Re: IP: Pac Bell says Net use may collapse phone system

[Dave Reid <dreid3@ent.agt.ab.ca>]

  >Maybe this would have been true 30 years ago, but the consumption of resources
  >required to attempt a call failed on busy is fairly minimal.  The main problem
  >is that the caller is off-hook demanding attention.  However, the inter-switch
  >communication needed to fail the call is trivial, since the circuit is not set
  >up until and unless the call succeeds.  That "busy signal" you hear is locally
  >generated, not piped down from the destination switch.
  
  I wouldn't speak about the AT&T #5 but in the DMS100 world you most certainly do
  consume important resources for non-connected calls. In DMS the incoming line
  is connected onto a line concentrator that may have upwards of 600 lines on it.
  This line concentrator is in turn connected to a line controller via a number of
  links. Typically there maybe an average of about 100 channels between the
  devices.
  To generate a busy tone back to the susbscriber a channel is held up to the line
  controller which is where the tone is generated. When the subsriber hangs up the
  channel is released. This also applies for most phases of the call such as
  digit collection etc. This supply of channels between the line concentrator
  and the line controller is where most of the congestion occurs. After the call
  leaves the line controller and enters the switching fabric your are into a
  pretty much nonblocking situation within the switch.