Let's assume I have a connection between a machine at a datacenter in France and a datacenter in Singapore and the RTT between the two is around 270 ms according to ping.

If I do a traceroute from the machine in France to the machine in Singapore, I see this as part of the traceroute output:

 8  ldn-bb2-link.telia.net (  3.888 ms  ae-2-3201.ear1.london1.level3.net (  4.109 ms  4.174 ms
 9  ldn-b5-link.telia.net (  3.977 ms ix-ae-20-0.tcore1.ldn-london.as6453.net (  4.150 ms  4.132 ms
10  tata-ic-300410-ldn-b5.c.telia.net (  3.967 ms  4.106 ms  4.084 ms
11  if-ae-15-2.tcore2.l78-london.as6453.net (  265.990 ms  266.042 ms  266.074 ms
12  if-ae-9-2.tcore2.wyn-marseille.as6453.net (  262.892 ms  263.072 ms  263.093 ms
13  if-ae-2-2.tcore1.wyn-marseille.as6453.net (  268.949 ms  269.093 ms  268.861 ms
14  if-ae-5-2.tcore1.mlv-mumbai.as6453.net (  264.870 ms if-ae-5-2.tcore1.mlv-mumbai.as6453.net (  271.598 ms  264.201 ms
15  if-ae-2-2.tcore2.mlv-mumbai.as6453.net (  263.022 ms  262.824 ms  262.757 ms
16  if-ina-2.tcore2.svw-singapore.as6453.net (  266.572 ms  261.399 ms  268.869 ms

Assuming that the city names in the PTR host entries match the geolocation, the packets would travel that way around the world (Note: In the traceroute output I cut the France-> Great Britain part off.):

France -> Great Britain -> France -> India -> Singapore

But then the RTT's clearly conflict with that assumption. One would assume a high RTT between Europe and Asia, but not between 2 peering networks in the same city (see hop 10 and 11).

So the conclusion seems to be that any "geolocation" info in the PTR record is basically useless? Also any whois information on IP geolocation seem to be useless, because if you do a whois query on the hops 11-15 you will get these country information:

11 (if-ae-15-2.tcore2.l78-london.as6453.net) -> "country: Great Britain"
12 (if-ae-9-2.tcore2.wyn-marseille.as6453.net) -> "country: France"
13 (if-ae-2-2.tcore1.wyn-marseille.as6453.net) -> "country: France"
14 (if-ae-5-2.tcore1.MLV-Mumbai.as6453.net)-> "country: France"
15 (if-ae-2-2.tcore2.MLV-Mumbai.as6453.net) -> "country: India"

As one can see even these are inconsistent (e.g. hop 14 - PTR record "Mumbai" vs whois record "France").

So my question is: Is the naming just arbitrary? Have I overlooked anything or is there really no connection between geolocation and router DNS records and even whois information?

  • Sadly, questions about networks that you don't operate are off-topic here.
    – Ron Trunk
    Apr 12 '17 at 11:28
  • Also, questions about protocols above OSI layer-4, e.g. DNS, are explicitly off-topic here.
    – Ron Maupin
    Apr 12 '17 at 13:16
  • I'm sorry. Before I asked this questions, I had used google in order to find the right place where I can ask my question. I found this . Based on that I thought this is the right place. To me my questions is not DNS specific. The background of my question is to get a better understanding of WAN. But if this is the wrong place, please move my questions to the right place.
    – 0x80
    Apr 12 '17 at 13:54

The thing is that you have an MPLS network in the middle of your traceroute. Actually results from traceroute are not really accurate when it comes to MPLS environment. What happens is that all ICMP TTL exceeded messages originated by a P device first have to be tunneled to the other edge using the original label stack before being sent back to the originator of traceroute operation. Therefore, you will always see end-to-end RTT across the provider core in the output. You need to trace from PE router toward a provider-scoped IP address to resolve this "layering" problem and observe proper RTT.

NOTE: DNS resolving is correctly configured.

Not the answer you're looking for? Browse other questions tagged or ask your own question.