1414 A Community Web Site Initiative Daniel, E., & Wilson, H. (2002). Adoption inten- WLRQVDQGEHQH¿WVUHDOL]HG$VWXG\RIHFRPPHUFH in UK SMEs. Journal of Small Business and Enterprise Development, 9(4), 331-348. Daniel, E., Wilson, H., & Myers, A. (2002). Adoption of e-commerce by SMEs in the UK: Towards a stage model. International Small Busi- ness Journal, 20(3), 253-270. DTI (2000). International bench marking survey for ICT use, UK. Department of Trade and In- dustry. Retrieved March 2005, from http://www. ukonline.gov.uk Brown, D. H., & Lockett, D. (2004). Potential of critical e-applications for engaging SMEs in e-business: A provider perspective. European Journal of Information Systems, 13, 21-34. Fillis, I., Johannson, U., & Wagner, B. (2004). Factors impacting on e-business adoption and GHYHORSPHQWLQWKHVPDOOHU¿UPInternational Journal of Entrepreneurial Behaviour and Re- search, 10(3), 79-88. Grandon, E. E., & Pearson, J. M. (2004). Elec- tronic commerce adoption: An empirical study of small and medium US businesses. Information and Management, 42, 197-216. Hamill, J. (1997). The Internet and international marketing. International Marketing Review, 14(5), 300-323. Herbig, P., & Hale, B. (1997). Internet: The mar- keting challenge of the 20 th century. Internet Research: Electronic Networking Applications and Policy, 7(2), 95-100. Kendall, J. D., Tung, L. L. Chua, K. H., Ng, C. H. D., & Tan, S. M. (2001) Receptivity of Singapore’s SMEs to electronic commerce adoption. Strategic Information Systems, 10, 223-242. Kiani, G. R. (1998). Marketing opportunities in the digital world. Internet Research: Electronic Networking Applications and Policy, 8(2),185- 194. 0DUWLQ/0DWOD\+³%ODQNHW´DS- SURDFKHVWRSURPRWLQJ,&7LQVPDOO¿UPV6RPH lessons from the DTI ladder adoption model in the UK. Internet Research: Electronic Networking, Applications and Policy, 11(5), 399-410. Mehrtens, J., Cragg, P. B., & Mills, A. (2001). A model of Internet adoption by SMEs. Information and Management, 39, 165-176. Poon, S., & Swatman, P. (1997). Internet-based small business communication. International Journal of Electronic Markets, 7(2), 15-21. Poon, S., & Swatman, P. (1999). An exploratory study of small business Internet commerce issues. Information and Management, 35(1), 9-18. Quayle, M. (2002). E-commerce: The challenge for UK SMEs in the 21 st century. International Journal of Operations and Production Manage- ment, 22(10), 1148-1161. Quelch, J. A., & Klein, L. R. (1996). The Internet and international marketing. Sloan Management Review, Spring, 60-75. Rao, S. S., Metts, G., & Monge, C. A. M. (2003). Electronic commerce development in small and medium sized enterprises: A stage model and its implications. Business Process Management Journal, 9(1), 11-32. Scupola, A. (2003). The adoption of Internet commerce by SMEs in the south of Italy: An environmental, technological and organizational perspective. Journal of Global Information Tech- nology Management, 6(1), 52-72. Storey, D. J. (1994). Understanding the small business sector. London: International Thomson Business Press. Teo, T. S. H., & Ranganathan, C. (2004). Adopt- ers and non-adopters of business-to-business 1415 A Community Web Site Initiative electronic commerce in Singapore. Information and Management, 42, 89-102. This work was previously published in Global Electronic Business Research; Opportunities and Directions, edited by N. Al Qirim, pp. 114-132, copyright 2006 by IGI Publishing (an imprint of IGI Global). Yin, R. K. (1994). Case study research: Design and methods. Thousand Oaks, CA: Sage. 1416 Copyright © 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited. Chapter 5.2 Testing, Measuring, and Diagnosing Web Sites from the User’s Perspective Leszek Borzemski Wroclaw University of Technology, Poland ABSTRACT Users perceive good Internet performance as characterized by low latency, high throughput and high availability. When browsing the Web, users are concerned with the performance of entire pages. Understanding and identifying the sources of the performance problems are very important issues, especially for e-business. Therefore, there is the need to have a service for testing and mea- suring e-business Web site performance from the perspective of the end-users. We present our contribution in this area, that is, the Wing free service that has been developed for the purpose of Web transaction visualization. Our Web client that probes a target Web site is a real Web browser (MS IE), so the user can observe how a particular browser uses the network. Such known tools use their own Web browsing methods. Therefore, the solutions can be different from that used by real browsers, and the results can be inadequate. :LQJKHOSVLGHQWLI\LQHI¿FLHQWQHWZRUNXVDJHE\ the browser and helps to tune Web pages to use WKHQHWZRUNHI¿FLHQWO\7KHUHIRUH:LQJFDQEH a good analysis tool for Web page and network application developers. INTRODUCTION In the past few years, the World Wide Web has grown from a speculative medium to a robust telecommunication infrastructure that handles several mission-critical business as well as re- VHDUFKFRPSXWLQJWUDI¿F:HEXVHUVSHUFHLYH good Internet performance as low latency, high throughput and high availability. Web quality of VHUYLFHLVH[WUHPHO\GLI¿FXOWWRVWXG\LQDQLQWH- grated way. It has never been easy to determine whether slow responses are due to network prob- lems or end-systems problems on both sides, that is, user and server sides. Moreover, because most of these performance problems are transient and very complex in the relationships between dif- 1417 Testing, Measuring, and Diagnosing Web Sites from the User’s Perspective IHUHQWIDFWRUVWKDWPD\LQÀXHQFHHDFKRWKHUZH therefore, cannot precisely diagnose and isolate the issue key sources. Understanding and identifying the sources of the performance problems are very important issues for Internet designers. There is the need to have a service for testing and measuring e-business Web site performance from the perspective of the end-users. This paper surveys and compares such services that are avail- able on the Internet. Almost all are commercial. We may also obtain free access to their functions however limited in time or functionality. Here, we also present our contribution in this area, that is, the Wing free Internet service that has been developed for the purpose of Web probing, visualization and performance analysis from the user perspective. Wing can be used in instant and periodic measurements of Web sites, including e-commerce solutions. The paper is organized as follows. The next section discusses related work and gives a back- ground. The third section presents the Wing ser- vice. Different tools are compared and discussed to show how Wing relates to other developments. The fourth section gives an illustrative example RILQVWDQWPHDVXUHPHQWRID:HESDJH7KH¿IWK and sixth sections show the application of the periodic measurements performed by Wing to WKURXJKSXW HVWLPDWLRQ DQG :HE SDJH ³PRUWDO- ity” evaluation studies. Finally, the last section concludes the paper. RELATED WORK AND BACKGROUND Network protocol visualization tools have been developed since the very beginning of computer networks. They are considered as tools for better understanding of computer network. They can help network administrators or end users in analysis of network reliability and performance. They can be general tools used for several network protocols RUGHYHORSHGVSHFL¿FDOO\WRDVVLVWVRPHFKRVHQ protocol or a suite of protocols. Due to the rapid development of the Internet, there is the need to have such tools for the TCP/IP protocol suite. Nowadays, the most important visualization challenge is HTTP protocol and especially Web page downloading. Internet uses IP network protocol, and all of the information is carried in packets. The data transfers are organized under control of TCP transport protocol that provides end-to-end reli- able connectivity. Web clients communicate with Web servers using HTTP protocol. The end users perceive good Internet perfor- mance through Web page downloading. Therefore both end users and Web site administrators are anxious for the knowledge found in how well the Web pages are downloaded. To study Web page download time, we can consider the following components, as illustrated in Figure 1: • DNS • DNS2SYN • CONNECT • ACK2GET • FIRST_BYTE • LEFT_BYTES ,QWKH¿UVWSKDVHRI:HEWUDQVDFWLRQWKHUHLV the need to determine the IP address based on the symbolic server name. DNS is the time to translate the server name into the IP address. Sometimes, this operation may take much time, therefore we need to include this component in the measure- ments of Web page download. Next is the time period to open the TCP connection by the browser. The client downloading the Web page opens the TCP connection using an exchange of SYN packet that initiates the three-way handshake. DNS2SYN is the time spent by the client between DNS reso- lution event and this SYN packet exchange. The elapsed time between transmitting the SYN to the server and receiving the SYN response is the CONNECT time. The connection phase begins 1418 Testing, Measuring, and Diagnosing Web Sites from the User’s Perspective when a client initiates connection request SYN, and ends when