The failure of yet another VOIP (voice over Internet Protocol) company reminds me of a question to which I've never got a satisfactory answer. Is there any technological reason for thinking that VOIP is a good idea? To clarify, IP is a packet-switching technology, which is great for data, but everything I've read suggests that circuit-switching is better for voice (that's what it was designed for, after all).
Whenever I've seen an explanation of the supposed advantages of VOIP, it has seemed to involve regulatory arbitrage. That is the technology is supposed to use unregulated substitutes for regulated voice lines, while typically relying on those same lines for access either via old-style modem or ADSL.
But all of this is just an impression. Can anyone set me straight?
Posted by jquiggin at September 24, 2004 10:30 AM | TrackBackJohn, I share your scepticism towards VOIP. I'm not aware of any technology benefits and it's a poor experience for users, who must co-ordinate their equipment in advance. Boosters say that it reduces call costs, but ho hum. It's pretty firmly in the dot.com camp as far as I'm concerned.
Posted by: Tony Healy at September 24, 2004 11:07 AMJQ,
Just to clarify, Comindico was not a "VoIP" company. Yes, it was offering VoIP services, but it was essentially an alternative telco using an IP network. Like many junior telcos, it burned through too much cash waiting for customers to hook up and provide it with revenue.
I'm not a telecomms engineer, but I'll have a stab at answering your question. Yes, there is a very strong technological reason for thinking that VoIP is a good idea. More importantly, there is a strong economic reason for thinking that VoIP is a good idea.
VoIP only works when your voice message is broken down into information packets, transmitted over an IP network and then re-assembled in the correct order at the right time at the destination. Each sound that you make has to arrive in sequence in real-time. This was historically difficult because (it was thought) you needed a lot of bandwidth to carry the information load, and you need to prioritise the packets so that they arrive at the right time.
Circuit switching provides extremely high reliability and service because it dedicates a single telephone line to a point-to-point communication - guaranteed bandwidth and prioritisation.
However, VoIP is superior to circuit switching in that it is much more efficient in its use of bandwidth. Just as the Internet was intended to use redundant bandwidth by distributing information packets throughout a network, VoIP is a much more efficient means of utilising a network's carrying capacity. It's not only technologically more efficient, it's also more economic in its use of telecomms infrastructure.
The marginal cost of an IP call is minuscule compared to the marginal cost of providing a dedicated circuit-switched line. Moreover, VoIP can transcend regional and national boundaries. Imagine making international calls for a couple of cents, or even fractions of cents. The economic potential is enormous, as the cost of providing bandwidth is dropping faster than the cost of computing. In fact, there is a tremendous amount of voice traffic currently carried in IP format over long-distance networks - it's just that the end closer to your telephone is circuit-switched.
The difficulty to date has been in getting VoIP to the same quality as PSTN (public-switch telephone network, or circuit-switched) phone calls. That said, IP's only been around for a couple of decades, whereas PSTN networks have had more than a century to evolve. The technology will improve and VoIP will grow in usage.
In the Australian context, VoIP seems to be possible only with a broadband connection, and not at a retail level just yet, for a variety of technological and administrative reasons. Needless to say, Telstra is watching the technology very closely, because it is extremely jealous of its monopoly hold on the "last mile" of copper that leads into your home that provides you with a cricuit-switched phone call.
If you could break that monopoly by offering phone calls that do not use Telstra's PSTN they could be placed under real revenue pressure.
Posted by: Fyodor at September 24, 2004 11:29 AMFyodor,
I agree:
The difficulty to date has been in getting VoIP to the same quality as PSTN (public-switch telephone network, or circuit-switched) phone calls.
By quality do you mean reliability of an IP network in the case of an emergency call, or making the call subject to the same powers of interception as apply to calls made over the PSTN?
Also as far as I knew, Comindico were more a wholesale or backbone connectivity provider, as opposed to a retail style telco?
