1) * k+y); Setting x the response time average increment of investigating in the window is AINC (x);

$\mathrm{AINC}\left(x\right)=\frac{\underset{s=1}{\overset{k-1}{\mathrm{\&Sigma;}}}{r}_t((x-1)\&times;k+s+1)-r_t((x-1)\&times;k+s)}{k},$

Step 3: transition probability between definition response time average increment class and calculated response time average increment class;

The value of minimum response time average increment is MIN in individual response time average increment,

The value of maximum response time average increment is MAX in

individual response time average increment,

MIN is divided into p type to the interval of MAX, p is a positive integer; Setting x the response time average increment of investigating in the window is map (x) to l mapping function of classifying, 1≤l≤p; And if only if $\mathrm{MIN}+\frac{\mathrm{MAX}-\mathrm{MIN}}{p}\&times;(l-1)\&le;\mathrm{AINC}\left(i\right)<\mathrm{MIN}+\frac{\mathrm{MAX}-\mathrm{MIN}}{p}\&times;l$ The time, map (x)=l;

Set MIN matrix of transition probabilities between m type and n the class in the interval of MAX and be TRAN (m, n), m, n are positive integer;

$\mathrm{ICOUNT}(m,l)=\left\{\begin{array}{cc}1& \mathrm{ifmap}\left(l\right)=m\\ 0& \mathrm{else}\end{array}\right.;$

$\mathrm{IJCOUNT}(m,n,l)=\left\{\begin{array}{cc}1& \mathrm{ifmap}(l+1)=\mathrm{nandmap}\left(l\right)=m\\ 0& \mathrm{else}\end{array}\right.;$

Step 4: under the computing system balance, the response time average increment is in the probability of each response time average increment class;

π is under the system balancing, and the response time average increment is in the probability vector of h response time average increment class; H the component of π is π (h), 1≤h≤p; π (h)=π (h) * TRAN (m, n);

Step 5: prediction t+1 response time constantly;

The desired value of setting the response time average increment under the poised state situation is EAINC;

$\mathrm{EAINC}=\underset{h=1}{\overset{p}{\mathrm{\&Sigma;}}}\mathrm{\&pi;}\left(h\right)\&times;(\mathrm{MIN}+\frac{(\mathrm{MAX}-\mathrm{MIN})\&times;(h-1)}{p}+\frac{\mathrm{MAX}-\mathrm{MIN}}{2p});$ Setting t+1 response time predicted value constantly is PRT; Calculate $\mathrm{EAINC}=\frac{\underset{w=1}{\overset{k-2}{\mathrm{\&Sigma;}}}{r}_t(t+1-w)-r_t(t-w)+(\mathrm{PRT}-r_t\left(t\right))}{k},$ Obtain PRT.

2. a kind of Web service response time Forecasting Methodology based on the increment steady-state analysis as claimed in claim 1 is characterized in that: the response time data sequence of the said web of obtaining service is obtained by the url address test of SOAP UI test platform to the web service.

Images