the connection is established, that is, when the server receives ACK packet from the client. Then the client starts the browsing phase. There can be another delay in browser action, namely ACK2GET time, after which the browser sends GET request for Web page components. The GET request can be for a base page (HTML) or an object embedded in a page. The FIRST_BYTE is the time between the sending of GET request and WKHUHFHSWLRQRIWKH¿UVWSDFNHWLQFOXGLQJUHTXHVWHG component. The LEFT_BYTES is the time spent for downloading the rest of the requested object. &RQWUDU\WRRWKHUGHYHORSPHQWVLQWKH¿HOGRI:HE page downloading and visualization, we propose to take into account DNS2SYN and ACK2GET time components. The Web transaction illustrated in Figure 1 is simple. Web transactions may be much more complex. Web pages have their skeletons and many objects embedded. They can be downloaded from different servers; the redirection occurs during the browsing phase. The browser usually employs more advanced HTTP protocol features such as connection persistency and pipelining. These features are widely used by the browsers to speed up page downloading. The end users perceive good Internet quality through the Web page download performance. They want to know how their browsers exploit the WWW network, either well or not well. They are especially interested in how leading browsers use the network and download Web pages. Passive and active measurement techniques can be used in order to achieve visualization of Web transactions. Most of them are active prob- ing projects that are based on the target Web page measurement performed using a special measurement tool. Some of them are non-intrusive passive measurement techniques. The comparison of HTTP measurement and visualization tools, including our Wing service, is given in Table 1. Probably the MyKeynote (www. mykeynote. com) service is the most advanced benchmarking service that measures Web site’s performance and availability from a world-wide network of measurement agents. It is a commercial service with some free (demo) functionality. In the demo version, the user can make some measurements of target URL and obtain valuable information on how the target page is seen by MyKenote. Un- fortunately, the MyKenote agents use HTTP/1.0 protocol that is not still commonly used in the Internet. Leading browsers, such as MS Internet DNS server Client Web server CONNECT FIRST_BYTE LEFT_BYTES emitemit SYN ACK, SYN ACK OK GET OK GET DNS DNS2SYN ACK2GET { { DNS server Client Web server CONNECT FIRST_BYTE LEFT_BYTES emitemit SYN ACK, SYN ACK OK GET OK GET DNS DNS2SYN ACK2GET { { Figure 1. Time diagram of Web transaction 1419 Testing, Measuring, and Diagnosing Web Sites from the User’s Perspective Explorer that is used by more than 96% (www. w3schools.com) users in Polish Internet, use more advanced HTTP/1.1 protocol. Other sources report similar data. Therefore, we need tools running like most popular browsers. Nevertheless, MyKenote gives worthwhile results for understanding how the page can be loaded. The Patrick.net (www. patrick.net) is a non-commercial service for test- ing our page as it is seen from California, USA, where this site is localized. There are no periodic measurements, and the returned information is not as detailed as in the case of MyKenote but yet useful, especially if included are Patrick’s com- ments which may be used by the user to improve page downloading. Because we are often focused on the continuous observation of the Web site, we would like to have the possibility to run periodic measurements. URLs would be measured in this ZD\E\ORDGLQJWKHEDVHSDJHIURPWKHVSHFL¿F measurement sites by the MyKenote, but not by Patric.Net service. Another approach is presented by the NAPA (Network Application Performance Analyzer) project (Yoder, 2002). In this open source project made by Intel, the user can freely download the NAPA, install it on his/her Windows workstation, and see Web page downloading timelines for all browsers that are installed in this operation system. It is simple and useful, but unfortunately, our experience showed that it has some bugs and is not stable. We cannot use Unix-based browsers to process off-line data because NAPA is only a Windows-based visualization system. There are also other developments featuring the Web page performance analysis, but unfor- tunately, without the graphical option showing the diagram of how the target Web page is downloaded in time (www.empirix.com; www. paessler.net; www.websiteoptimization. com; www.softwareqatest.com). However, we think that the visualization is very important in the analysis