1

Is it possible to use ping or tracert (Networking tools) to approximate any of the following:

  • Propagation delay to ISP
  • Distance to ISP
  • Link bit rate
  • 2
    ICMP is usually low priority traffic which is queued or dropped in favor of "real" traffic. The performance of ICMP really doesn't reflect how real traffic performs, and many ISPs will redirect ICMP traffic in order to disguise their internal networks. The path traceroute takes may not reflect anything like the path your other traffic takes. It can even be directed to secondary transit links which you other traffic only takes in the event of a failure. – Ron Maupin May 6 '16 at 1:05
  • @RonMaupin I'm sure that that is not the answer his teacher is expecting :) – Law29 May 6 '16 at 7:09
  • Probably not. Maybe it gets extra credit for how it really works. An interesting thing is that the traceroute RFC was never on the Standards track, and it was always Experimental status, but it has now been deprecated to Historic status because the ISPs never bought into it. – Ron Maupin May 6 '16 at 7:21
  • @Law29 Yes it's a question for an undergraduate project. All my attempts on my ISP resulted in unexpected values. That's why i asked if it's even possible as a crude approximation. – usfmohy May 6 '16 at 10:25
  • @RonMaupin I found alot of data missing like the size of udp packets, routers processing/queuing time and transmission speed. Also, are both times a packet take to reach a node and be sent back equal? – usfmohy May 6 '16 at 10:38
0

I found alot of data missing like the size of udp packets, routers processing/queuing time and transmission speed. Also, are both times a packet take to reach a node and be sent back equal?

First, ping uses ICMP, not UDP, and you specify the packet size when you run the command, or the OS may have a fixed packet size. It probably defaults to 64 bytes.

Which OS you are using dictates which type of traceroute you use (e.g. ICMP, UDP, TCP), and whether or not it uses options, all of which affects the size of the packets.

The router queuing and processing time will vary, based on the router model, how it is configured, and how busy the router link is, which may vary by day of the week, time of day, etc. For instance, a shared medium, like a cable modem, may get very busy in the evening when people come home, while a business ISP router is probably busiest during working hours.

I assume you mean bandwidth, rather than the transmission rate (which is well known for electrons traveling through copper wire, or light through fiber). The serialization rate (bandwidth) will vary by hop, and it is more important than the actual transmission speed. Also, the layer-2 protocol on each link has an affect. You will have headers for the layer-2 frames, and the size of those vary by the layer-2 protocol used on a link. For instance, ethernet will add 26 to 30 octets for the frame, and it has a 12-octet gap between frames. If a link has a half-duplex medium, you can be delayed because the medium is in use by other hosts.

ICMP is usually low priority traffic which is queued or dropped in favor of "real" traffic. The performance of ICMP really doesn't reflect how real traffic performs, and many ISPs will redirect ICMP traffic in order to disguise their internal networks.

The path traceroute takes may not reflect anything like the path your other traffic takes. It can even be directed to secondary transit links which your other traffic only takes in the event of a failure.

The time it takes a packet to reach the other end can be different than the time the return packet takes to reach you, and the time it takes to process the packet at the far end and generate a return packet can vary. For instance, ping gives your a round-trip time which is cumulative of the time the packets takes each way and the processing time.

All in all, using ping and traceroute for measurements is crude, fraught with problems, and may not reflect how your other traffic performs. It's even possible that the delay for these in your own network (short physical distance) is greater than the delay from your network perimeter to the ISP perimeter (longer distance).

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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