Re: (Off Topic) Re: FCC_ups
At 02:34 AM 8/12/96 -0400, Rabid Wombat wrote:
Anyone that wants to carry a large volume of traffic via the 'net will find that either the market will dictate that they pay for the bandwidth they use, or the FCC will. I don't see the FCC getting involved, unless the "phone service via internet" providor tries to use the courts to get out of paying for the bandwidth they use.
Bandwidth costs almost nothing, unless you are doing full motion video. What is expensive is cutting that bandwidth up into little pieces and delivering those pieces to the people who want to use it at the time that they want to use it. Thus those who retail bandwidth will have the bulk of the revenue, rather than those who wholesale it. --------------------------------------------------------------------- | We have the right to defend ourselves | http://www.jim.com/jamesd/ and our property, because of the kind | of animals that we are. True law | James A. Donald derives from this right, not from the | arbitrary power of the state. | jamesd@echeque.com
On Sun, 11 Aug 1996, James A. Donald wrote:
At 02:34 AM 8/12/96 -0400, Rabid Wombat wrote:
Anyone that wants to carry a large volume of traffic via the 'net will find that either the market will dictate that they pay for the bandwidth they use, or the FCC will. I don't see the FCC getting involved, unless the "phone service via internet" providor tries to use the courts to get out of paying for the bandwidth they use.
Bandwidth costs almost nothing, unless you are doing full motion video.
I disagree - the cost doesn't stop w/ the cost of the circuit itself - installation, CPE, maintainance, management, customer service (re the purchaser of the bandwidth, not the "voice customer") all contribute to the cost. None of these costs are realy "fixed" - all are fixed+marginal, so the costs increase as bandwidth is added. Also, I've worked with a number of large sites, as well as a large ISP - the amount of RealAudio traffic is amazing.
What is expensive is cutting that bandwidth up into little pieces and delivering those pieces to the people who want to use it at the time that they want to use it.
Yes, but having enough bandwidth to service your customer base means having enough bandwidth to service peak periods of demand - otherwise, customer satisfaction will be low. What is your point? If x bandwidth is sufficient during "off-hours", but 6x bandwidth is needed for peak periods, then either the voice providor leases 6x bandwidth, or a higher-tier providor must provide the capacity for 6x bandwidth, and lease "bandwidth on demand" to the voice providor. The infrastructure must still support 6x bandwidth, with accompanying costs. Do you think "BigTelco" will lease bandwidth on demand without charging more for high levels of use during peak hours? Not on a large scale. This reminds me of the old arguements that "bandwidth costs would come down so fast that the Internet would be essentially free." Haven't seen that happen yet, either.
Thus those who retail bandwidth will have the bulk of the revenue, rather than those who wholesale it.
The breakup and the addition of competition and second-tier voice providors hasn't killed AT&T yet. If anything does, it will be bloated corporate structure, not 2nd tier competition. (They're working on that - funny how years ago we spent gobs of tax dollars breaking up AT&T, and now they've gone and done it to themselves voluntarily). Most second-tier providors differentiate themselves through marketing and/or customer service. Customer service is a big expensive pain in the butt. Many larger companies out-source it anyway, when they can get away with it. Larger companies will still retail large accounts, and leave the smaller, less profitable crumbs to the niche marketers. I still see voice over the Internet as a hobby - nobody is going to spend $800 for a telephone, er "voice terminal." If I decide to, who will I call? Only others with similar equipment? That's not very functional. Yes, this new carrier could implement equipment in each CO to convert my call to POTS systems, so I can call my computerless Grandmama from my PC - what will this do to costs? Most of corporate America is still grappling with learning to use email. I don't see the business customer buying into voice services via PC anytime soon. The equipment costs per user are too high. I started working with a whole slew of telephony and convergence products years ago - they've been very slow to catch on. Yes, one day voice, video, and data will all be carried on the same infrastructure - using ATM, or a similar technology. The convergence is inevitable. Large corporate phone switches will communicate with the outside world via packet switched, rather than circuit switched networks. This isn't the same as the "save money on phone calls by using cheap Internet bandwidth and your PC" sales pitch, however. ob crypto (for anyone who read this far): When packet switched voice systems become a reality, how can secure calls be placed to any number? Key exchange during call set-up? How long will this make the call set-up? Ideas? -r.w.
In article <Pine.BSF.3.91.960812153240.26183B-100000@mcfeely.bsfs.org>, Rabid Wombat <wombat@mcfeely.bsfs.org> wrote:
ob crypto (for anyone who read this far): When packet switched voice systems become a reality, how can secure calls be placed to any number? Key exchange during call set-up? How long will this make the call set-up?
This is easy. Just use end-to-end encryption. No sweat. (So what if call setup takes a half a second to do a public key encryption? The phone rings for a couple of seconds before the other guy picks it up anyhow.) Well, there's that nasty key distribution and management problem (e.g. who certifies the millions of public keys corresponding to everyone's phone number?), but that's not specific to voice traffic, and this is a well-known annoying problem. The *real* challenge: how do you support sender- and recipient- anonymous phone calls with strong security? Have fun.
David Wagner wrote: | The *real* challenge: how do you support sender- and recipient- anonymous | phone calls with strong security? Have fun. Caller calls 1-900-stopper via an international callback service. Caller uses Stopper to reach callee's phone number. Callee, taking responsibility for their own privacy, uses a forward that she placed on a pay phone in Grand Central to a cheese box* in the Seychelles to her real phone. Oh, you want authentication and MITM protection? Only caller<-->callee needs authentication, for the DH key that they share for the call. The other encryption is point to point transport layer stuff; its nice that its there, but a MITM can listen in, and only get one or two phone #s. The chain is as strong as its strongest link, namely the photuris style authentication of the caller<->callee. (A cheese box is a forwarder that works outside of the switch; call #1, it dials #2, then connects it to line 1. So called because the first one the police found was in a cheese box.) Adam -- "It is seldom that liberty of any kind is lost all at once." -Hume
David Wagner wrote:
The *real* challenge: how do you support sender- and recipient- anonymous phone calls with strong security? Have fun.
If you're only encrypting/decrypting at each end, couldn't a key exchange like Diffie-Hellman work? Or is this not the "anonymous" feature you were looking at? -- Wyntermute
participants (5)
-
Adam Shostack -
daw＠cs.berkeley.edu -
James A. Donald -
Justin Card -
Rabid Wombat