| FAQ last modified | Sat Jun 9 16:35:41 2007 GMT |
|---|---|
| Current time | Sun May 19 21:44:35 2013 GMT |
1.0. General Questions
1.1. What is the Sarangworld Traceroute Project?
1.2. How does it work?
1.3. Why don't you use the WHOIS record? Why is this better than the IP to Latitude/Longitude program?
1.4. If the WHOIS record is accurate for 90% of all domains, why bother with pattern matching and code conversion?
1.5. Why traceroute to a given host instead of just converting that host directly to a location?
1.6. What about DNS LOC (RFC 1876)?
1.7. Why not use top-level country codes (like .uk) to determine the location of a machine?
1.8. Is pattern-matching and code conversion the best way to find the locations of Internet-connected machines?
1.9. Whose traceroute servers do you use?
1.10. What if I don't want my server listed?
1.11. What sorts of traceroute servers do you support?
1.12. Are there any other (online or offline) programs that map machines to geographical locations?
2.0. Code Questions
2.1. Where can I see a list of codes you use?
2.2. What does the code [START] mean?
2.3. Why are the codes in the code database biased towards the United States?
2.4. Isn't it possible for two different ISPs to use a single code to mean different things?
3.0. Map Questions
3.1. How is the map created?
3.2. Why isn't the map in GIF format?
3.3. Why do the city names sometimes overlap on the map?
3.4. Why does the map sometimes show lines leading to/from nowhere?
3.5. (OBSOLETE) Why does the map only show locations in the United States?
3.6. What is the projection for the maps?
3.7. Why do the maps draw inaccurate "shortest path" lines?
3.8. What is the status of the world map?
3.9. How is the world map created?
4.0. Technical Questions
4.1. Where can I see a list of the patterns you use?
5.0. How Can I Help?
5.1. How can I help?
5.2. Should I give you the physical location of my machines?
6.0. Miscellaneous Questions
6.1. What are your feelings on privacy issues related to this program?
6.2. What if my question isn't answered here?
6.3. (OBSOLETE) According to the pull-down menu, Sarangworld is located in Ogden, UT, USA. Do you live in Ogden?
7.0. Errors
7.1. When using the program, I sometimes see a listing under code, but there's no city/state/country/latitude/longitude listed?
7.2. When using the program, I sometimes get weird errors.
8.0. Tracking spammers, crackers, and abusers
The Sarangworld Traceroute Project is an attempt to map the path between Internet-connected machines to geographic locations
1.2. How does it work?
As a first step, the program will perform a traceroute from a public traceroute server to any given hostname. By default, the traceroute is performed from a randomly chosen traceroute server to your IP address, but this can be changed. After the traceroute is performed, each hostname is matched against a list of patterns to see if it matches a known pattern. If it does match a known pattern, a "code" value is extracted and this code is then converted to a city, state, country location. Finally, using a database of city latitudes and longitudes, the location of the hostname is mapped.
1.3. Why don't you use the WHOIS record? Why is this better than the IP to Latitude/Longitude program?
The IP to Latitude/Longitude program use the WHOIS record of a hostname to determine location. The WHOIS record yields the location of the administrative office for the organization/person that owns that domain. For local ISPs and many others, the WHOIS record provides an accurate location. However, for large ISPs (such as alter.net), the WHOIS record is necessarily sometimes inaccurate, because Earthlink has routers/machines all over the country. For example, the WHOIS record for alter.net shows they are in Falls Church, VA. However, the machines "132.ATM3-0.XR2.SEA1.ALTER.NET", "198.ATM7-0.XR2.BOS1.ALTER.NET", and "110.ATM7-0.TR2.NYC1.ALTER.NET" are clearly in Seattle, Boston, and New York respectively.
1.4. If the WHOIS record is accurate for 90% of all domains, why bother with pattern matching and code conversion?
The 90-10 rule applies here. Roughly speaking, 90% of the Internet's traffic is carried by 10% of all domains (actually, it's probably more like 99.99%-00.01%, since there are millions of domains and only a few dozen major ISPs). So, while the WHOIS record might be accurate for 90% of all domains, the domains its inaccurate for are actually far more important. You'll notice that in most traceroutes, the same top-level domains (alter.net, earthlink.net and its subsidiaries, verio.net, flash.net, etc) show up over and over again. All of these companies have country-wide and sometimes worldwide prescense, so a single WHOIS location could never be accurate for them.
1.5. Why traceroute to a given host instead of just converting that host directly to a location?
Converting a single host to a location is actually harder than using traceroute to find the location. As an example, if you're dialing up from a local ISP, your IP address maybe "dialup-17.somelocalisp.net". Unless that local ISP is in my list of patterns (which it's usually not), there's no way to convert that address to a location. On the other hand, if a traceroute to your IP address shows the penultimate step as "dca0.iad3.verio.net", it's pretty easy to see that machine is in Washington DC (both DCA and IAD are airport codes for airports in Washington DC). Verio.net is included in my list of patterns because it's a major provider. From this, I can guess you're in Washington DC as well. This isn't always accurate, but it's more accurate than trying to convert just your IP address with no additional information.