and decision making. One picture tells m u c h m o r e t h a n n u m b e r s . I n m a ny t e c h n o l o g y a n d VFLHQFH¿HOGVWKDWLVLQVWDWLVWLFDOGDWDDQDO\VLV and data mining, we have taken advantage of the visualization tools for the past several years. The browsers download Web pages in dif- ferent ways so the Web page downloading time chart can be different, and, as a result, the actual perceived performance is different. Therefore, it is very important to have the tools which perform in the same manner as the browser under obser- vation. The best way is to build-in a tool inside the browser, but this may have a great impact on System Feature Wing NAPA MyKeynote Patrick.net WebPerf. org (a)ctive (p)assive measurements a p a a a Different browsers + + – – – Level of visualization: (p)ackets (o)bjects (t)ransactions opot o DNS Lookup (UDP protocol) + - + + ? Results presented on a Web page + - + + + Periodic measurements + - + - + Storing data for further off-line analysis + - + - - Free instant full page measurements +++++ Free periodic full page measurements + Table 1. Comparison of http measurement and visualization tools 1420 Testing, Measuring, and Diagnosing Web Sites from the User’s Perspective browsing itself, and on performance, especially. Of course, such an approach requires that this functionality is organized on the user side. But that is not a good way; we rather need a service with such functionality. The tools discussed in the previous section were directed this way with the exception of one thing — they could not work as real browsers. WING SERVICE The Wing (Web ping) Internet free service has been developed for the purpose of Web probing, visualization and performance analysis from the user perspective (Borzemski & Nowak, 2004a). Our service downloads a target page as the real Web browser. Today’s implementation is done for MS IE, which is the most popular Web browser in the Internet, but the service is developed to monitor activity of any browser. Using a real Web browser, Wing supports real-life studies and tests of Web-based applica- tions. It is very important to notice that we may observe how a real browser uses the network. We can compare browsers and Web sites; we can compare our design and other designs, as well. Such known services, unfortunately, use their own Web-browsing method. Browsing by specially developed browsers (e.g., they use HTTP/1.0 protocol only) is different from that used by the most popular browsers; therefore, the result is inadequate. Wing downloads a target page to the service location and returns to the user a page showing an HTTP timeline chart and both detailed and aggregated information about Web page down- loading. At the moment, Wing is localized in Wro- claw University of Technology (WUT) campus computer network at WASK location. WASK is the Metropolitan Academic Computer Network in Wroclaw, Poland. Our Web client that probes a target Web site is a real Web browser (MS IE), so the user can observe how a real browser uses the network. Such known tools use their own Web browsing methods. Therefore, the solutions can be different from that used by real browsers, and the results can be inadequate. Wing measures, visualizes and stores all Web page download time components shown in Figure 1. It supports IP, TCP, and UDP, DNS and HTTP protocols, logging a dozen parameters of HTTP transactions and TCP connections, thus facilitat- ing a much deeper analysis. For example, we used it in the research on the estimation of the HTTP throughput and round-trip time. Our solution is based on passive monitoring that originated with TCPdump protocol packet capture program (Jacobson, 1994; Ostermann, WKDWDOORZVUHFRUGLQJRIDOOQHWZRUNWUDI¿F WKDWÀRZVWKURXJKDOLQN7KLVDVVXPSWLRQJLYHVD possibility to take into account both unix (Linux) and Windows-based browsers. Wing traces all steps performed during browsing. We trace the entire communication between our Web client and target server. Tracing is done on the raw IP packages level. After tracing, the browsing time chart is reconstructed. The result is the same as the user achieves a browser under consideration. Hence, if for example, the browser features HTTP/1.1 persistent connection, the Wing shows that activity as well. If there is the compressed transmission, then such communication is also shown. Additionally, Wing stores collected data in a local database for further post-processing. This feature was exploited in the analysis of generic Web characteristics from the perspective RIWKHHQGXVHUDVVKRZQLQWKH¿IWKDQGVL[WK sections. The main emphasis in the Wing project was to develop a system for supporting real-life us- age of different Internet browsers. The most popular browser is MS IE running under MS :LQGRZV7KHUHIRUH¿UVWZHEXLOWDV\VWHPLQ the Windows environment. Wing uses P-CAP API. Unfortunately, our initial programming showed that there was a disadvantageous effect of the browsing process on the monitoring process. 