System, method, and program for measuring performance in a network system

CROSS-REFERENCE TO RELATED APPLICATION

0001[0001] This application is a continuation of and claims the benefit of application Ser. No. 09/310,502 filed on May 12, 1999.

BACKGROUND OF THE INVENTION

0002[0002] 1. Field of the Invention

0003[0003] The present invention relates to a method, system and program for measuring client performance in network system in general and the Internet in particular.

0004[0004] 2. Description of the Related Art

0005[0005] One of the problems facing electronic commerce over the Internet, i.e., World Wide Web, is the delays in processing electronic transactions and returning information to a customer or user. Electronic commerce includes both purchasing products or services over the Internet as well as including advertisements in web pages that are otherwise freely available. A customer at a remote location from a merchant's server may experience a delay of several seconds to a couple of minutes to perform each step of an electronic transaction, including the steps of retrieving information on the transaction, retrieving an order form to display in the customer's Web browser, transmitting the filled out form to the merchant's Internet server form, and then receiving confirmation of the completion of the transaction. Further, the customer may experience substantial delays when purchasing and downloading multi-media files or programs over the Internet. Still further, users web pages that include requested information, such as newspaper articles, along with advertisements, may experience significant delays in retrieving the desired web page. The sum total of delays for each of these operations may discourage users from engaging in electronic commerce over the Internet.

0006[0006] There are many possible causes of delays in processing Internet transactions. A user may have a slow communication device, such as a low speed modem which substantially limits bandwidth. Further, a merchant's server may be experiencing significant traffic which increases the system load and causes transactions to be queued, i.e., wait-listed. Still further, delays may be caused by the route a file takes when transmitted across the Internet. To transfer a file across the Internet, the file is broken into many packets, which are then transferred across multiple routers to the end user, where they are assembled to form the complete file. The transmission of a packet over a router is referred to as a hop. Each time a packet is forwarded to a router a hop occurs. Transmission time increases as a packet makes hops. Delays further include the time for the graphical user interface (GUI) at the customer computer to display the information transmitted over the Internet.

0007[0007] Currently, there are utilities that can measure certain performance aspects in the route of Internet transmissions. For instance, there are utilities which test a server's ability to handle Internet file requests by load and stress testing the server with file requests. Traceroute and Packet Internet Groper (“Ping”) programs determine the route of a file transmitted over the network, i.e., all the intermediary routers and hosts through which the packets of the file passed. Traceroute and Ping are helpful for determining whether specific IP addresses are functioning, whether file packets are being lost or dropped when transmitted over the network, and the response times for requests. Large variances in the response time of Pings, also referred to as “jitter,” can indicate a cause of poor Internet performance.

0008[0008] Notwithstanding the usefulness of the present Internet performance measurement utilities, as electronic commerce extends world wide, electronic commerce merchants need tools to estimate and determine the response times for electronic Internet transactions from customers located at IP addresses throughout the world. Further, current performance utilities may measure certain network transmission times and paths, but do not measure specific client response times, such as the time it takes for the customer to display information requested from the server.

0009[0009] The lack of tools to adequately measure response time performance can prevent merchants from accurately assessing the cause of delays. Further, without adequate performance measurement tools, merchants cannot optimize and tailor the design of Web pages to accommodate the bandwidth limitations of both the customer computers and network communication paths between the merchant server and customers. If merchants cannot accurately estimate the response time from the customer end, then they may not realize problems with the design of the web page, especially delays in downloading merchant web pages, that are discouraging customers from using electronic commerce to purchase products over the Internet. Moreover, the importance of the Internet as a sales and marketing channel is increasing exponentially. Thus, the need for accurate response time measurement tools is ever more apparent and necessary to operate in the new “on-line” economy.

SUMMARY OF THE PREFERRED EMBODIMENTS

0010[0010] To overcome the limitations in the prior art described above, preferred embodiments disclose a system, method, and program for generating network performance data. Performance information including a performance time to download a page and execute the page within an application program is received. The received performance information is processed and then performance information output indicating network performance is generated in response to processing the performance information.

0011[0011] In further embodiments, executing the page comprises displaying the entire contents of the page. Still further, executing the page may comprise executing a script program to input data into the displayed page. The page including the input data is then transmitted and a response concerning the transmitted page is received. In such case, the performance time further includes the time to enter the input data into the page, transmit the page including the input data, and receive the response.

0012[0012] In still further embodiments, a network address at a location remote from the server is determined. The performance information output may then be used to estimate a performance time for a hypothetical device at the network address to download and execute the page.

0013[0013] With preferred embodiments, performance data may be gathered from remote computers and then used to assess network performance and/or estimate performance times at any network address, such as an IP address, worldwide. In preferred embodiments, the performance time includes both the network delay time and the client performance time. The network delay time includes the time to transmit a request to the server, have the server process the request and generate a response page, and then transmit the response page back to the client. The client performance time includes the time for the client processor and system resources to display the received response in a browser. Further, the client performance time may include time for the client to enter data into the received page, transmit the page to the server for processing, and then receive confirmation of the page with the inputted data from the server.

0014[0014] In this way, the performance time includes all the delay time a user, such as a customer, experiences when engaging in transactions with a remote server. With commercial transactions, the performance time can include the time to download a purchase order from a merchant's server, enter data into the purchase order, submit the purchase order to the server, and receive confirmation that the purchase order was processed. Merchants can use actual performance time information to estimate performance time at IP addresses at any location in the world. These estimated performance times provide an accurate estimate of all delays a consumer might experience, including network delays and client performance delays. With this information, a merchant or other web site host can determine whether the design of their web site, including the number of pages and content within the pages, produces unacceptable delays in user/customer processing that needs to be addressed in order to avoid losing customers dissatisfied with the delays.

BRIEF DESCRIPTION OF THE DRAWINGS

0015[0015] Referring now to the drawings in which like reference numbers represent corresponding parts throughout:

0016[0016]FIG. 1 is a block diagram illustrating a computing environment in which preferred embodiments of the present invention are implemented;

0017[0017]FIG. 2 is a block diagram of software components in which preferred embodiments of the present invention are implemented;

0018[0018]FIG. 3 illustrates logic to gather network performance data in accordance with preferred embodiments of the present invention;

0019[0019]FIG. 4 illustrates logic to measure network performance time in accordance with preferred embodiments of the present invention;

0020[0020]FIG. 5 illustrates data structures for maintaining performance data in accordance with preferred embodiments of the present invention; and

0021[0021]FIG. 6 illustrates an example of gathered and calculated data in accordance with preferred embodiments of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

0022[0022] In the following description, reference is made to the accompanying drawings which form a part hereof and which illustrate several embodiments of the present invention. It is understood that other embodiments may be utilized and structural and operational changes may be made without departing from the scope of the present invention.

Computing Enviromment

0023[0023]FIG. 1 illustrates a hardware and software environment in which preferred embodiments are implemented. A computing environment <b>2</b> comprises a server <b>4</b>, server probe <b>6</b>, multiple remote probes <b>8</b><i>a, b, c, d, </i>and a network <b>10</b> providing communication therebetween. The server <b>4</b> may be any server device known in the art capable of servicing network requests, e.g., the IBM NETFINITY®, AS/400®, and RS/6000®; and POWEREDGE® from Dell Computer Corporation.** In preferred embodiments, the server <b>4</b> would include Hypertext Transfer Protocol (HTTP) software to service HTTP requests from the server probe <b>6</b> and remote probes <b>8</b><i>a, b, c, d. </i>Further, the server probe <b>6</b> would also include HTTP software to service requests for documents from the remote probes <b>8</b><i>a, b, c, d. </i>The server <b>4</b> and server probe <b>6</b> may also include software to handle document requests in data transfer protocols other than HTTP, such FTP, Gopher, WAIS, NNTP, SMTP, etc. The server probe <b>6</b> would additionally include software to function as a client generating document requests in any of the protocols. The server probe <b>6</b> and remote probes <b>8</b><i>a, b, c, d </i>include one or more browser programs <b>12</b> that can submit document requests to the server <b>4</b>. In preferred embodiments, the browser programs <b>12</b> are Hypertext Markup-Language (HTML) browsers, e.g., NETSCAPE® Communicator, the MICROSOFT® Internet Explorer, Spry Mosaic, NCSA Mosaic, Lynx, Opera, GNUscape Navigator, etc.**, that include HTTP software to function as HTTP clients capable of transferring requests for documents to the server <b>4</b>. The HTTP server protocol is described in “Hypertext Transfer Protocol—HTTP/1.1,” Network Working Group, Request for Comments No. 2068 (January, 1997), which publication is incorporated herein by reference in its entirety. The browser program <b>12</b> may also include plug-ins or additional program features to process files in other media formats, such as sound files, movie files, JAVA™ applets**, etc. <!--footnotes removed-->

0024[0024] The network <b>10</b> may be a TCP/IP network, such as the Internet and World Wide Web, or any other network system known in the art, e.g., LAN, Ethernet, WAN, System Area Network (SAN), Token Ring, LocalTalk, TCP/IP, the Internet, etc. Alternatively, there may be separate and different networks between the server <b>4</b>, server probe <b>6</b>, and remote probes <b>8</b><i>a, b, c, d. </i>Further, the computing environment <b>2</b> may include a separate network <b>14</b>, such as an Intranet, to provide additional communication paths between the server <b>4</b> and server probe <b>6</b>. In preferred embodiments, the server <b>4</b> and server probe <b>6</b> are in the same geographical location, i.e., location A, such as the same building, room, city, plant site, etc. In alternative embodiments, the server <b>4</b> and server probe <b>6</b> may be in separate geographical locations. The remote probes <b>8</b><i>a, b, c, d </i>may be dispersed throughout geographical locations, such as different cities, states, countries, etc.

0025[0025] The server <b>4</b> and server probe <b>6</b> further include in a storage area, e.g., hard disk drive, tape drive, cache, etc., pages <b>20</b><i>a, b, c, </i>which are files intended for transmittal over the network <b>10</b>. The pages <b>20</b><i>a, b, c </i>may include any text or media data known in the art, such as text, graphical images, sound files, movies, Java applets, etc. Certain of the pages <b>20</b><i>a, b, c </i>may be forms including fields in which the user may enter data for the purpose of resubmitting back to the server <b>4</b> for further processing. For instance, certain pages <b>20</b><i>a, b, c </i>may be customer purchase order forms in which the customer enters purchase related information. Alternatively, the forms may include fields in which the user requests information, which causes the server <b>4</b> to perform database search operations. In preferred embodiments, the server <b>4</b> and server probe <b>6</b> maintain the same pages <b>20</b><i>a, b, c. </i>

0026[0026] In preferred embodiments, the server probe <b>6</b> and remote probes <b>8</b><i>a, b, c, d </i>are comprised of computer systems that are representative of the computer systems a likely user or customer would use. The merchant or web site host setting up the computing environment <b>2</b> to measure Internet performance for an electronic commerce web site would want to implement the server <b>6</b> and remote <b>8</b><i>a, b, c, d </i>probes in computers that would likely be used by their customers. For instance, the server <b>6</b> and remote <b>8</b><i>a, b, c, d </i>may be implemented in a personal computer that includes a common processor, such as an INTEL PENTIUM processor at a speed that is commonly used, e.g.,166 MHZ, a common operating system, such as Microsoft WINDOWS®, OS/2®, etc., and a common browser <b>12</b>, e.g., Netscape, etc.** For instance, if the merchant is selling software most often used in personal computers, then an Intel Pentium type machine with Microsoft Windows may be used to implement the probes <b>6</b>, <b>8</b><i>a, b, c, d. </i>However, if the merchant is selling UNIX®** products, then the probes <b>6</b>, <b>8</b><i>a, b, c, d </i>would be implemented in hardware and software typically used by UNIX users. In preferred embodiments, the server probe <b>6</b> is implemented as the same computing system as the remote probes <b>8</b><i>a, b, c, d </i>to allow the server probe <b>6</b> to measure performance with respect to the server <b>4</b> that is not subject to distance factors. <!--footnotes removed-->

0027[0027] In preferred embodiments, both server probe <b>6</b> and remote probes <b>8</b><i>a, b, c, d </i>include a response time measurement (RTM) program <b>16</b>, illustrated in FIG. 2. The RTM program <b>16</b> may be implemented as a stand alone application program that includes the program logic in which the performance measurement tools of the preferred embodiments are implemented. Preferably, these programs are written in an object oriented computer language, such as JAVA**, that may be executed on different operating system platforms. <!--footnotes removed-->

0028[0028]FIG. 2 provides details of the program components within the RTM program <b>16</b>. The RTM program <b>16</b> includes a performance measurement program <b>30</b> to measure actual performance data from downloading pages <b>20</b><i>a, b, c </i>from the server <b>4</b>. The performance measurement program <b>30</b> in the remote probes <b>8</b><i>a, b, c, d </i>is configured to transmit performance measurement data to the RTM program <b>16</b> in the server probe <b>6</b>. A screen scraping routine <b>34</b>, included within the performance measurement program <b>30</b> measures the time to download and display or further process a page <b>20</b><i>a, b, c. </i>The RTM program <b>16</b> further includes a data analysis program <b>36</b> to analyze performance data collected by the probes <b>6</b>, <b>8</b><i>a, b, c, d. </i>A performance estimator program <b>38</b> utilizes data produced by the data analysis program <b>36</b> to estimate the time in which remote computers at distant locations would download and process the pages <b>20</b><i>a, b, c. </i>Thus, merchants and web site hosts would estimate the time to complete electronic transactions with the performance estimator program <b>38</b>.

0029[0029] In preferred embodiments, the same RTM program <b>16</b> is included in all the probes <b>6</b>, <b>8</b><i>a, b, c, d. </i>In the above discussed embodiments, the RTM programs <b>16</b> in the remote probes <b>8</b><i>a, b, c, d </i>are configured to transmit performance data to the server probe <b>6</b> for data analysis and performance estimation at the server probe <b>6</b>. System managers may then use the data analysis <b>36</b> and performance estimator <b>38</b> programs at the server probe <b>6</b>, which is preferably at the same location as the server <b>4</b>, to analyze network performance data. In alternative embodiments, one of the remote probes <b>8</b><i>a, b, c, d </i>may be configured to receive the performance data from the other remote probes and the server probe <b>6</b>. For instance, to have remote probe <b>8</b><i>a </i>receive data, the RTM programs <b>16</b> in remote probes <b>8</b><i>b, c, d </i>and server probe <b>6</b> would be configured to transmit their performance data to the remote probe <b>8</b><i>a; </i>remote probe <b>8</b><i>a </i>would be configured to store its performance data. A system manager at such remote probe <b>8</b><i>a </i>may then run the data analysis <b>36</b> and performance estimator <b>38</b> programs to perform network performance analysis at the remote probe.

0030[0030] If there are many remote probes, then the performance of the probe that receives the data and its ability to run its own performance tests could be degraded by continually receiving performance data from many probes. In such case, a server computer known in the art may be added which includes a copy of the RTM program <b>16</b>. All the probes <b>6</b>, <b>8</b><i>a, b, c, d </i>may be configured to send their performance data to the added server. Data analysis and performance estimation may then be performed at the added server. The use of the added server insures that the probes' performance and ability to generate performance data that reflects the performance of a common customer or user is not adversely affected by having to receive data from all the other probes and perform data analysis operations. Still further, the added server may provide a Web page or other network accessible file that displays or provides in real-time network performance data from the probes and performance estimation data.

Gathering Performance Measurement Data

0031[0031]FIG. 3 illustrates logic implemented in the performance measurement program <b>30</b> to gather performance data on the time to download, display, and/or process web pages from the server <b>4</b>. Control begins at block <b>50</b> with the remote probe <b>8</b><i>a, b, c, d </i>or server probe <b>6</b> executing the performance measurement program <b>30</b>. The program <b>30</b> may be set to execute at predetermined time periods to measure performance data at different times of the day and/or on different days. For instance, the web site host may want to measure performance data during peak usage hours, such as during business hours or in the early evening when users have time after work to access the Internet, and at non-peak usage, such as in the late evening or weekends. The predetermined times to run the performance measurement program <b>30</b> may also be set to time periods when customers are most likely to access the merchant's web site to engage in commercial transactions. In preferred embodiments, the web site host may set such predetermined times at which to execute the program <b>30</b> at the server probe <b>6</b>. The predetermined times may then be transmitted to the remote probes <b>8</b><i>a, b, c, d </i>to set their times. Further, both the server <b>6</b> and remote <b>8</b><i>a, b, c, d </i>probes are preferably configured to execute the performance measurement programs <b>30</b> at the same time and date to insure that both gather performance data under similar network <b>10</b> circumstances, as the network response time may vary significantly throughout the day and on different days.

0032[0032] After initiating the performance measurement program <b>30</b>, a 1K ping is generated from the remote probe <b>8</b><i>a, b, c, d </i>to the server <b>4</b> to determine a response time in milliseconds. Details of “pinging” are described in the published Request For Comments (RFC) no. 792, entitled “Internet Control Message Protocol” (September, 1981), by J. Postel, which document is incorporated herein by reference in its entirety. Any pinging method known in the art may be used. A loop then begins (at block <b>52</b>) to gather performance data for i pages, where i is an integer from 0 to n. For instance, if a commercial transaction requires the user to download and, in some instance, process a sequence of i pages, then the performance data would be gathered for each of the i pages involved in the commercial transaction. The loop begins (at block <b>54</b>) with the performance measurement program <b>30</b> generating and transmitting a request for a page i, e.g., a HTTP GET command for web page i at a Universal Resource Locator (URL) at the host web site. The performance measurement program <b>30</b> then (at block <b>56</b>) calls a screen scraping routine <b>34</b> (the details of which are described below with respect to FIG. 4) to determine the time to download and display page i in the browser <b>12</b> graphical user interface (GUI). The term “performance time” with respect to a page i, as used herein, refers to the time to download and display page i. The term “performance data” as used herein refers to the collection of performance times over numerous instances at one or more of the probes <b>6</b>, <b>8</b><i>a, b, c, d. </i>In this way, the performance time captures not only the network transmission time to download a page, but also the time to display or otherwise execute the downloaded page in the browser <b>12</b>. Total performance time is impacted by both the network transmission time, which is affected by network factors and connection capabilities, and the probes <b>6</b>, <b>8</b><i>a, b, c, d </i>processing capabilities, which is preferably representative of the customer's processing capabilities. If the page <b>20</b><i>a, b, c </i>includes media other than images and text, such as sound files, movie files and/or Java applets, then the “display” time would include the time to complete presenting the information in the sound file, movie file or Java applet to the user within the user's browser <b>12</b>. In this way, a measurement of performance time includes the total wait time the user must endure to obtain information, regardless of the media in which the information is implemented.

0033[0033] After determining the performance time for page i (at block <b>56</b>), the performance measurement program <b>30</b> stores (at block <b>58</b>) the measured performance time for page i, along with other information, such as the time and date the performance time was measured. The performance measurement program <b>30</b> then determines (at block <b>60</b>) whether a script routine is provided to perform further processing of page i, which at this moment is displayed or executing in the browser <b>12</b>. If there is a script program for page i, then the performance measurement program <b>30</b> starts (at block <b>62</b>) measuring the time to further process page i. If there is no script program for page i, then another iteration (at block <b>64</b>) of the loop beginning at block <b>52</b> is performed if there are further pages <b>20</b><i>a, b, c </i>to download and process. After starting time measurement (at block <b>62</b>), the performance measurement program <b>30</b> executes (at block <b>55</b>) the script to perform the further processing operation. For instance, if page i, is a form in which the user or customer enters information, then the script may cause the probe <b>6</b>, <b>8</b><i>a, b, c, d </i>to enter input information into the form fields of page i in the same manner a user would. The script may include other processing operations, such as saving the page i at a specific location or including in a message to transfer to another location. There may be a separate script program unique to each page i, as the fields included in different pages <b>20</b><i>a, b, c </i>may require different script operations.

0034[0034] If the script program enters input data into form fields in page i, then the performance measurement program <b>30</b> transmits (at block <b>68</b>) the page i, including the input data, to the server <b>4</b> and then executes the screen scraping routine <b>34</b> (at block <b>70</b>) to determine the time to receive a response from the server <b>4</b> indicating that page i has been processed, e.g., the electronic purchase was received and is being processed by the merchant. After receiving a response (at block <b>70</b>), the performance measurement program <b>30</b> stops measuring the performance time to further process page i, and adds the performance time to process page i with the script program and receive confirmation to the time to download and display page i, which was stored at block <b>58</b>. In this way, the performance time with respect to a page i that is downloaded and further processed includes the time to download and display the page <b>20</b><i>a, b, c, </i>the time for the user to enter information into the page, submit the filled-in page to the server <b>4</b>, have the server <b>4</b> process the page, and then receive confirmation that the transmitted filled-in page was processed. The performance measurement program <b>30</b> may then perform (at block <b>64</b>) another iteration of the loop beginning at block <b>52</b> if there are further pages <b>20</b><i>a, b, c </i>to process.

0035[0035]FIG. 4 illustrates the program logic of the screen scraping routine <b>34</b>. Control begins at block <b>100</b> with a call to the screen scraping routine <b>34</b> from within the performance measurement program <b>30</b> to determine the performance time for an action, such as downloading and displaying a web page or transmitting a page and receiving a response. In preferred embodiments, the performance measurement program <b>30</b> calls the screen scraping routine <b>34</b> with a parameter of the first character or character string that will be displayed at the top of the page <b>20</b><i>a, b, c </i>for which the performance time is being measured. Orientation information of this character string on the page <b>20</b><i>a, b, c </i>being downloaded may also be provided. After being called (at block <b>100</b>), the screen scraping routine <b>34</b> determines the brand type of the browser <b>12</b>, e.g., Netscape, Windows Explorer, etc. The screen scraping routine <b>34</b> will include different sets of code to perform certain operations for different browser types if the probes <b>6</b>, <b>8</b><i>a, b, c, d </i>include multiple browsers. For instance, the code to determine whether the browser <b>12</b> is displaying a “done” indicator differs according to the type of browser <b>12</b>. The screen scraping routine <b>34</b> then begins a loop (at block <b>104</b>) of scraping operation to perform continuously until the loop is broken. After beginning the loop (at block <b>104</b>), the screen scraping routine <b>34</b> determines (at block <b>108</b>) whether the character string provided with the call is displayed at the top position in the browser <b>12</b> GUI. If so, the screen scraping routine <b>34</b> then determines whether a “done” message is displayed in the browser <b>12</b> GUI, at the location where such message is displayed for the browser <b>12</b> type. If the character string is not displayed at the top of the browser <b>12</b> GUI (at block <b>106</b>), then the screen scraping routine <b>34</b> determines (at block <b>110</b>) whether the loop has been executing longer than a predefined time-out period. If so, the loop and screen scraping routine <b>34</b> terminates (at block <b>114</b>) with a return of an error message indicating that the page <b>20</b><i>a, b, c </i>did not successfully download. If the time-out period has not been exceeded, then the screen scraping routine <b>34</b> performs another iteration of the scraping operation loop (at block <b>104</b>) after a predetermined time has elapsed since the previous iteration of the loop at block <b>104</b>. For instance, the scraping operation could be performed sixty times per minute to accurately determine the time to completely download and display or run the requested page <b>20</b><i>a, b, c. </i>

0036[0036] If the “done” message is displayed (at block <b>108</b>), then the screen scraping routine <b>34</b> terminates the loop starting at block <b>104</b> and returns (at block <b>112</b>) indication that the page has been successfully downloaded and displayed in the browser <b>12</b>. Otherwise, if “done” is not displayed, then the screen scraping routine <b>34</b> proceeds to perform another iteration of the loop at block <b>104</b> if the time-out period is not exceeded.

0037[0037] If the performance measurement program <b>30</b> returns a downloaded page <b>20</b><i>a, b, c </i>filled-in with data to the server <b>4</b>, then the reply time would include the time for the server <b>4</b> to process the returned page <b>20</b><i>a, b, c, </i>such as time to do any database searches or processing requested by the page <b>20</b><i>a, b, c </i>or that must be performed to process the returned page <b>20</b><i>a, b, c. </i>For instance, if the user enters search information into the page <b>20</b><i>a, b, c </i>submitted to the server <b>4</b>, then the reply time would include the time for the server <b>4</b> to execute the search, retrieve requested information, and return such information to the probe <b>6</b>, <b>8</b><i>a, b, c, d. </i>Alternatively, the server <b>4</b> may have to execute database searches using data entered in the returned form to verify certain information before completing the transaction and returning verification to the probe <b>6</b>, <b>8</b><i>a, b, c, d. </i>

0038[0038] If the screen scraping routine <b>34</b> is measuring the time to generate a sound file or display a movie file, in the event that the page <b>20</b><i>a, b, c </i>is in alternative media formats, then the screen scraping routine <b>34</b> would include logic to monitor information displayed in plug-in programs and/or the browser <b>12</b> to determine when the presentation of information in such alternative media formats has completed.

Analyzing Performance Data

0039[0039] After the performance measurement programs <b>30</b> have gathered performance times with respect to downloading and processing the pages <b>20</b><i>a, b, c, </i>then the remote probes <b>8</b><i>a, b, c, d </i>would transmit the gathered performance data to the server probe <b>6</b> for analysis and estimation. The data analysis program <b>36</b> processes the performance times to produce information that the merchant may review or the performance estimator program <b>38</b> may use to estimate worldwide delay times. The data from the probes <b>6</b>, <b>8</b><i>a, b, c, d </i>includes the identity information on the probe <b>6</b>, <b>8</b><i>a, b, c, d </i>that conducted the test, test time, ping time, and the performance times by page <b>20</b><i>a, b, c. </i>

0040[0040]FIG. 5 illustrates a preferred embodiment of how the data analysis program <b>36</b> may group the performance data received from the performance measurement programs <b>30</b> into various fields. The grouped fields of data may be stored in a database program or other storage format. In FIG. 5, a group of fields <b>150</b>-<b>156</b> include information on a particular remote probe <b>8</b><i>a, b, c, d </i>and a performance test. Field <b>150</b> includes a remote probe <b>8</b><i>a, b, c, d </i>identifier; field <b>152</b> indicates the hardware and operating system for the remote probe <b>8</b><i>a, b, c, d </i>identified in field <b>150</b>; field <b>154</b> indicates the connection device included in the remote probe <b>8</b><i>a, b, c, d </i>identified in field <b>150</b>, e.g., cable modem, telephone modem, T-1 line, etc.; field <b>156</b> indicates the geographical location of the remote probe <b>8</b><i>a, b, c, d </i>identified in field <b>150</b>; field <b>158</b> identifies the date and time when a performance test was initiated; and field <b>160</b> indicates the ping reply time in milliseconds between the remote probe <b>8</b><i>a, b, c, d </i>identified in field <b>150</b> and the server <b>4</b> at the performance time.

