U.S.B. speeds

[Lennart Sorensen]

On Wed, Jun 16, 2010 at 08:47:56AM -0400, colin davidson wrote:
> Every communication protocol has protocol overhead. The more
> connections deviate from a simple point-to-point communication between
> two and only two devices, the heavier that overhead will tend to
> become. As an example, the original ethernet had a physical transfer
> rate of 10 million bits per second, but it was widely held that the
> maximum useful data throughput was around 3 million bits per second.
> Thus data transfer rates are almost always lower than the physical bit
> rates (though in some cases data compression can give a higher data
> transfer rate than physical bit rate, but only for data with fairly
> high redundancy.

Once people got switches (instead of hubs or worse yet coax cables),
the full rate became perfectly easy to achive.

> Protocol overhead isn't the only reason for poor performance, however.
> It can also be the result of poor configuration (settings that don't
> work well with the actual connection), poor topology, inteference,
> signal loss and so on. These all apply to every form of communication,
> though the causes and sources of the problems can vary massively
> depending on the technology.
> 
> If your problems are caused by protocol overhead, there's nothing you
> can do except change your connection technology. If you are
> experiencing significant problems from other causes, however, there
> will often be things you can do about it to gain significant
> performance improvements.
> 
> There, that deals enough with the theoretical. Everyone else can deal
> with the practical :-)

-- 
Len Sorensen
--
The Toronto Linux Users Group.      Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists