0

Internet is slow today. So, I ran tracert.exe against a nearby server (German edition of Windows XP):

>tracert inka.de

Routenverfolgung zu inka.de [193.197.184.1]  über maximal 30 Abschnitte:

  1     2 ms     2 ms     9 ms  speedport.ip [192.168.0.17]
  2  2685 ms  2508 ms  2375 ms  217.0.117.14
  3  1435 ms  1680 ms  2103 ms  217.0.81.130
  4  1302 ms  1729 ms  1856 ms  f-ed5-i.F.DE.NET.DTAG.DE [217.5.95.22]
  5  1759 ms  1688 ms  2153 ms  ffm-b12-link.telia.net [62.115.12.45]
  6  1542 ms  1072 ms  1172 ms  ffm-bb1-link.telia.net [213.155.136.196]
  7   978 ms   999 ms   967 ms  ffm-b2-link.telia.net [213.155.132.205]
  8   895 ms   836 ms  1072 ms  belwue-ic-130164-ffm-b2.c.telia.net [213.248.88.26]
  9  1136 ms  1191 ms  1366 ms  Karlsruhe-RZ-1-10GE-0-3-0-2.belwue.net [129.143.57.177]
 10  1448 ms  1842 ms  1577 ms  fe0-553.cix.ka-ip.net [129.143.166.162]
 11  1609 ms  1901 ms  1830 ms  tapac.inka.de [193.197.184.1]

Ablaufverfolgung beendet.

Line two tells me that RTT for an ICMP package between me and 217.0.117.14 is more than two seconds on average. However, line seven tells me that RTT between me and ffm-b2-link.telia.net is less than a second, although there are five more hops.

I assumed that RTT increases with each hop, at least on average.

Is that a false assumption? If not, then what could be the explanation that RTT drops considerably for some hops?

Improve this question

1 Answer 1

Reset to default
0

It is very possible that the fowarding devices on the path use some kind of route caching. The first time a pdu for a specific destination arrives at a node, the device must discover which interface to forward the pdu from. When the next traceroute pdu arrives (with a higher TTL) then the node does not need to rediscover which interface to forward the pdu to and can forward the pdu much faster.

It is also possible that a similar thing is happening with Address Resolution Protocol (ARP) - the first time a node needs to forward a frame it must request the next hop via ARP, this is cached so that next time a frame for the same next hop is received it can be forwarded quicker.

Improve this answer
1
  • If caching were to blame, I would expect one (probably the first) RTT in each row to be considerably higher than the other two. However, this is not the case here. The RTTs vary considerably by row, and in each row, they are about the same.
    – feklee
    Nov 29, 2013 at 16:30

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .