Sprint Moves Phone Network to IP

Ryan Barrett writes "Sprint announced that it has 'begun transforming its telephone network so voice calls are transmitted in packets.' AP article here. Combined with a recent /. story about Telus doing the same thing, this sets an interesting precedent. Many telcos already use packet-switching to handle a significant chunk of their calls. Is this the beginning of the end for circuit-switched networks?"

Comments

Telus Calls Sprint

[yoey]

Hello, Sprint? This is Telus. You stole our idea you son-of-a-bit....

Re:Telus Calls Sprint

[Anonymous Coward]

Ah, another dang IP IP legal fight.

Re:Telus Calls Sprint

[Anonymous Coward]

MORPHEOUS DIES! Right after he fucks Trinity! Then Neo reaches into his chest like in Mortal Combat and grabs his heart!! And eats it!!!

Re:Telus Calls Sprint

[Anonymous Coward]

UH, no. I have been working on this project for over 6 months. How long has Telus been doing it? It just happened to make the news after Telus. First office success, all your circuit are packet now. ;-)

The USA is Dying

[Michael's a Jerk!]

It is official -- The UN is now confirming: The USA is dying

One more crippling bombshell crushed the already beleaguered American economy
when x-rates.com confirmed that the American Dollar has dropped yet again,
now down to .9 of a Euro. Coming on the heels of a recent Usenet survey which plainly states that The American Economy is in a recession, this news serves to
reinforce what we've known all along. The USA is collapsing in complete disarray, as fittingly exemplified by failing dead last [samag.com] in the recent Sys Admin comprehensive networking test.

You don't need to be a Kreskin to predict the USA's future. The hand writing is on the wall: The USA faces a bleak future. In fact there won't
be any future at all for the USA because it is dying. Things are looking very bad for America. As many of us are already aware, the US continues to lose relevence. Red ink flows like a river of blood.

The IT industry is the most endangered of them all, having lost 93% of its core
developers. The sudden and unpleasant departures of long time IT jobs to india only serve to underscore the point more clearly. There can no longer be any doubt: The American IT Industry is dying.

Let's keep to the facts and look at the numbers.

Microsoft Encarta states that there are 291,065,636 people in America. What is the US's national debt? Let's see. According to the The Debt clock the USA's National Debt is 6,465,271,811,559.14. Therefore each American
is $22,212.42. in debt. In fact, the USA's national debt has continued to
increase an average of $992 million per day since September 30, 2002. Indeed,
it can clearly be seen the the US is going broke faster then the Soviet Union did

Due to the troubles of American Meddling, An Capitalist Gorvernment and so on,
South Vietnam was attacked was taken over by North Vietnam who sell another
a more compassionate government. Now Iraq is also dead, its corpse turned over
to feed the US media.

All major surveys show that the USA has steadily declined in the world economy.
America is very sick and its long term survival prospects are very dim. If
the USA is to survive at all it will be among a broken collection of warring
factions. America continues to decay. Nothing short of a miracle could save
it at this point in time. For all practical purposes, the United States of America is dead.

Fact: The USA is dying

Re:The USA is Dying

[Anonymous Coward]

LOL. That was a nice troll, gotta give props.

Re:The USA is Dying

[Anonymous Coward]

Mr Rumsfeld acknowledged that Iraq may have destroyed all its chemical munitions and weapons of mass destruction before the US-led invasion last March.

Re:The USA is Dying

[Anonymous Coward]

It is official -- Slashdot is now reporting that circuit switching is dying.

One more crippling bombshell crushed the already beleaguered circuit-switching community when slashdot.com community didn't care that the use of circuit switches has dropped yet again. Coming on the heels of a recent Usenet survey which plainly states that circuit switches are boring, this news serves to reinforce what we've known all along. Circuit switch use is collapsing in complete disarray, as fittingly exemplified by falling dead last in the recent Cowboy Neal polls.

You don't need to be a Kreskin to predict circuit switching's future. The hand writing is on the wall: Circuit switching faces a bleak future. In fact there won't be any future at all for circuit switching because it is dying. Things are looking very bad for circuit switching. As many of us are already aware, the circuit switch continues to lose relevence. Red ink flows like a river of blood.

Fact: Nobody cares Timmy.

WOOHOO!

[Anonymous Coward]

FROST PISS!!!

Yeah!

ffsdffsfsdf
ds
fsdf
sf

ahoy!

[count_sporkula]

FP all round, yarr!

Re:ahoy!

[count_sporkula]

arse.

Re:ahoy!

[Anonymous Coward]

feck. grrrls.

Re:ahoy!

[Anonymous Coward]

Drink! Drink! Drink!

Re:ahoy!

[Anonymous Coward]

Dick! Balls!

Re:ahoy!

[Anonymous Coward]

nuns! NUNS!

What's better?

[Anonymous Coward]

(A) An IP phone network

or

(B) Sex with a mare?

Re:What's better?

[Anonymous Coward]

I would have to say A, thank you very much.

Re:What's better?

[Anonymous Coward]

An IP phone network sounds disgusting!

Ouch, now I have to remember IP addresses too

[infonography]

I am so gonna miss the (555) 123 4567 type numbers. Will you have to use the * key to make the dots in the quads?

Reminds me of that old Dogbert Joke about having a Tilde in the phone number. I wonder how long it will be till them move to IPv6, won't that be a joy to dial.

Re:Ouch, now I have to remember IP addresses too

[Anonymous Coward]

h0h0!

Er, unless they really fuck up the QoS, it should be
transparent to everyone.

Re:Ouch, now I have to remember IP addresses too

[Anonymous Coward]

Wrong. You couldn't be any further off the mark. Now, do us all a favour and put your nose over a steaming pool of C11H26NO2PS and inhale deeeeeeply.