1421 Testing, Measuring, and Diagnosing Web Sites from the User’s Perspective Therefore, we decided to use two computers, one IRUEURZVLQJDQGRQHIRUPRQLWRULQJWKHWUDI¿F The third computer is used for collecting data in a database. The Wing service has a distributed local infrastructure consisting of three components (Figure 2): • Linux-based Wing controller, • Local Web client, and • Database server. Wing is activated by a remote user request sent to the Linux-based Wing controller targeting the URL to be tested. As the service is invoked, the Wing controller starts a Web local client that issues GET request to target URL as well as GET requests for all objects that are embedded in the target page. Wing works like a sonar-location system, send- ing requests for the target Web page and waiting for the answer. The Wing controller monitors and time stamps all of the browser’s actions, determines the end of the Web page uploading, and pre-process gathered data into the format convenient for further statistical and data mining analysis, as well as for visualization. Data is also stored in a database server. The most complex processing is done after collecting the row data from the network. Then, the whole Web trans- action must be reconstructed. Transactions are usually performed in parallel TCP sessions with an active persistency feature. During the analysis, it is important to remember that the data can be compressed by the Web server; the packets can arrive in another sequence than were sent or can be doubled. We need to identify embedded objects and track all redirections in URL addresses. Also, it is not a commonplace problem to reconstruct + 7 0 /G R FX PH QW VE XL OW I U RP I U D P HV : LQ J I XO ¿ OO V all requirements in this area. The return page shows the HTTP timeline chart and a number of detailed and aggregated data about Web page downloading. Wing can be used for either instant or periodic measurements. The fourth section presents an example of instant measurement, whereas in the ¿IWKVHFWLRQZHGLVFXVVWKHDSSOLFDWLRQRISHULRGLF measurements in the estimation of the through- put and latency of network links between user’s browser and Web servers in the Internet. Database Linux Domain name o IP Connection establishment HTML skeleton downloading First object downloading Second object downloading . . . z z Internet z z z Executive Module Control and communication module e Windows Executive Module Control and communication module Figure 2. The Wing architecture 1422 Testing, Measuring, and Diagnosing Web Sites from the User’s Perspective AN ILLUSTRATIVE EXAMPLE OF INSTANT MEASUREMENT To show the issue, consider the following example of Web page downloading. We targeted the home page of the APIEMS2004 Conference. We perform and compare Web page visualizations prepared by MyKeynote (Figure 3) and Wing (Figure 4), respectively. As we can see MyKeynote uses an individual TCP connection for every object (there are 17 TCP connections as shown by Connect bars) whereas the real browser (MS IE) stimulated and observed by Wing uses only two connections to download all objects. Moreover, we also used Web Page Analyzer (www.weboptimization.com) and Paessler Site Inspector (PSI) (www.paessler.com), two popular tools for the analysis of Web pages but without the visualization feature. Table 2 shows a comparison of the measured components. The mismatches between the results are displayed in red color. LE stands for the message sent by 0\.H\QRWH³ Byte Limit Exceeded” but this limitation is well MXVWL¿HG1$LVWKHPHVVDJHGLVSOD\HGE\6LWH Inspector. MyKeynote omits the objects that are really transferred through the network by the real browser and used for improving further navigation (items 4-9). Web Page Analyzer and PSI do not properly recognize items 1 and 2. PSI recognizes Item Wing Bytes Keynote Bytes Web Page Analyzer Bytes PSI Bytes 0 + 12076 + 12076 + 12062 + 12076 1 + 30737 + 30737 + 30737 + - 2 + 27400 + 27400 - - - - 3 + 164625 + LE + 164625 + - 4 + 261 - - - - - - 5 + 287 - - - - - - 6 + 288 - - - - - - 7 + 266 - - - - - - 8 + 238 - - - - - - 9 + 280 - - - - - - 10 + 43 + 43 + 43 + 43 11 + 1584 + 1584 + 1584 + 1584 12 + 1281 + 1281 + 1281 + n/a 13 + 1870 + 1870 + 1870 + n/a 14 + 1337 + 1337 + 1337 + n/a 15 + 1919 + 1919 + 1919 + n/a 16 + 1638 + 1638 + 1638 + n/a 17 + 1420 + 1420 + 1420 + n/a 18 + 537055 + LE + 537055 + 537055 19 + 465578 + LE + 465578 + 465578 20 + 32504 + LE + 32504 + 32504 21 + 4167 + LE + 4167 + 4167 22 + 41410 + LE + 41410 + 41410 Table 2. Measured components 1423 Testing, Measuring, and Diagnosing Web Sites from the User’s Perspective item 2, however does not count its size. However, both tools recognize them correctly in the case of individual analysis of these objects. :LQJ KHOSVLGHQWLI\ LQHI¿FLHQW QHWZRUN XV- age by the browser and helps to tune Web pages WRXVHWKHQHWZRUNHI¿FLHQWO\7KHFRPSDULVRQ shows Wing can be a good analysis tool for the Web page and network application developers. 