Ping with datagram size.

Unanswered Question
Apr 24th, 2010
User Badges:

Hi Netpros,



Please help me to understand the difference between below two working of "Extended ping with size"



1) For Eg,  We having a 1Mb internet connectivity to remote end(10.10.0.2). We are pining (extended ping with repeat size of 500) to remote end from our end router (10.10.0.1) with datagram size of 1000 i.e.


ROUTER#ping 10.10.0.2 repeat 500 size 1000



How much load(in sense of Mb) this datagram size of 1000 actually put on 1Mb remote end connectivity. How about if we can give datagram size of 1500 on same connectivity. Is there any standard design to check "that we can check the particular bandwidth with particular datagram size only" . please share...

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
Giuseppe Larosa Sat, 04/24/2010 - 14:53
User Badges:
  • Super Silver, 17500 points or more
  • Hall of Fame,

    Founding Member

Hello Netbeginner,

be aware that even when we use extended ping we use a 1 packet window:

that is the sender sends out icmp request n, waits up to 2 seconds to receive ICMP reply related to request and only after one of the two events ICMP request n+1 is sent.


The effective loading on the link under test can be found taking in account the statistics about the RTT round trip time of packets.

With no losses:


RTT includes the time for the ICMP request n to be sent (a timestamp is present in the packet) and the time at which the ICMP reply n is received at sender.

You could at the average RTT that is the second time in results


so the load on link can be see as:

(packet size * bits / byte  ) / (RTT/2)


example 1000 bytes with average RTT 22 msec


(1000*8) / (11 10^-3) = 727723 bps


but if there are losses, missing packets count as 2 seconds each and of course load on link is lowered by the fact that sender waits 2 seconds before going to next ICMP request.


Now, if you increase the packet size also the average RTT will increase.


Edit:

more practical approach:

a "poor man" traffic generator, if the objective is to load the link, is made by using very big  packets like 10000 bytes.

These packets cannot travel on the link if the MTU is 1500 bytes, so each ping ICMP request becomes a small burst of fragments of the original packet  that travel from sender to receiver and then back from receiver to sender.

The difference in load is effective


Hope to  help

Giuseppe

George Stefanick Sat, 04/24/2010 - 23:13
User Badges:
  • Purple, 4500 points or more
  • Community Spotlight Award,

    Best Publication, October 2015

Hi Net,


If you are looking to 'test' your bandwidth on your pipe you can also use a free tool called IPERF. Google it. Its a client-server solution.

Actions

This Discussion