David
Posted by: David at September 24, 2004 11:40 AMDavid,
I meant quality and reliability in the sense of sound quality and not dropping out. This is important for ease of use and emergency situations. I've been told that VoIP in retail trials is currently about on a par with mobile phone calls, but is improving. The "problem" with interception/phone tapping etc. is another matter...
I agree with you on Comindico - the retail element wasn't strong.
Posted by: Fyodor at September 24, 2004 11:52 AMComindico's demise seems more related to poor management than uptake on technology. I base that on knowing people who used to work there. They were burning through money from the get go. If such a compay had less grandiose plans they may have made a good go of it.
VOIP has begun to take off. But its adoption in Australia will depend on the uptake of broadband. Fyodor is on the mark re Telstra. There are rumblings in the US that VOIP could mean the end of the traditional Telcos.
Posted by: Irant at September 24, 2004 12:16 PMJohn:
The real advantage of VoIP occurs in third world countries with monopoly suppliers which overprice the cost of calls - as for example, Việt Nam. Normally, the cost of a call from VN to Australia is about AUD $1.40 a minute.
That's why VoIP is exteremely popular, given that (a) buying a VoIP card costs you about $ 3-5 dollars per hour; (b) $1.40 is often a hourly wage over here; and © the locals have a lot of relatives overseas. Who doesn't want to talk to their brother in Melbourne or Milwaukee?
Personally, it has been a god send for the locals and the odd expatriates who live here. It also (indirectly) explains why Internet cafes are popular. Or you could have ADSL, but very few private people have it (I do).
Posted by: Peter Murphy at September 24, 2004 12:24 PMWhat VOIP does is end the sham where telcos pretend they are circuit switching when their whole system after the exchange is packet switched . VOIP ends this particular form of tickling the customer for more revenue.
Posted by: Bill O'Slatter at September 24, 2004 12:24 PMOf course, the push for VoIP would be considerably reduced if the telco dinosaurs stopped trying to support their doomed business model, but anyway...
Posted by: Robert Merkel at September 24, 2004 12:26 PMVoIP is to PSTN as hydrogen fuel cells are to the internal combustion engine. VoIP is theoretically a much better idea, but PSTN has well over a hundred years head start.
Areas where VoIP has a technological advantage:
- more done in software, simplifying hardware infrastructure and changing upgrade economics.
- IP can convey many kinds of services - broadband internet, TV and phone on one cable. (OK, Optus does that already)
- simplified provision of enhanced voice services such as teleconferencing, stereo hi-fi, video. Expect to see products from telcos along the lines of Apple's new iChat (I'd just like to take a moment, John, to say how much I covet your shiny new Cinema display)
- privacy is simpler (though still not simple) to achieve with VoIP.
- VoIP hardware and software standards are being implemented by networking vendors as well as telco vendors. This increases consumer choice. All going well, the standards will evolve to reduce lock-in, too.
VoIP is doing fairly well in the PABX market. This is interesting, because a PABX has many of the same issues as a regular telephone network; for example, voice quality and emergency override, are more important in a PABX than the public network. There are other benefits too, such as having only one network port at each desk to service both phone and LAN, which is a significant saving in a large fit-out.
Posted by: Alan Green at September 24, 2004 12:38 PMOne fundamental problem with VoIP is that it requires a separate power supply. If the power goes out, you won't be able to use the phone to make an emergency call.
Posted by: Robert at September 24, 2004 12:49 PMThe basic reason for VOIP is that eveyone is going to get a broadband IP connection anyway, so you might as well use it for phone calls as well. IP is certainly easily capable of handling phone calls, and probably at much higher quality than PTSN. All that's needed is for the technology to mature a bit and become more standardised.
It makes no sense to have an internet connection and a seperate voice connection. Telephone traffic is just another form of data, albeit with higher demands on network performance.