0041[0041] Fields <b>162</b>-<b>170</b> include performance times with respect to a page <b>20</b><i>a, b, c </i>and calculated information. In preferred embodiments, there is an instance or record of fields <b>162</b>-<b>170</b> for each page <b>20</b><i>a, b, c </i>downloaded and/or further processed as part of a performance test at a particular remote probe <b>8</b><i>a, b, c, d, </i>identified by fields <b>150</b>-<b>160</b>. Thus, for each group of fields <b>150</b>-<b>160</b> that identify a remote probe <b>8</b><i>a, b, c, d </i>and performance test there are one or more instances of fields <b>162</b>-<b>170</b> for each performance test performed with respect to a page <b>20</b><i>a, b, c. </i>Field <b>162</b> indicates page <b>20</b><i>a, b, c </i>identification information on the page <b>20</b><i>a, b, c </i>subject to the performance test, including information on the page size; field <b>164</b> provides the performance time to request, receive and display or run, and, perhaps, perform further processing of the page <b>20</b><i>a, b, c </i>indicated in field <b>162</b>; field <b>166</b> provides the performance time for the server probe <b>6</b> to perform the same operation that generated the performance time for field <b>162</b>. Preferably, the server probe <b>6</b> would perform the operation at the same time the remote probe <b>8</b><i>a, b, c, d </i>performed the operation; field <b>168</b> provides a network delay time calculated by the data analysis program <b>36</b> by subtracting the performance time in field <b>166</b> from the performance time in field <b>164</b>; and field <b>170</b> provides a network delay to ping ratio that the data analysis program <b>36</b> calculates by dividing the network delay time in field <b>168</b> by the ping response time in the related field <b>160</b>.

0042[0042] The information in the fields is useful in assessing performance with respect to customers attempting to access pages <b>20</b><i>a, b, c </i>in the server <b>4</b>. The performance time in field <b>164</b> indicates the total time for a remote probe <b>8</b><i>a, b, c, d </i>to process a page <b>20</b><i>a, b, c </i>from the server <b>4</b>. This information indicates not only network transmission time, but processing time in the remote probe <b>8</b><i>a, b, c, d </i>to display or generate the transmitted page <b>20</b><i>a, b, c </i>or the time for the remote probe <b>8</b><i>a, b, c, d </i>to enter information and submit a downloaded page <b>20</b><i>a, b, c </i>back to the server <b>4</b> for further processing and confirmation. The performance time, thus, provides the total time the user or customer must wait to retrieve and view a page <b>20</b><i>a, b, c </i>from the merchant's server and/or submit a page, perhaps including a purchase order. Field <b>166</b> indicates a performance time measurement for the same operation yielding the performance time in field <b>164</b>, but performed between the server probe <b>6</b> and server <b>4</b>. This value includes delay factors that may be inherent in the server <b>4</b>, i.e., the time for the server to retrieve and return a page <b>20</b><i>a, b, c, </i>including overall network load and other network transmission factors that do not include the transmission distance factors of transmitting a page <b>20</b><i>a, b, c </i>between the server <b>4</b> and the remote probe <b>8</b><i>a, b, c, d </i>at a distant geographical location. As discussed, the server probe <b>6</b> is preferably located in the same location A as the server <b>4</b>.

0043[0043] The network delay time in field <b>168</b>, which the data analysis program <b>36</b> calculates by subtracting the performance time in field <b>166</b> from the performance time in field <b>164</b>, indicates a network transmission time that includes primarily the transmission delay due to routing a page through the network <b>10</b> and across the geographical distance between the server <b>4</b> and remote probe <b>8</b><i>a, b, c, d. </i>For instance, transmission from the server <b>4</b> to the remote probe <b>8</b><i>a, b, c, d </i>may involve the transmission of the file in multiple packets through multiple routers, i.e., hops. At the routers, the transmission may be routed to various intermediary routers before reaching the target location indicated by the IP address. Thus, the network delay time in field <b>168</b> indicates delays associated with routing the transmission packets through multiple routers between the server <b>4</b> and remote probe <b>8</b><i>a, b, c, d, </i>including the multiple routers, regional network traffic, and the router traffic management algorithms that affect transmissions. The performance time, indicated in field <b>166</b>, for transmissions between the server <b>4</b> and server probe <b>8</b><i>a, b, c, d </i>is typically less than the performance time, indicated in field <b>164</b>, because the transmission between the server <b>4</b> and physically proximate server probe <b>6</b> requires far fewer hops than the transmission between the remote probe <b>8</b><i>a, b, c, d </i>and the server <b>4</b>. In this way, by subtracting the performance time in field <b>166</b>, the data analysis program <b>26</b> is subtracting delay times due primarily to server <b>4</b> factors, such as load, server traffic, and the server input/output processing capabilities.

0044[0044] The network delay time to ping ratio in field <b>168</b> provides a correspondence of the actual network delay time for a particular page with a 1K ping, which is page size independent, between the remote probe <b>8</b><i>a, b, c, d </i>and server <b>4</b>. This ratio in field <b>168</b> is used to estimate, as discussed below, performance times between the server <b>4</b> and hypothetical clients at IP addresses at remote geographical locations. The network delay time to ping ratio in field <b>168</b> is maintained for each of the different pages <b>20</b><i>a, b, c </i>subject to performance testing. In preferred embodiments, the pages <b>20</b><i>a, b, c </i>are selected to represent a range of page sizes <b>20</b><i>a, b, c </i>that will be transmitted between a customer at the remote probe <b>8</b><i>a, b, c, d </i>and the server <b>4</b>.

0045[0045] A Multiple Socket Simultaneous Feed Connection Factor (MSSF Page factor) is a network delay factor concerning delays caused by increases in file size. Files transmitted over a TCP/IP network, such as the Internet, are broken down into multiple packets and then separately transmitted through routers to reach the target destination. In network transmissions, the algorithms in the routers queue larger files for longer periods of time than smaller files. Further, the larger files are broken down into smaller packets at the routers, and may be transmitted through different routes to reach the target computer. For these reasons, it takes longer to route a larger file through the Internet, and through the various routers, than a smaller file. This file size factor that increases network delay is the MSSF Page Factor. The network delay time in field <b>168</b> and network delay to ping ratio in field <b>170</b> are maintained for each page <b>20</b><i>a, b, c </i>included in the performance test. Thus, differences in the network delay time among the pages <b>20</b><i>a, b, c </i>having different sizes reflects the MSSF Page Factor.

0046[0046]FIG. 6 provides a table <b>200</b> including an example of the values that may be included in fields <b>162</b>-<b>170</b> for performance tests at a remote probe <b>8</b><i>a, b, c, d. </i>Information for the performance tests on the remote probe <b>8</b><i>a, b, c, d, </i>the time of the tests, and ping time would be maintained in related fields <b>150</b>-<b>160</b>. The table <b>200</b> includes instances of the fields <b>162</b>-<b>170</b> for performance tests for pages <b>20</b><i>a</i>, <b>20</b><i>b</i>, and <b>20</b><i>c. </i>In this example, the ping response time in field <b>160</b> is 2 milliseconds and page <b>20</b><i>a </i>is smaller in size then page <b>20</b><i>b, </i>which is smaller in size than page <b>20</b><i>c. </i>As can be seen from the column for field <b>164</b> in FIG. 6 that provides the performance time between the server <b>4</b> and remote probe <b>8</b><i>a, b, c, d, </i>as the page size increases, the performance time increases from 12 seconds to 15 seconds to 17 seconds. This reflects the additional network transmission time to transmit a larger page over the network <b>10</b> which is likely caused, in part, by the MSSF Page Factor. The third column includes field <b>166</b>, i.e., the performance time to transmit the page between the server <b>4</b> and server probe <b>6</b>. This time is less than the performance time for the remote probe <b>8</b><i>a, b, c, d </i>because, in preferred embodiments, the server probe <b>6</b> is physically proximate to the server <b>4</b>. The network delay time, i.e., field <b>168</b>, increases as the page <b>20</b><i>a, b, c </i>size increases. As mentioned, this increase in network delay is due, in part, to the MSSF Page Factor, reflecting the longer time to transmit over the Internet network <b>10</b> a larger page than a smaller page. The network delay to ping ratio, i.e. field <b>170</b>, is calculated for each page <b>20</b><i>a, b, c </i>and increases as the page size increases also reflecting, in part, the greater amount of network transmission time to transfer a larger page.

0047[0047] Other techniques may be used to calculate the values described above. For instance, during data gathering, instead of measuring the performance time between the remote probe <b>8</b><i>a, b, c, d </i>and server <b>4</b>, the performance time can be measured between the server probe <b>6</b> and remote probe <b>8</b><i>a, b, c, d </i>by having the remote probe <b>8</b><i>a, b, c </i>request and submit pages to the server probe <b>6</b> in the same manner the remote probe <b>8</b><i>a, b, c, d </i>would request and submit pages to the server <b>4</b>. The logic of FIGS. 3 and 4 may be used to gather performance time data for operations between the remote probe <b>8</b><i>a, b, c, d </i>and server probe <b>6</b>. Performance times between the server probe <b>6</b> and remote probes <b>8</b><i>a, b, c, d </i>would reflect a network transmission time that does not incorporate load and traffic burdens on the server <b>4</b> that cause delays in response times as the server probe <b>6</b> does not experience the network traffic that the server <b>4</b> experiences. Alternatively, to incorporate some, but not all, of the delays due to server load and traffic, the performance time between the server <b>4</b> and the remote probe <b>8</b><i>a, b, c, d </i>in field <b>164</b> can be calculated by averaging the performance time between the remote <b>8</b><i>a, b, c, d </i>and server <b>6</b> probes and the remote probe <b>8</b><i>a, b, c, d </i>and server <b>4</b>. This will incorporate some aspect of server <b>4</b> load and traffic as such server load and traffic factors effect the real network transmission time customers at the remote probes <b>8</b><i>a, b, c, d </i>experience.

0048[0048] The performance estimator program <b>38</b> may utilize the data processed and generated by the data analysis program <b>36</b> to estimate the remote probe to server performance time using equation (1) below:

(ND/PING RATIO)*(IP PING)+(S-SP PT)  (1)

0049[0049] The ND/PING RATIO is the network delay time to ping ratio maintained in field <b>170</b>. The IP PING is the response time when the performance estimator program <b>38</b> pings an IP address at a location in the world to estimate the time to retrieve and/or further process a web page from that IP address location. The S-SP PT is the server to server probe performance time, which is maintained in field <b>166</b>, which is added back to the estimated network delay time to provide an estimated remote probe to server performance time. This estimated remote probe to server performance time is the estimated time for a customer at a computer, having a configuration similar to the configuration of the probes <b>6</b>, <b>8</b><i>a, b, c, d, </i>at the location of the IP PING.

0050[0050] The performance estimator program <b>38</b> may average the network delay to ping ratios, in field <b>168</b> for a given page <b>20</b><i>a, b, c </i>and across remote probes <b>8</b><i>a, b, c, d </i>to obtain an averaged ND/PING RATIO and average the server to server probe performance times for a given page <b>20</b><i>a, b, c </i>and across remote probes <b>8</b><i>a, b, c, d </i>to obtain an average S-SP PT. These averaged ND/PING RATIO and S-SP PT values may then be used with an IP PING to estimate the performance time at a hypothetical user or customer computer at the pinged IP address. Alternatively, the performance estimator program <b>38</b> could average the ND/PING RATIO and S-SP PT for a given page <b>20</b><i>a, b, c </i>across remote probes <b>8</b><i>a, b, c, d </i>in a distinct geographical location to provide more accurate geographical estimates. For instance, the averaged ND/PING RATIO and S-SP PT could be specific to a geographical area, thereby providing an estimate that incorporates network delays specific to that geographical location, as well as page size and traffic specific to the time and date of the performance test.

0051[0051] Other ways of averaging and combining the gathered data may be used to generate averages of the ND/PING RATIO and S-SP PT to use with equation (1). For instance, averages can be taken across test times, i.e., field <b>158</b>.

0052[0052] Preferred embodiments thus allow merchants or other web site hosts to estimate the time for a hypothetical user or customer at a remote IP address to download, display the web page, and perform further precessing, such as submitting a filled in downloaded page to the server to process and provide confirmation. Preferred embodiments allow such estimated wait times to be specific to particular dates and times, geographical area, and page size. In this way, the web site host or merchant can determine delay times at different times of day, e.g., peak usage hours or off-peak usage hours, at different geographical locations, e.g., Japan, United States, California, Britain, etc., and for different page sizes. Providing an estimate that is based on performance data for a particular web page size incorporates MSSF Page Factor delays in transmitting the web page between the hypothetical user at the IP address and server. Further, the estimated wait times would incorporate wait times for specific computer systems the user or customer would have because the values in equation (1) that are used to estimate the performance times are based on the hardware and software configuration of the probes <b>6</b>, <b>8</b><i>a, b, c, d. </i>

0053[0053] Based on the estimated delay times for hypothetical customers at IP addresses around the world, a merchant can then determine whether the estimated network delays are acceptable, i.e., unlikely to discourage users or customers from requesting the pages or performing electronic commerce through the pages. If the delay times are unacceptable, then the web site host or merchant may modify, if necessary, the web pages to reduce the network delays customers around the world would experience when accessing the web pages that comprise the commercial transaction. To minimize delay times, the merchant may reduce the content in the web pages to reduce the size of the pages, thereby reducing delays attributed to the MSSF page factor, or consolidate the content from multiple pages into fewer pages. Alternatively, if the merchant or web site host believes that the network delay time could be increased without discouraging access, the merchant or host may increase the content and/or number of web pages involved in the transaction to provide more information, additional advertisements or enhance the presentation of the information.

Conclusion

0054[0054] This concludes the description of the preferred embodiments of the invention. The following describes some alternative embodiments for accomplishing the present invention.

0055[0055] The preferred embodiments may be implemented as a method, apparatus or article of manufacture using standard programming and/or engineering techniques to produce software, firmware, hardware, or any combination thereof. The term “article of manufacture” (or alternatively, “computer program product”) as used herein is intended to encompass one or more computer programs and data files accessible from one or more computer-readable devices, carriers, or media, such as a magnetic storage media, “floppy disk,” CD-ROM, a file server providing access to the programs via a network transmission line, holographic unit, etc. Of course, those skilled in the art will recognize many modifications may be made to this configuration without departing from the scope of the present invention.

0056[0056] In preferred embodiments the probes <b>6</b>, <b>8</b><i>a, b, c, d </i>and server <b>4</b> were described as having particular configurations. However, in further embodiments the probes <b>6</b>, <b>8</b><i>a, b, c, d </i>may be comprised of any network client device and the server <b>4</b> may be comprised of any network server device. Further, as discussed additional computers, such as servers, that are not functioning as probes may be used to gather the probe data and run the data analysis and performance estimation programs.

0057[0057] Preferred embodiments of the data fields used to maintain performance data were described with respect to FIG. 5. However, in alternative embodiments, additional data types may be added or data fields removed. Further, alternative arrangements and relations of the fields and performance data may be provided according to database management techniques known in the art, including relational database management techniques.

0058[0058] In preferred embodiments the gathered data was used to estimate delay times at hypothetical computers around the world. In alternative embodiments, the web site host or merchant may just review a presentation of the gathered data to analyze the acceptability of network delay times without performing estimates of delay times using the performance estimator program.

0059[0059] Preferred embodiments discussed the advantages of estimating network delay times with respect to commercial transactions, such as purchasing products. However, web site hosts in general would be interested in the preferred embodiment estimation technique even if users are not purchasing products from the web site host. For instance, many web site hosts, such as newspapers, provide information for free and present advertisements with the free information. Significant network delays for such “free” information could discourage browsing activity at the web site, which would reduce the number of people viewing advertisements and, hence, reduce advertisement revenues. Thus, web site hosts providing advertisements along with free information would want to estimate network delay times to determine the maximum number of advertisements that may be included on a page and still create acceptable network delay times that would not unduly discourage browsing activities. For such uses, the network delay time to estimate would include the time for a user to request a page, the server to obtain the requested page and perform any necessary database searches, the time to transmit the requested information to the user, and the time for the user computer to display the requested information.

0060[0060] In summary, preferred embodiments disclose a system, method, and program for generating network performance data. Performance information including a performance time to download a page and execute the page within an application program is received. The received performance information is processed and then performance information output indicating network performance is generated in response to processing the performance information.

0061[0061] The foregoing description of the preferred embodiments of the invention has been presented for the purposes of illustration and description. It is not intended to be exhaustive or to limit the invention to the precise form disclosed. Many modifications and variations are possible in light of the above teaching. It is intended that the scope of the invention be limited not by this detailed description, but rather by the claims appended hereto. The above specification, examples and data provide a complete description of the manufacture and use of the composition of the invention. Since many embodiments of the invention can be made without departing from the spirit and scope of the invention, the invention resides in the claims hereinafter appended.