[雙語(yǔ)翻譯]計(jì)算機(jī)外文翻譯--開發(fā)人員視角下的java web服務(wù)性能評(píng)估（英文）

1、Communications and Network, 2010, 2, 200-206 doi:10.4236/cn.2010.23029 Published Online May 2010 (http://www.SciRP.org/journal/cn) Copyright © 2010 SciRes.

2、 CN Performance Evaluation of Java Web Services: A Developer’s Perspective Sanjay Ahuja, Je-Loon Yang School of Computing, University of North Florida, Jacksonville, Florida, USA E-mail: {sahuj

3、a, n00168553}@unf.edu Received March 7, 2010; revised July 2, 2010; accepted July 3, 2010 Abstract As the population of the Internet grows rapidly the development of web technologies becomes extremely important. For th

4、e evolvement of Web 2.0, web services are essential. Web services are programs that allow computers of different platforms on the web to interactively communicate without the need of extra data for human reading interf

5、aces and formats such as web page structures. Since web service is a future trend for the growth of internet, the tools that are used for development is also important. Although there are many choices of web service fr

6、ameworks to choose from, developers should choose the framework that fits best to their application based on performance, time and effort for the framework. In this project, we chose four common frameworks to compare t

7、hem in both qualitative and quantitative metrics. After running the tests, the results are statistically analyzed by SAS. Keywords: Web Service, Framework, Performance, Java, Developer 1. Introduction For going on trips

8、 to other states or countries, the person usually requires to buy airplane tickets, rent a car, and make reservations for hotels to stay at. When dealing with airplane tickets, most of the time the person even has to

9、 buy several tickets for some stops instead of a ticket that takes the person directly to the final destina- tion. Looking up the airplane arrival and departure times to connect each flight might take hours of searchin

10、g and planning. What if there was a virtual agent that could do this all in just a few seconds? So usually people would look for agents to do this for them. But what if this agent was actually is actually a virtual a

11、gent online. If the per- son just enters the location he wants to start from, the destination, the desired time for departure or arrival, and all the information required into the computer, in a snap, the computer sho

12、ws all the results for the person to choose from and purchase the tickets. Even better, such virtual agents could have possibilities to also show information of the car rentals and hotels at the destination and reser

13、ve them for you. By using this type of virtual agent, could save much effort and time and can also be more accurate than human agents. Such technology relies on the development and widespread of web services. Instea

14、d of developing a web service application from scratch, there are many open source frameworks that make development much easier. Which of these frame- works would be a better choice for web service application develop

15、ment? This study compares four popular open source frameworks both qualitatively and quantitatively by doing several tests and analysis. The four frameworks are Apache Axis, JBossWS, XFire, and Hessian. More introduc

16、tions of web services are done in Section 2. Section 3 describes the four frameworks that are used in this study. In Section 4, the metrics that are used to measure the frameworks are explained in more detail. Sectio

17、n 5 introduces the statistical analysis methods that are use to analyze the measured results. In Section 6, the test results are shown and analyzed. The conclusions are in Section 7. 2. Web Service Frameworks Since we

18、b services are designed to transfer data in com- mon ways, several companies and groups developed web service frameworks for the convenience of web service developers so that they do not need to write a complete web s

19、ervice from scratch. Some of the popular frame- works are Apache Axis, JBossWS, Codehaus XFire, and Caucho Hessian. In this section, these frameworks are going to be introduced. S. AHUJA ET AL. 202 wide variety of m

20、emory, such as cache memory, flash memory, random access memory (RAM), virtual memory, etc. Either memory, they are all limited in server due to the cost and space. A framework that uses less memory would have an adv

21、antage of allowing higher capacity for the server. 3.4. CPU Usage Central Processing Unit (CPU), also known as processor, is a component in a computer used to interpret program instructions and process data. Although

22、it is only able to process one task at a time, when there are multiple tasks to be done, instead of finishing a task then going to an- other, the CPU is designed to switch to other tasks before finishing one if necess

23、ary, so that it would act like if it is executing multiple tasks at the same time. However, large tasks might occupy a lot of CPU time, which de- creases the time scheduled for other tasks. A framework that uses less

24、CPU would allow the server to have more time to execute other tasks. 3.5. Source Lines of Code The source lines of code (SLOC) used in a framework can indicate the transparency and abstraction of the framework. The ma

25、in goal of a framework is to save the developer’s time and effort by not having to write the entire code from scratch. Thus the less lines of code that is required for a framework, the more time and effort is saved b

26、y this framework. However, lines of code cannot be exactly accurate since some lines might be long while some lines are short. So the number of files and size of files also is a consideration. 4. Statistical Analysis

27、Methods After retrieving the test data to compare the performances, we need a method to analyze the results. By simply calculating the average response times and making them into a graphs is not sufficient for the ana

28、lysis. Looking at the average response times 1.5 seconds and 1.6 seconds, we can not be sure if that is a great difference or not. Therefore, statistical analysis methods are required to tell if the difference is sig

29、nificant or not. In this project, the general linear model (GLM) [9] and two-way analysis of variance (two-way ANOVA) is used for statistical analysis. Furthermore, the Statistical Analysis System (SAS) [10] is used

30、as a tool for aiding the calculations of the statistical analyses required. 4.1. The SAS System The SAS system is statistical analysis software that has a wide variety of statistical modules and procedures. They use a

31、fourth-generation programming language (4GL) for their code and the programs are composed by three main components – the data step, the procedure step, and the macro language. The data step is for entering data, like

32、 inserting the data in the code or read from data files. The procedure step is the use of the statistical methods and models to analysis the data that was read in the data step. The macro language is for decreasing th

33、e redundancy of functions that are used again and again throughout the program. 4.2. The GLM Model The GLM model is a statistical linear model that is used in general cases. It is the foundation of many statistical a

34、nalyses, such as t-test, ANOVA, Analysis of Covariance (ANCOVA), etc. The easiest case to understand how the GLM model works is the two-variable case. The goal of this analysis is to use a way to accurately describe t

35、he information in this plot. Using the GLM model, we try to find a straight line that is closest to all the dots in the plot. This line would be written as: y = b0 + b1x + e, where y is the y-axis variable, x is the

36、x-axis variable, b0 is the intercept (the value of y when x equals 0), b1 is the slope of the straight line, and e is the error. By solving b0 and b1, we can get information about this linear line that describes the

37、dots in the plot. In other cases with more than two variables, the formula can be extended as: y = b0 + b1x1 + b2x2 + b3x3 + ... + bnxn + e, where n is the number of variables of the situation. But the mechanism for

38、solving such problems is the same. 5. Results and Analyses In order to get best results from SAS analysis, each case is tested twenty times. Since four frameworks with five different amounts of clients are tested, ther

39、e are twenty different cases. Adding the twenty test times for the twenty different cases would result 400 data sets to be calculated by SAS. Besides the amount of clients, the size of data is also considered. With a

40、lso five different sizes of data sent, there would be twenty cases with a total of 400 data sets. The response time is measured by recording the time right before invoking the web service and recording the time right

41、 after the data requested is received then subtracting the time difference. 5.1. Results 5.1.1. Client Scenarios For testing the four different frameworks in different scenarios, web service applications to send out d

42、ata are created. The five scenarios based on amount of clients for testing performance of the four frameworks are 1 client, 5 clients, 10 clients, 15 clients, and 20 clients each retrieving 1 MB of data. The average