Cisco Support Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Announcements

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. If you'd prefer to explore, try our test area to get started. And see here for current known issues.

Basic question

Hi,

Whether there is any formula or method that we can calculate the bandwidth of link based on delay(ms) in ping request ?

d:\> ping 10.1.1.2

Reply from 10.1.1.2: bytes=32 time=302ms TTL=46

Note :  Current the link 10 % usage.

Everyone's tags (3)
2 REPLIES
Cisco Employee

Re: Basic question

Hello,

Using pings to determine the available bandwidth is very unreliable. While it is true that it holds: amount of transferred data = bandwidth x time from which the bandwidth can be computed given the total size of the ping packet and the transfer time, the time given by the ping command is usually the total round trip time that consists of the time necessary for the ECHO REQUEST packet to traverse from source to destination, the time until the destination processes it and sends an ECHO REPLY packet, and the time necessary for this reply to get back. Replying to ping packets is usually a low-priority process on the destination device and may, under circumstances, take several milliseconds. Also, different queueing and processing delays contribute to the overall round-trip time. Therefore, even if approximating the one-way delay as RTT/2, it is a very imprecise approximation and the gathered data may fluctuate wildly. Even if the bandwidth is computed this way, it has a property of an immediate available bandwidth in the moment of the ping, however, because of bursty nature of network traffic, such measurement is not a representative of the average bandwidth that should not vary over time.

The available bandwidth should be measured over a certain period of time using a particular data flow. A simple file download may give you a decent approximation of the available bandwidth, however, it is best to use a flow in which there is no upper-layer flow control - such as UDP, for example (the flow control mechanisms generally prevent congestion at the price of slightly lower bandwidth utilization).

To sum it up - I would personally suggest not to try to infer the bandwidth using pings.

Best regards,

Peter

Re: Basic question

Peter ,

Thanks for yours reply.

Assume the scenario is point to point.  Please suggest on this .

401
Views
4
Helpful
2
Replies