Talk:Measuring network throughput
- Oppose - I understand goodput to be a term of the art which is pretty much unknown outside a small networking community. I believe there is room for both articles. In my opinion, the edits to what was Measuring data throughput have made the article less accessible, and less easy to understand. Goodput also encompasses more than simply the maximum throughput of data (less overheads), but includes packet loss; whereas people attempting to measure throughput are often attempting to determine the maximum possible data throughput under ideal conditions and may not appreciate just how large some overheads may be, before even beginning to consider items like serialisation delay, packet loss and other items. WLD 22:32, 5 November 2006 (UTC)
- Okay. I have withdrawn my own merge suggestion. However, the "measuring network throughput" article needs to be rewritten in many ways. Perhaps it was good once uppon a time. Protocol overhead and kibiByte/kByte confusion only affects the throughput calculation a few percent as is not important. Protocol overhead calculations may be moved into the goodput article, or to a new article on protocol overhead. The focus of the "measuring network throughput" article should be on the key factors why the achieved throughput some times differ very much from the maximum throughput. Examples are TCP flow control and that the network capacity is shared with other users. These factors are described in the throughput article, but may be more pedagogically explained. (limited TCPSome of the details on protocol overhead may be moved to the goodput article. Mange01 08:42, 10 November 2006 (UTC)
The Evaluation Engineering article on how to accurately measure bandwidth is an irrelevant reference I will remove it within 48hrs. Kendirangu 15:14, 21 February 2007 (UTC)
Move detail on prefixes to another article?
I think the section "Kibi, mebi, gibi, tebi, and pebi prefixes", while interesting is not exactly on topic. The section "Nomenclature" neatly summarises kilo vs kibi differences. Would not a link to another article with this additional detail be better? -drd 184.108.40.206 13:29, 22 February 2007 (UTC)
Once I hit the line about hard drive marketing the rest of the article suddenly seemed to veer into increasing irrelevance —Preceding unsigned comment added by 220.127.116.11 (talk) 16:59, 29 April 2009 (UTC)
More facts on actually measuring throughput
Techniques, different methods and their pros and cons, touching subjects such as instantaneous, short term and long term bandwidth estimation. Also missing information about frequency, bursts, and usage scenarios for the measurement results. Which methods are applicable for certain network types?
TCP Window Sizing
In regards to the article stating max TCP Window Size of 65536 bytes: TCP Window sizing has different default sizes for different Operating Systems, and can be tunable on most operating systems. 18.104.22.168 (talk) 22:13, 12 January 2008 (UTC)
Format of definition
I thought the format of the definition was a bit strange
=> Max. Throughput = TCP Window Size / Round-trip time.
Is this really how definitions should be formated on Wikipedia?