Having a broadband connection is not enough, though. You need a reliable high quality ISP on the other end of it as well. As most internet gamers could tell you there is quite a variation in quality of service provided by ISP's, even on a broadband connection. Most ISP's broadband services do not run at their claimed high performance speeds consistently enough to run a stable VOIP connection as dependably as current PTSN. This will be a stumbling block to VOIP replacing PTSN altogether in the medium term but not to it encroaching on PTSN.
Posted by: still worknig it out at September 24, 2004 02:12 PMSkype is trying to decentralise by using Peer-to-Peer technology. I do not know how they work exactly, but I do know that I talk to my friends in London with crystal-clear clarity for free. However it is still not perfect- talking to friends in Canada is frustrating. I suspect there's still bugs to be worked out in the Skype system.
Posted by: Scott Wickstein at September 24, 2004 02:46 PMHere are a few other advantages on top of those above:
- it's more robust. Circuit switching relies on a direct link. If any part of that link goes down, the call stops. A packet can just redirect itself along a link that's up.
- it's more effecient. A packet can decide which links and nodes have more capacity and direct itself along them.
- it's not geographically dependent. You can check your email from anywhere, but your circuit switched landline number just goes straight to your house. A VOIP phone (or device such as a PC acting as a phone) can be plugged into any IP network connection and recieve calls to your number. It authenticates with a server, much like email.
- as it's data-linked, it can provide additional info. Who's online, for example (like a messenger service). You could set your phone to say "I'm busy" or "I'm out" and those on your "friends" list can see your status without having to try and call you.
- it offers the possibility of only having one ID to contact anyone by voice, messenger, email, videocall, fax etc. Rather than having a number, username etc for each one, you could have one client device which allows you to select someone and decide how you want to contact them (call PK, fax PK, email PK etc).
- also, as stated above, it's more efficient as there's only one system, the TCP/IP link rather than two, one for data and one for voice.
Posted by: PK at September 24, 2004 03:04 PMScott, Skype's voice quality has nothing to do with "peer-to-peer technology". Once the call is set up, data is shared point-to-point like any other internet transfer would be. The great voice quality compared to your regular phone is mainly due an artificial limitation in the PSTN, as discussed on this page. The public telephone network filters all frequencies lower than 300 Hz and higher 3000 Hz; Skype doesn't. The other thing that Skype probably does, but they don't talk about, is some tricks to cope with "dropped packets" (a problem inherent in the Internet but not with the normal phone network), but as Skype's technology is proprietary I don't know for sure.
Posted by: Robert Merkel at September 24, 2004 03:26 PMTelstra has been using packet technology to route voice messages for several years now, but there is a strong commercial reason - and not technical reason- that this is not extended to the last mile.
Currently Telstra makes a lot of money selling data transfer via ISDN to businesses. Pushing packet switching to the last mile will mean that they will no longer be able to differentiate between business data (high margin) and voice traffic (low margin).
Packet switching also means they will not be able to differentiate between long distance calls (high margin) and local calls (low margin).
Imagine what happens to airlines who are unable to discriminate between casual travellers (willing to stay weekend away from home) and business travellers (not willing to stay weekend away). They'll all be forced into a commodity business with slim margins.
However, VOIP does not enjoy the huge installed base of users, and consequently suffer the network disadvantage. The appearance of VOIP to PSTN gateways will soon pose a serious threat to the old telecom monopolies. Unfortunately, the barrier of entry to supplying VOIP-PSTN gateways are quite low (witness the number of calling card companies around), and hence it's a pretty cut-throat business in the VOIP end too. Because of this, VOIP roll out is slower, and is not enjoying the mass acceptance that it deserves.
In the future, most voice communication will be over VOIP rather than PSTN, just as email has largely replaced snail mail for letters. The bad news for companies is - like email, the network is not going to be owned by anyone, and companies will find it hard to profit from it.
Interestingly, when the cost of call comes down to zero, VOIP spamming will become a major issue. Some US companies are already working on VOIP spam filters.
Chui
Posted by: Chui at September 24, 2004 04:23 PMThe per phone call transit cost a fiber optic cable link - eg from Perth to Melbourne - is a couple of hundredths of a cent. There is virtually no economically basis for distance charging on most phone calls in Australia. Some newer regimes like mobile plans don't distance charge. So how do they get away with it? History and expectations.
VoIP as a consumer product is is just a way of sidestepping telcos' historical charging regimes.
As an office system, VoIP is potentially a big saving for companies as it coalesses their voice and data systems into a singe infrastructure and allows various hybrid services. While VoIP is new and fashionable so manufacturers of handsets, routers, PABXs, etc have been able to charge wild premiums for fairly simple technology. Prices are falling. Allowing these VoIP office systems to connect between companies will progressively eliminate call charges. Eventually you'll be able to download a public domain VoIP PABX application, load it on your internet connected linux box and plug your $10 chinese VoIP handsets into your local network. Not quite yet.
I wouldn't rate Telstra shares as a long term hold.
Everthing is packet switched past the the exchange these days and increasingly more before. The theoretical efficiency advantage of circuit switching is only realized when bandwith is expensive. As bandwidth got cheaper, packet switching's flexibility and aggregation take over. More practically, packet switching requires more sophisticated technology, but this becomes cheap as chips.
Posted by: Jim Birch at September 24, 2004 04:59 PMReturning to JQ's question, my interpretation of VOIP as a term is that it refers to the consumer and office product, rather than to the technology of packet switching. As others have pointed out, that product has value in corporate settings, but not so much in consumer settings. Further consideration of the value falls into engineering idealism and different regulation of telecommunications, rather than market reality. That's why I class it as dot.com.
I don't really care one way or the other.
Posted by: Tony Healy at September 24, 2004 06:23 PMPrevious commenters (especially Fyodor) have explained the technical aspects of VoIP much better than I could. The pricing advantage arises from greater efficiency of using existing pipes and the Internet's basic packet-switching system rather than dedicated lines with circuit switching. In a practical sense the advantages are even greater.
The online law degree that I have developed and am reponsible for co-ordinating (which is Australia's first and one of the few fully online degree programs in the world) relies heavily on VoIP via voice-activated chatrooms in which we conduct real-time tutorials where students from all over Australia and the world interact freely, fully and flexibly with tutors, almost as if they were all in the same room together. In some ways it's actually better than a live tutorial experience, because the chatroom includes a built-in web browser so that students and tutor can surf the web together and "workshop" cases, legislation, journal articles etc together, and a fairly sophisticated whiteboard as well.
The disadvantage at present is that there is no visual interaction, which is a significant component of normal face-to-face communication. The chatroom system we use has the capacity to add multiple video windows, but our testing to date has indicated that this doesn't work satisfactorily for students with dialup Internet connections (whereas pure voice/audio does). Once the majority of our students have broadband connections, we WILL be moving to a full video and voice over IP system. Of course, it's possible to buy tele-conferencing services using old-fashioned dedicated line circuit-switching technology that offer all or most of these facilities, but the cost is prohibitive and will remain so.
Anyone who remains sceptical of the benefits and enormous potential of VoIP (e.g. John Quiggin) is most welcome to drop into one of CDU Law School's e-tutorials and listen and observe them in operation. I guarantee you'll be a convert once you can see how it actually works. There really isn't any comparison, either in cost or performance terms, with circuit-switching-based models. You simply don't need the potentially higher voice quality of a dedicated line (which mostly isn't realised anyway) for ordinary conversation, and the lower cost and vastly greater flexibility and potential of integrating voice with the Internet is enormous. CDU is on mid-semester break for next week, but any readers who may be interested in seeing/hearing how the technology works in practice in a higher education environment is most welcome to contact me and I'll arrange guest access.
Posted by: Ken Parish at September 24, 2004 06:32 PMWhat most of the others said about VOIP above. Except I don't believe its reliable enough to rely on in emergencies. So we will all have POT, plus a mobile (or two), plus VOIP /broadband to cover all contingencies.