7KH³ODVWEXWQRWOHDVW´IHDWXUHLVWKDW:LQJFDQ process scripts so we can automate the usage of the service and use it in the advanced Internet measurements. PERIODIC FULL PAGE MEASURE- MENTS — WEB PERFORMANCE STUDY :HETXDOLW\LVH[WUHPHO\GLI¿FXOWWRVWXG\LQDQ integrated way. End users perceive good Web quality mainly in the context of good performance, availability, security and accessibility. Network, Web site and Web infrastructure solutions (DNS UHVROXWLRQ FDFKLQJ WUDI¿F VKDSLQJ FRQWHQW distribution networks, load balancing, etc.) have deep impact on Web quality. It has never been easy to determine whether bad performance or non-availability of a particular Web site is due to either network problems or end-systems problems, or both. The relationships between different fac- WRUVWKDWPD\LQÀXHQFHHDFKRWKHUDUHWUDQVLHQW and very complex; therefore, we cannot often exactly diagnose, isolate and evaluate key sources of problems. The evaluation of Web quality using ad hoc LQVWDQWPHDVXUHPHQWVRQO\LVQRWVXI¿FLHQW:H need to measure Web quality in periodic full page measurements. Wing supports such an active periodic full page measurements mode. Here we show how Wing was used in the empirical study of Web quality. We measured the Web from the Wroclaw University of Technology campus to evaluate how the Web is seen (in the context of perceived quality) from the perspective of users located in our campus (Borzemski & Nowak, 2004b). The users perceive Web quality mostly by latency and throughput. The main goal of our work was to answer for a question whether it is possible to develop a general model describing :HESHUIRUPDQFHIRUWKHXVHUVVXU¿QJWKH:HE from some site. Our model takes into account network latency and throughput. We investigated the correlation between latency and throughput to examine whether connections with shorter latency tend to transfer more data. Almost 60% latency, as perceived by end-us- ers at their microscopic level while accessing the Web server by the browser, refers to the network latency that is the delay between sending the UHTXHVW IRU GDWD DQGUHFHLYLQJ WKH ¿UVW ELWRI the reply. The latency is evaluated by the round trip time (RTT) which is a measure of the time it takes for a packet to travel from a computer, across a network to another computer, and back. RTT UHÀHFWV WKH,QWHUQHWSDWK SURSHUWLHVHVSHFLDOO\ the cumulative end-to-end queuing delays. The lower the latency, the faster we can do low-data activities. RTT is commonly used to check whether the target server is reachable. Then RTT samples are taken by sending ICMP messages by Ping or Traceroute tools. Ping-based RTT estimate is not very valuable in Web quality evaluation studies as ICMP packets do not match usual Web-based WUDI¿F)RUH[DPSOH,&03SDFNHWVFDQEHEORFNHG E\¿UHZDOOVDQGWKHURXWHUVRIWHQSURYLGHGLIIHU- H Q W S U LR ULW L] LQ JIR UW KH PW KD QI R U ³Q R U P DO ´ W U DI ¿F which is made of IP packets within TCP sessions used for HTTP transfers. Therefore, we need a more suitable estimation technique of RTT, that is, at the TCP (HTTP) protocol level. 7KURXJKSXW LV WKH ³QHWZRUN EDQGZLGWK´ metric which tells us about the actual number of bytes transferred over a network path during D¿[HGDPRXQWRIWLPH7KURXJKSXWGHWHUPLQHV WKH³VSHHG´RIDQHWZRUNDVSHUFHLYHGE\WKHHQG user. The higher the throughput (at the HTTP level) of Internet connection, the faster the user can surf the Internet. . of Trade and In- dustry. Retrieved March 2005, from http://www. ukonline.gov.uk Brown, D. H., & Lockett, D. (2004). Potential of critical e -applications for engaging SMEs in e -business:. Research: Electronic Networking Applications and Policy, 7(2), 95-100. Kendall, J. D., Tung, L. L. Chua, K. H., Ng, C. H. D., & Tan, S. M. (2001) Receptivity of Singapore’s SMEs to electronic. Internet Research: Electronic Networking, Applications and Policy, 11(5), 399-410. Mehrtens, J., Cragg, P. B., & Mills, A. (2001). A model of Internet adoption by SMEs. Information and Management,