Posted by William Gill on 09/28/07 18:42

cwdjrxyz wrote:

>
> I get up to a maximum number of GB of bandwidth per month. If that is
> exceeded, I will get charged for the excess bandwidth at the end of
> the month. ...

Bandwidth is an old analog term. In general it refers to how much
spectrum is used/available, and thus the capacity. The term did not
port well to the digital world, because its use introduced too much
ambiguity. i.e. 1000 Khz means 1,000,000 full cycles every second.
Seconds are the agreed to standard interval or period. You won't see
cycles per week, cycles per hour, or anything else. When a hosting
service uses "bandwidth" to mean how many gigabits per month, he/she may
be using a "pipe" that only works at 45 megabits per second to connect
to the internet at large. That "pipe" is shared by all his/her
customers. When he/she tries to feed more bits than that, the excess is
queued or dropped.

> ... They download part of the data at a time, The time between
> download "spurts" becomes longer and longer as more and more people
> are using the server.

Well... Sometimes the "spurts" are the result of requests being queued
or the bursty "packet" nature of TCP/IP ("data" is broken up; put into
packets numbered; sent; received; resequenced if necessary; and
reassembled).

> ... This situation will not do for a busy streaming media
> site, such as a web radio or TV station. A certain minimum download
> rate is required to keep the media streaming. Thus a special media
> server often is set up to limit the number of people viewing it at one
> time. When that limit is exceeded, no one else can get on until some
> people sign out. Typically when too busy, you get a message that the
> service is too busy to use.
>

The media is always streaming. The question is; is it streaming
efficiently enough.

To prevent the interval between packets in any source/destination stream
from being intolerable to a specific "service", it is necessary to have
faster "pipes", faster "pumps" (servers), and/or fewer recipients. Also
keep in mind that the "last mile" (the part between the net and the
recipient) is probably the biggest choke point (the slowest). In the
old telephone world echo was not a problem on shorter connections,
because even if the echo was "relatively loud", it was happening almost
in sync with what it was echoing, so the brain didn't notice. On longer
calls the delay made the echo separate itself in time from the original
(delay). We couldn't fix the time delay, but we could suppress the
volume of the echo in an effort to keep the brain happy. A
non-technical example is a Casino. The dealer(server) can only deal
cards (packets) so fast, the table can only be so small (faster
delivery), but some players won't be happy if they are losing their
money too slowly. So we limit the number of seats at the table (limit
the delay between cards/packets) so that the dealer can get that unhappy
player his/her next card faster.

<disclaimer> all of this may not hold up to intense technical scrutiny,
but it's close enough for government work. Anyway, I think it's way
beyond the scope of this group</disclaimer>

[Back to original message]