Talk:Bandwidth-delay product

Examples
I believe many of the examples are incorrect. The round trip time should be used instead of the one-way delay. Dividing the TCP windows size by the bandwidth delay product will provide an accurate estimate of the percent efficiency of the link. BruceBarnett (talk) —Preceding undated comment added 18:53, 23 April 2010 (UTC).


 * The examples are indeed incorrect, but not for the reason you state. The math is incorrect, inconsistent with the text, and inconsistent from one example to another, the formatting is poor, and the assumptions are incorrect.  For instance, the RTT from my home computer (2 Mbps DSL line in Norway) to en.wikipedia.org is nowhere near 200 ms—more like 50 ms.  I don't quite understand what is meant by "Server on a long-distance 1 Gbit/s link", either.  DES (talk) 09:57, 13 January 2011 (UTC)
 * ...not to mention that 10 Mbps is hardly "high-bandwidth" these days, and T1 lines are usually terminated at your ISP, not routed over satellites. DES (talk) 10:04, 13 January 2011 (UTC)
 * I've replaced the examples with others based on real-world numbers: one based on my own DSL line, one based on the HSDPA modem in my laptop, and one based on two servers I have in colo at two different ISPs. DES (talk) 10:13, 13 January 2011 (UTC)
 * BDP is bandwidth delay product, not bandwidth latency product. Delay is end-to-end, hence unidirectional. That's why 2*BDP is used for buffer calculations. --113.61.79.131 (talk) 04:45, 4 August 2011 (UTC)


 * RFC 1072 as RS clearly states "The significant parameter is the product of bandwidth (bits per second) and round-trip delay (RTT in seconds); ... ". The other two sources agree on that as well so I've changed the text, also making the lead fit to the examples. Zac67 (talk) 20:20, 10 September 2014 (UTC)