1.6. What about DNS LOC (RFC 1876)?
DNS-LOC is an addition to the DNS protocol that lets machines provide their physical location (latitude and longitude) as part of the DNS record. If and when it becomes widely supported, there will no longer be a need for my project. However, at the moment, few machines support this protocol.
1.7. Why not use top-level country codes (like .uk) to determine the location of a machine?
Actaully, country codes are no longer a good indication of the physical location of a machine in many cases. As an example, Register.Com allows people worldwide to register domains with country codes .uk (United Kingdom), .ro (Romania), .ms (Montserrat), .il (Israel), .sh (St. Helena), .gs (South Georgia), .vg (British Virgin Islands), .nz (New Zealand), .fm (Micronesia), .to (Tonga), .tf, .as (American Samoa), .dk (Denmark), .lt (Lithuania), .kz (Kazakhstan), .tc (Turks & Cairn Islands), .md (Moldova), .ly (Libya), .ac (Ascension Island), .za (South Africa), .ky (Cayman Islands), and .ws (Samoa), without regard to where the machines hosting these domains will be located. Note that I'm not affiliated with register.com in anyway and link to them for example purposes only. I don't even register my domains through them (GoDaddy.Com is cheaper, though I'm not affiliated with them either!).
1.8. Is pattern-matching and code conversion the best way to find the locations of Internet-connected machines?
Probably not -- I originally started the project when I noticed several machine names had "obvious" cities embedded in them. I tried to find a quick and dirty way to convert these machine names to locations, and this is what I came up with. The method I use turns out to be remarkably easy-to-use (for example "DFW" means Dallas, Texas, USA, regardless of which ISP uses it-- so I don't need a seperate set of entries for each ISP), but is probably non-optimal in other ways. I'd be interested in hearing about other ways to do this.
1.9. Whose traceroute servers do you use?
I use publically available traceroute servers only. I started with the Boardwatch Traceroute Server Index (no longer available online), and omitted those servers that were down or would only traceroute to a specific IP address. I also omitted those servers that didn't support the ISINDEX/GET protocol. Currently, my program only supports traceroute servers that can be accessed using a URL followed by a question mark followed by a hostname/IP address. In other words, "http://some.server.net/somepathname.ext?TARGET_HOST_OR_IP". As of 11/22/2002, I created a small "proxy" program to access traceroute servers that don't use the ISINDEX/GET protocol, but this is currently experimental.
1.10. What if I don't want my server listed?
If you would prefer not to be part of this project, please send me an email.
1.11. What sorts of traceroute servers do you support?
Currently, only those traceroute servers supporting the ISINDEX/GET protocol are supported. I access "SERVER_URL?TARGET_HOST_OR_IP" to perform the traceroute. As of 11/22/2002, I created a "proxy gateway" to support other types of traceroute servers, but this is currently experimental.
1.12. Are there any other (online or offline) programs that map machines to geographical locations?
There are several. Martin Dodge maintains an excellent list at http://www.cybergeography.org/atlas/routes.html.
2.2. What does the code [START] mean?
This is a dummy code used to indicate the point where the traceroute starts. Since I already know the locations of the traceroute servers themselves, the first location displayed/mapped is taken directly from the location of the traceroute server and not from the code.
2.3. Why are the codes in the code database biased towards the United States?
This project started out as an attempt to map Internet hosts in the United States only. Now that it's been expanded, I hope to add more international codes as well.
2.4. Isn't it possible for two different ISPs to use a single code to mean different things?
Yes. This isn't a problem for codes like "LAX" which almost universally will mean Los Angeles, CA, USA, but a code like "POR" might mean Portland, OR, USA for one ISP and Portland, ME, USA for another-- and these two Portlands are 3000+ miles apart. I haven't run into this situation so far, but, if I do, I'll alter the list of patterns to separate these cases (eg, "POR" found in an alter.net host could be recodified as "POR.alter.net" and "POR" found in a local-maine.net host could be recofidied as "POR.local-maine.net", so the two codes would no longer be the same). As of November, 2002, I have run into this situation, and the top few lines of the patterns file now compensate for this.
The blank map of the United States is a static image which I originally downloaded from the U.S. Census Bureau TIGER mapping service. The city dots and names are then added using a libgd-extension to PHP3.
3.2. Why isn't the map in GIF format?
The map is in PNG (portable network graphics) format, because our version of PHP3 doesn't have GIF support installed for copyright reasons. I support open-source software, and prefer using PNG anyway. If you can't view PNG files, click on the checkbox marked "Use JPEG format?" to see the map in JPEG format. Because JPEGs aren't as good for line drawing (such as maps), the map appears a bit fuzzier in JPEG mode.
