- From: <D.White@mcs.surrey.ac.uk>
- Date: Wed, 12 Mar 1997 21:23:21 +0000
- To: www-talk@w3.org
- cc: D.White@mcs.surrey.ac.uk
Hi all, Does anyone have any ideas on how to find country or geographical info or timezone info for an arbitrary domain name or IP address? I'm trying to see if it's possible to perform a timezone-based analysis of our Web server logs, to see the distribution of accesses by timezone (and hence country). The Web server logs record either the IP address or the FQDN of the requesting machine. I've already assumed that I can look at FQDNs with country codes at the end, and use that to determine which country the machine's in. First question: does anyone know where to find the official country code listings, and how to relate these to timezone, ie. GMT+n? However, many FQDNs fall into the categories of .com, .edu, .gov etc, and unfortunately these do not necessarily imply being in the USA. Many UK companies have registered themselves in the .com domain rather than the .co.uk domain (and I'm sure the same is true of many other domains). Also, many machines don't even HAVE FQDNs at all - just IP addresses. So, I find myself in need of, at base, a mechanism to identify the timezone from an IP address. Now, I know this may not always be possible in the extreme case of a portable computer moving between timezones, but that is a pretty extreme case! I've been discussing this with my colleagues, and we've come up with the following possible approaches. None are guaranteed to work in all cases, but together they might make an impact on all the .com, .edu and numeric IP addresses in typical Web access logs! 1). make a daytime or time service request via telnet or similar libraries (eg expect or Perl Net:: libraries) Problem: This would only work on machines running a time service. Mainly only Unix machines, we suspect! We further suspect that most Unix machines would be DNS registered.. Still, this might work to prove that many .com and .edu machines were really in the USA. 2). try locating a nearby mail server using MX records in the DNS, talk to it and see if it puts the current time of day into the first 220 response (many do). Problem: this will only work if we can find a nearby mail server - this may help with many .com and .edu addresses. But how do you find a machine nearby to an IP address? 3). Try replacing the last octet of the IP address with special values such as 1, 2, 3.. and ping each one and then try approaches (1) and (2) on them! Problem: totally hit and miss. 4). Use a tool like traceroute to see where the packets go, and try contacting machines topologically close to the destination, applying techniques (1) and (2) on them.. Problem: most intermediate hardware are routers, not computers! Also, topologically close ain't necessarily physically close! All these approaches are heuristics. Does anyone know any more definitive mechanisms (or better heuristics!)? Or should I forget the whole idea of analyzing the past, and maybe set up a form to ask people (with the problems of self-selection and randomness in sampling that causes)? Unfortunately, the DNS doesn't seem to have any such info - it was my first idea! I've been trawling the InterNIC Web site (http://rs.internic.net/) for several hours now, just in case they maintained a master list of IP addresses (class A, B and C, etc) -> country code/timezone.. cheers, duncan ------------------------------------------------------------------------------ Duncan C. White, Senior Computing Officer, Dept of Maths and Computing Science, University of Surrey, Guildford, Surrey GU2 5XH, UK. Email: D.White@mcs.surrey.ac.uk Phone: +441 483 259632 URL: http://www.mcs.surrey.ac.uk/showstaff?D.White Fax: +441 483 259385 PGPkey: http://www.mcs.surrey.ac.uk/Personal/D.White/pgpkey.html Key fingerprint = 91 93 0D 90 D0 5E 62 BF 57 39 08 56 43 FC E5 C8 ------------------------------------------------------------------------------ "After all, this is a species whose principal means of population control are famine, abortion, a high infant death rate and war." Intervention (page 442) - Julian May ------------------------------------------------------------------------------
Received on Wednesday, 12 March 1997 16:49:27 UTC