At the moment I speak to my daughter in Taiwan each week using one of those telephone cards over the Telstra phone. Costs me $10 for 5 hours. It is obviously normal phone at either ends for the last mile but seems like its VOIP the middle bit. We now use a card for all long distance calls in oz as well.
Posted by: Francis Xavier Holden at September 24, 2004 09:28 PMLike Scott, I am also a very happy Skype user (which is no longer just P2P). Give it a try and you will be impressed. It has the added advantage of not contributing to Telstra's dividend to the Commonwealth, which then gets wasted on public schools and hospitals (JQ may not appreciate that last feature!)
Posted by: Stephen Kirchner at September 25, 2004 11:07 AMhttp://www.itee.uq.edu.au/~coms7902/Lecture%20Notes/Lecture_7.pdf
Posted by: Bill O'Slatter at September 25, 2004 11:36 AMJohn,
I appreciate your comments regarding VoIP. I’m in the States and work as a telecom consultant so I may come it this a little different than an IT mentality. I’ve watched a several companies and schools have made the move to IP. I’ve sat in presentations of why VoIP is the ONLY solution. In what I have read about your focus on how IP works and prioritizing packages as well as bandwidth issues have be resolved are not complete.
I’ll admit that prioritizing voice has improved and the technology is far ahead of where it was a couple of years ago. But as I speak with other independent consultants we are coming to the same conclusions. Is VoIP a cost affective, reliable service? At best whether you are looking at running it over your own WAN or using it to speak to the outside world you’d have to be nuts to give up your PSTN 100%.
Here’s the real question, is it a better solution that PSTN/PBX in cost, time investment and quality? My answer is no. 1) Equipment (handsets) fails sooner 2) Cost more, 3) takes more maintenance to keep in working, 4) has more down time than a PBX. Some have to add staff to keep up with the issues.
In my office I use both VoIP and switched circuits. It’s never as clear using VoIP, and at times calling into an AA, it cannot identify my keystrokes to sent me to the right extension.
A dedicated T1 circuit would resolve most of the issues around QoS (Quality of Service) however I still don’t see the justification of moving away from 100% circuit switching compared to partial or total VoIP.
Dan
PS. This is my first time to respond to this format, would you mind sending me your response by email to make sure I am aware of your comments? danb@corptele.com.
I'm not 100% sure about what people are comparing here, but I'd like to offer some (belated) comments.
I think that it's misleading to talk about aggregation gain being the reason for VOIP.
While packet switching does increase efficiency, any first year telecoms student will tell you that the QUEUE you create in getting that aggregation gain will often make the service unusable for voice. How does a telco get around that problem? They massively overdimension their (say) ATM bearers so that there is very little queuing. Is there an aggregation gain? Yes. But it's pretty much lost trying to control latency and jitter. So - The idea that VOIP is more efficient because of statistical multiplexing is overly simplistic and ignores the queuing & subsequent overdimensioning that takes place.
A couple of people mentioned bandpass filtering, etc. Obviously there is still arbitrary bandpass filtering involved in VOIP - if there weren't we couldn't do the digital to analogue conversion, so I'm not sure what's being got at.
Of some interest is the particular scheme used for compressing sound- obviously depending on the technology you use, voice quality and required data rate vary.
In summary, John, I think that you can approach this from the point of view of a market and just ignore the technical side. Yes there are issues, but the KEY driver here is about people's expectation for pricing and how they VALUE a service. Voice & Data are really the same thing (with some latency, jitter and bandwidth guarantees) but they are charged differently. VOIP is really nothing more than a means by which people exploit a price discrepancy.
From the point of view of a telco, initially it is simply a threat (truly commodity bit carrier) but has some appeal for allowing simplification of the core network (Internally for a telco there are a number of heterogenous networks which for simplicity ought to be knitted together).
Posted by: CJ at October 6, 2004 11:23 PM