3.3. Why do the city names sometimes overlap on the map?
The placement of city names is done in a very simplistic manner-- the city name is place to the "southeast" of the dot representing the city. If two cities are nearby, their names will overlap (it might be possible to work around this, but I haven't explored this). This problem is especially prevelant near the East and West Coast, since the US East Coast has several large cities packed together tightly, and both the East and West coast have major hubs of Internet activity -- Mae-West in San Jose, California and Mae-East in Washington, DC.
3.4. Why does the map sometimes show lines leading to/from nowhere?
Since the map only represents the US, locations outside the US are not displayed. However, lines leading TO/FROM those locations ARE still displayed. So, the lines leading to/from nowhere are actually going to/from locations that aren't displayed on the map.
3.5. (OBSOLETE) Why does the map only show locations in the United States?
I originally started the project to map Internet machines in the United States. Since then, I've decided to expand the project to be worldwide, but I haven't had a chance to update the map. Using a world map for each and every traceroute probably isn't a good idea (especially considering how scrunched together some cities are already), so I'm trying to find a good compromise.
3.6. What is the projection for the maps?
The world map is projected rectangularly, so each degree of latitude and longitude takes up exactly the same number of pixels (4.444 pixels/degree to be specific). The US map is projected elliptically, which looks better when mapping smaller areas. I considered projecting the world map elliptically as well, but it didn't look as good.
3.7. Why do the maps draw inaccurate "shortest path" lines?
The shortest distance between two points on Earth (remaining on or above the surface of the planet) is a "great circle", which is not a straight line on either elliptical or rectangular projections. This is less noticeable on the US map (since it's a smaller area), but can be quite noticeable on the world map. In particular, if you fly from Albuquerque, NM to Tokyo, Japan, you fly northwest in real life, but, on the map, Tokyo appears far to the east of Albuquerque. Similarly, Australia is actually west of California, not east as the map shows.
3.8. What is the status of the world map?
The world map is in experimental/beta status. I realize that it's uglier than the US map, and that it shows the whole world even when the traceroute is limited to only a small portion of the world. It would be neat to have a world map drawn dynamically that only shows the extent of the traceroute, but I haven't gotten that working yet.
3.9. How is the world map created?
I downloaded a static image of the world from the XEROX PARC MapWeb Server. The city names are added the same as for the US map.
There are several ways you can help:
5.2. Should I give you the physical location of my machines?
If you work for an ISP that carries traffic to many locations countrywide or worldwide, then yes please! On the other hand, if you work for a small ISP with all machines in one city, this information wouldn't be as useful to me-- in this case, most likely, all of your traffic comes through a major ISP router and it would be the location/translation of the major ISP router that I would want.
I'm a strong advocate of privacy and anonymity. I realize not everyone wants to announce their physical location. At one point, I even asked flash.net to change their machine names so as not to reveal the physical locations of their dialup users (they never answered my query). I believe this project is interesting because it automatically translates the information already provided by ISPs, but it's not invasive and doesn't seek to extract additional information. If it came to a choice between continuing this project or convincing ISPs to stop providing location information on their users, I would choose the latter.
6.2. What if my question isn't answered here?
Please contact me via email and I'll see what I can do.
6.3. (OBSOLETE) According to the pull-down menu, Sarangworld is located in Ogden, UT, USA. Do you live in Ogden?
No, I live in Albuquerque, NM, USA. However, my Internet provider (webpipe.net) is located in Ogden.
The pattern-matching will sometimes end up extracting a code for which the code database doesn't have a translation (a "bogus code"). Whenever this happens, I'm alerted by email and will try to find a code-to-city conversion. If you see this happening and know of a code conversion, please send me an email to let me know.
7.2. When using the program, I sometimes get weird errors.
The Internet works very well, but not perfectly-- sometimes servers are slow or down, and the program will be unable to perform a traceroute. The best strategy is to try again later. However, if you run into a problem that can be repeated consistently, let me know about it.
If you need to trace someone for legal reasons and/or want to report a network abuser, my program will not be of much help. At best, it will show you the city (not street address) of the abuser... and many times my program's inaccurate enough that it won't even provide that information. For legal and abuse tracing purposes, please use the ARIN WHOIS Database Search.
Featured links:
Please do not email this address: 6b1481d4-519947c3@st.sitesuck.com
8.1. Can I use your program to trace spammers, crackers, and other network abusers? 1. General Questions
2. Code Questions
3. Map Questions
4. Technical Questions
5. How Can I Help?
6. Miscellaneous Questions
7. Errors
8. Tracking spammers, crackers, and abusers
Last modified stardate:
Warning: date() [function.date]: It is not safe to rely on the system's timezone settings. You are *required* to use the date.timezone setting or the date_default_timezone_set() function. In case you used any of those methods and you are still getting this warning, you most likely misspelled the timezone identifier. We selected 'America/New_York' for 'EDT/-4.0/DST' instead in /home/larryday/sites/SARANGWORLD/footer.php3 on line 17
20070609.123541