tro@uunet.uu.net (Tom Olin) (10/02/90)
In our last episode... I had asked for an explanation of how phone calls were supposedly sliced and diced to achieve statistically based multiplexing (although not in those exact words). I posed the hypothetical situation of a fully loaded network doing 50% duty cycle on each call, then having all callers jump to 100% duty cycle. I wanted to know what would happen to the network and to the individual calls. Everybody who responded to my question said the same thing: The current phone system (with a very few exceptions, maybe) does not packet switch. It uses time division multiplexing only. Every call is allotted a constant bandwidth, whether the callers are silent or screaming. THUS, A DATA (MODEM) CALL CURRENTLY USES NO MORE OF THE NETWORK'S BANDWIDTH THAN DOES A VOICE CALL. The respondents also pointed out that techniques for packetizing calls are actively being investigated, so in the future, a modem call might indeed cost more than a voice call. But not yet. Thanks to: Wayne Sung <uunet!mcnc.org!sung> uunet!tabasco.lcs.mit.edu!ath (Andrew Heybey) uunet!turing.cs.rpi.edu!borcherb (Brian Borchers) uunet!ihlpl.att.com!mea (Mark E Anderson) kabra437@pallas.athenanet.com (Ken Abrams) uunet!xavax.COM!alvitar (Phillip Harbison) And thanks to goldstein@carafe.enet.dec.com (Fred R. Goldstein) for his recent posting to c.d.t on the subject. Tom Olin uunet!adiron!tro (315) 738-0600 Ext 638 PAR Technology Corporation * 220 Seneca Turnpike * New Hartford NY 13413-1191