COS 116 The Computational UniverseLaboratory 8: Internet Structure and Congestion ControlYou saw in lecture that the Internet is a loose, decentralized network of computers that communicate with each other using TCP/IP (Transmission Control Protocol / Internet Protocol). We also saw that IP breaks data transmissions into packets. The Internet should not be confused with the World Wide Web (WWW); in fact it existed for many years before the WWW. In everyday usage the term WWW is used as shorthand for all useful content that resides on the Internet, but it also has a technical meaning: it refers to web pages that were all created using languages such as HTML (used for your course blog). These languages allow web pages to have hyperlinks to each other, thus giving rise to a conceptual network or graph which we thought about earlier in the context of web search. This conceptual network resides on the Internet, since web pages are stored on server computers (such as www.princeton.edu), which use the Internet to communicate with each other. These servers use the HTTP protocol, which, like every other means of Internet communication, uses IP.Today’s lab will give us a peek inside the Internet. Of special interest will be the phenomenon described in the lecture as “dependence on the kindness of strangers.” There is no centralized control or predetermined limit on the amount of data communication allowed to a particular user or computer. Nobody can stop us all from going online at the same time and begin downloading our favorite files. However, it is clear that this could result in chaos: the Internet would get overloaded with packets and, unable to deliver all of them, would drop most. This used to happen a lot in the 1980s. Ultimately, most makers of Internet software agreed upon an informal convention for sharing the resources of the network that has so far kept the Internet running smoothly. It is interesting to speculate—from political, social, and technological viewpoints—how much longer this informal agreement will continue to last. In the meantime, it provides us with a good case study of congestion issues in society and their solutions. For this lab you’ll be using Clack, a powerful network simulator tool created by Dan Wendlandt for his bachelor’s thesis at Stanford. (Dan is now at Carnegie Mellon and has helped tremendously with this lab.) We have to resort to a simulator for experiments because the decentralized nature of the Internet makes it impossible to run controlled experiments. Clack creates a virtual world that merges seamlessly with the real Internet. Your browser thinks it is communicating with a server on the real Internet, but the requests from your browser are actually directed into a virtual world. Do not try to think too hard about where the real network ends and the virtual one begins, as that is not the point of this lab. (But, we are happy to answer any questions.) As usual, feel free to discuss with other students or a TA. However, you are not allowed to copy another student’s answers.Hand in your lab report at the beginning of lecture on Tuesday, April 17. Include responses to questions printed in bold. (Number them by Part and Step.) Part 1: Entering the “Matrix”Let us enter the virtual world of Clack.1. Start Internet Explorer and visit the following web page to launch Clack:http://www.clackrouter.net/princeton/lab.html(If you are prompted with a security warning, click “Ok” and “Yes” as needed.)2. You should see the Clack application window, as pictured above. Clack is set up to simulate a small network consisting of two servers (“www-server” and “ftp-server”) connected to the Internet through a router (“vrhost”) and a firewall. In this picture, you and your lab mates exist in the cloud labeled “Internet.” We will refer to this picture as your virtual world.3. IP addresses: Like all Internet devices, the ones in this virtual world are identified by numeric IP addresses consisting of four numbers separated by periods. As mentioned in lecture, each of the four numbers lies between 0 and 255. In the above picture we’ve circled the IP address of the virtual web server, 171.67.240.142. This address changes each time you restart Clack. Also, your neighbor is seeing a different IP address in his/her virtual world.4. Write down the IP address of www-server in your virtual world and mention it in your report. (It probably will not be the same as the address in the picture.) In the instructions that follow, whenever we write [www-server] you should replace that with the IP address you wrote down in this step. COS 116 – Lab 925. Open a new Internet Explorer window and enter the address http://[www-server]/princeton/ If you watch the Clack window while the page loads, you should see some of the virtual network links flash green as data moves across them. Each flash corresponds to a single packet traversing the network. 6. Tell your neighbor your [www-server] address and ask him/her to open this address in their browser. Report what you see. (We emphasize that this IP address is 100% real; feel free to call up mom on the cell phone and ask her to open this address in her browser. Do you see her request in your virtual world?)7. The web page you loaded from http://[www-server]/ princeton/ contains hyperlinks to several large files. Download one of these files such as “Dan’s Magnum Opus”. (Right-click on the link, select “Save target as…”, and save the file anywhere.) Watch the download window for a minute, focusing on the transfer rate in kilobytes per second (kb/s). How does that number change? Stop the download and go on to Part 2, which will give you a glimpse into what is really going on. Part 2: Data flow and congestion1. Return to the web page at http://[www-server]/princeton/ and click again on the link labeled “Dan’s Magnum Opus”. You will use Clack to observe TCP’s congestion control behavior while this file is downloading.2. In the Clack window, double-click the blue icon labeled “vrhost” to enter the detailed view of the virtual world shown below:COS 116 – Lab 93This screen shows a detailed view of how data moves through the simulated network. The window has more information than we will need, so ignore most of it. The important thing is that there are three entities attached to the network—a web server (eth1), an ftp server (eth2), and the
View Full Document
Unlocking...