[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: ISP phone ratio's and Interoffice trunkage for voice

  At 05:20 PM 7/12/96 -0400, James Love wrote:
  >thanks, i understood some of this.  Now I take this comment to say that 
  >company's measure calling by smaller units, something called 36 CCS's to 
  >an hour, and that they assume a particular  distribution of calling 
  >(Poisson distribution??)..  
  Yes, they assume a Poisson distribution, hence the name of the formula
  (which was actually invented by E.C. Molina).   There are two common 
  measurements of traffic in the US, Erlangs (call hours) and CCS (hundred
  call seconds).  At least it's not like in the UK where they often use
  ODECS (18 seconds of calling, as in "octo-decimal call seconds").
  >  I'm not an engineer, so bear with me.  
  >Now, if an Online caller makes call which are longer in duration than 
  >phone calls (say one hour rather than 6 minutes), what does this mean?  
  >Does this make the online calls more or less expensive (on a per minute 
  >basis, holding total calling constant), than modem calls?  
  On a per minute basis, little impact.  In some cases the apparent impact
  of longer average (but Poissonian) calls might be to make the burstiness
  happen slower, but that's not really an issue here.  I've never actually
  seen this quantified.  Poisson tables (Molina formula) don't even ask call
  duration.  Now when you have a queueing network (not relevant here), the
  average queue duration is expressed as a fraction of average call duration,
  per the Erlang C formula.  But phone switches block, not queue, calls.
  >  Also, what does it mean when you say switch bandwith is cheaper than 
  >trnnkage bandwidth?  Doesn't everything have to go over a switch at some 
  >point?  jamie
  Sure.  But an intra-switch call uses one switch and no trunks.  A trunk
  call uses at least two switches plus at least one trunk.  Traditionally,
  transmission costs more than switching.  But in today's fiber optic
  local networks, the *incremental* cost of transmission is very low too.
  Fred R. Goldstein     k1io    fgoldstein@bbn.com   +1 617 873 3850
  Opinions are mine alone.  Sharing requires permission.