May 2004

Net cartography
Does it take rocket science to build an eye-catching Internet map?
by Salman Siddiqui
Currently the world’s population of more than six billion is increasing by four births per second. Likewise, online population has grown steadily and is projected to stand at 945 million for 2004; Pakistan claims 1.2 million surfers among them. In addition, the Internet Software Consortium indicates that the number of hosts is up to almost 250 million, where each host might represent either a small network or a cluster of thousands. Cyberspace is expanding, this much is obvious.But while satellite images have been used to gauge the demographics and geography of the world, what do we do if we need a visual map of the Internet? This is where different mapping projects come into the picture. Talented network cartographers have begun to use specialized software to illustrate the physical underlying of the Internet.
Trace routeMost mapping projects make use of a tool called ‘traceroute’ to determine the path between the machine that runs the tracing software and the destination host. This utility sends a data packet containing a Time-To-Live (TTL) numeric field that decreases each time it passes through a computer on the way towards its destination network. When this field is set to zero, the packet dies and an error message containing the IP address of that computer is sent back to the source. So for example if TTL is set to 3 then the error message would indicate the IP address of the third computer. Therefore, by varying this numeral, one can find out the addresses of all the hosts in a network. Later these addresses are identified using Domain Name Servers (DNS) and other Internet databases.You can see this tool in action on your Windows machine by going to Start->All Programs->Accessories->Command Prompt. Here, type: tracert http://www.spider.tm/ and press ‘Enter’. This should show a table containing a list of routers between your PC and the web server that hosts the SPIDER website. Once this data is obtained, maps can be drawn using diverse graphic engines.
The Internet Mapping ProjectSince 1998, research scientists Bill Cheswick and Hal Burch have been maintaining a comprehensive topological database using traceroute to map the Internet.But instead of running a daily scan of the entire Internet and overloading it with intrusive requests, the Lumeta Corporation’s Internet Mapping Project (IMP) scans 10% of its list of networks each day and the entire list on the first of each month. It randomly selects a target host via a network protocol called Borderless Gateway Protocol (BGP), or as listed in various Internet databases. Moreover, the color-coded visualization techniques employed here have nothing to do with the geographical boundaries. Besides drawing attention from art enthusiasts in the late 90s, IMP has found commercial applications for the scanning of large corporate networks as well. The inspections help a corporation to identify the vulnerable areas of their networks. In figure A, a corporate network is shown for which the trace from IMP clearly indicates the presence of an unauthorized path between the corporate network and the Internet. The project has also been successful at portraying infrastructural damage caused due to natural disasters or calamities. For example, during the civil war in Bosnia-Herzegovina in 1999, the IMP successfully recorded the disappearance of the country’s networks.
The Opte ProjectOnly six months ago, Opte revived the buzz associated with Internet mapping by releasing some dazzling pictures of the Internet. The colorful maps of the Net looked more like the Milky Way than images created from a 3D graphics program used for modeling proteins. Inspired by the Internet Mapping Project, Barrett Lyon, a 26-year old network engineer from California had commenced this project as a challenge to churn out complete maps of the Internet every day instead of waiting for six months. Although it uses the same utility, traceroute, Opte traverses only smaller network level addresses that aren’t reserved, thus reducing its address space to only about 13 million IP blocks. With the code capable of scanning 512 million Internet addresses per second, Barrett claims that the scanning of the entire Internet takes only about 10 hours and another hour or two for the visual map output.
Are the maps complete?The maps generated from both of the above-mentioned projects render only a portion of the Internet, since a number of web hosts may be unapproachable or inaccessible due to the placement of firewalls. Both mapping techniques also use only a single computer and Internet connection, resulting in maps that use the mapper as the reference point for all the other points on the Internet. In other words, this is equivalent to designing a geographic map where Karachi is the reference point for every city worldwide, with only connections between Karachi and other cities defined, but no creation of links that join the several cities independent of Karachi. An all-inclusive map can be developed only if mapping is done from multiple hosts. Although nothing conclusive has been achieved at Opte’s end yet, the project has plans to develop such a map by taking a distributed approach to the problem, whereas IMP has started to use more than one host for their scan results only recently.
Projects @ largeAlthough mapping projects have nothing to do with the physical geological boundaries of the world, approximate diagrams of the Net can be developed using advanced algorithmic techniques to map IP addresses to geographic locations. John S. Quaterman has is a leading authority on the subject. CAIDA is very similar to the Internet Mapping project and it too daily maps the Internet as part of its Skitter project. Skitter focuses more on the paths to Web and DNS. Other projects, such as Mercator, attempt to discover networks with limited pre-knowledge and RocketFuel builds a very accurate map of a particular ISP’s network. Furthermore one may attempt to map the fiber optic infrastructure of the networks instead of only the routers in a network. The Department of Homeland Security in the USA got a wakeup call after 9/11 when a graduate student mapped out the nation’s entire information technology network as part of his PhD dissertation. The map apparently illustrated the manner in which different government organizations and bodies including nuclear power plants were connected to their Intranets and to the Internet.So what good are these maps to us at the end of the day? Well, the application of maps depends on the way you want to use them. You could either set one as the wallpaper on your computer desktop, just like you stick a real world map on the wall to decorate your room, or you could use the map for constructive applications, such as receiving early warnings of congested areas, market opportunities and Net growth in the same way you might use a geographic map to navigate your ways in an unknown country. If you’re thinking of giving your study that extra geeky touch then order a customized Internet map poster from Peacock.com. Or you can keep that copy of the world map from Urdu Bazaar handy too.
Sites:
Atlas of Cyberspace
www.geog.ucl.ac.uk/casa/martin/atlas/geographic.html
CAIDA
www.caida.org
Cyber Geography
www.cybergeography.org/
DNS
www.dns.net/dnsrd
Internet Mapping Project's daily Database
research.lumeta.com/ches/dbInternet
Maps at Peacockmaps.com
www.peacockmaps.com
Internet Software Consortium
www.isc.org
John S. Quatermanorder.mids.org/~jsq/index.htmlLumeta Corporation
www.lumeta.com/mapping.html
Mercator
www.mercator.com
Opte
www.opte.org
RocketFue
l www.rocketfuel.com
Skitter
www.caida.org/tools/measurement/skitter
Traceroute
www.traceroute.org