Re:Ouch, now I have to remember IP addresses too

[Ryan Amos]

Well, that's what we have DNS for, isn't it? :) You could easily rig up a service like DNS (say, .phone) and have (country).(areacode).(exchange).(number).phone just like we have now. For IPV6, I'd imagine you have to. The human brain gets really confused when numbers get longer than 7 or 8 digits; as in you start transposing digits, forgetting parts, etc unless you can come up with a mnemonic device to remember them with. So I think it's safe to say it'd have to closely resemble the system we have today. At least from an end-user perspective.

Re:Ouch, now I have to remember IP addresses too

[Jeremiah Cornelius]

"This boy shouts, and this boy screams!"

Re:Ouch, now I have to remember IP addresses too

[miu]

No need. E.164 number and DNS

Re:Ouch, now I have to remember IP addresses too

[Anonymous Coward]

Somebody needs to read up on how DNS works. Domain names are resolved from back to front.

Re:Ouch, now I have to remember IP addresses too

[hatrisc]

do you really want google searching your dns/phone entry? ugh i can just imagine.

Is Mr or Mrs. slashdot there? When would be a good time to call back?

Re:Ouch, now I have to remember IP addresses too

[Anonymous Coward]

that's what H.245 is for as I recall.

Re:Ouch, now I have to remember IP addresses too

[maxphunk]

What happens when we go to IPv6?? Try and get Grandma to remember your 128-bit phone number in hex...

Re:Ouch, now I have to remember IP addresses too

[Ryan Amos]

I realize this :) But phone numbers are front to back, and it wouldn't be too hard to have the .phone root resolve things backwards. Basically, the point is to provide an interface as close to what we have today as possible, because it works and people know it well.

Re:Ouch, now I have to remember IP addresses too

[Anonymous Coward]

i doubt u can forget phone numbers. my read is that numbers stay. in fact, the key is ur voice is packetized ... so u dont get a dedicated 64kbps circuit span for each conversation. u get a "virtual tunnel" by packetizing ur voice and sending it over IP at the mtso. the last mile should still be circuit based however. as u can see, this isnt publicized as it works better for them than it does for us (not that it doesnt work for us) ...

just my $0.02 ...

More like the middle of the end

[tgma]

The beginning of the end started when the equipment manufacturers started producing boxes that allowed VoIP calls to have the same quality as circuit-switched ones. We all probably make a lot more IP calls than we are aware of.

The quicker companies do this, the better it will be for their margins - this news from Sprint probably doesn't mean much for their users, but their shareholders should be happy. The cost of carrying VoIP is much lower, which is what allows those calling card companies to stay in business.

Re:More like the middle of the end

[Smidge204]

And I can already imagine the price going up. Supposedly the hone companies already bill thie customers for stuff they don't even have, I can only imagine the further charges when they start arguing they have to maintain their switching systems as well.

Crooks, every last one of 'em. I hope that as soon as they all to to VoIP, someone figures out how to connect to it from a standard computer and bypass the local phone company completely! (And yes, I know there already exists a way for calling someone on your PC, but that still requires a third party company to do the IPSwitch network translation IIRC...)
=Smidge=

Re:More like the middle of the end

[otmar]

I doubt that Sprint will be able to keep charging their users the old rate. The savings will be passed on to the customers sooner or later.

There is a really good article on the economics invoved by Clay Shirky. Recommended reading.

/ol

Re:More like the middle of the end

[stinky wizzleteats]

The beginning of the end started when the equipment manufacturers started producing boxes that allowed VoIP calls to have the same quality as circuit-switched ones.

They did? I've recently been in some training where we discussed the problems of getting VOIP to work on networking gear. It was interesting to see scenarios where one slight change to MTUs, etc., caused a 180 degree reversal of QoS behavior. Nothing I've looked at, including carrier class boxes, handles VOIP worth a damn. Data boxes, and more importantly, the protocols that run on them, simply weren't designed to handle realtime media streams.

Even if QoS is working correctly, VOIP is a train wreck. I mean, this is worse than when we got the bright idea to centralize faxing onto a server with a brooktrout card in it.

Re:More like the middle of the end

[lynx_user_abroad]

Yes, but...

There is a really good article on the economics invoved by Clay Shirky.

However, there's a new component here: the "legislative" layer.

In the ZapMail scenario, individual businesses could replace the FedEx service simply by buying a fax machine; but that's only because of FCC rules which a) demanded that common carriers (the phone companies) could not discriminate against different users of the network, and b) allow any non-destructive device to be connected to the Public Switched Network. In other words, the fax machine revolution was sparked by FCC rules which created an open and equal (Lessig would call it "flat and end-to-end") network on top of which others could build and innovate.

However, the FCC has chosen a different path with "broadband" these days. The FCC has already begun to rule (and appears ready to go whole hog with more rulings) that companies which provide broadband services own their network. If we were replaying ZapMail today, that means the phone companies would be allowed to prevent individual businesses from using their network to transfer documents via fax. Customers who wanted to deliver a document would have to use either an authorized corporate partner (in this case, FedEx) or the services of the telephone company itself.

We're already seeing manifestations of this in the Internet today; Most ISP's won't allow individuals to use port 25 (SMTP) so if you want to send email, you have to use the server provided by the ISP. That service is no longer available to customers, even the ones who have already bought equipment capable of sending and receiving email direrctly.

Consider AOL's position concerning mailing lists: If you want to provide a mailing list (free or fee) service to AOL subscribers, you must either a) run your list from an "approved" (read: corporate partner) server, or b) trudge through a lengthly approval process to get your mailing list onto the "whitelisted" list. It's not a far stretch to see the day when there will be a fee to mailing list managers in order to service AOL subscribers, and that will be the end of the free mailing list.

So, the next thing to fail will be the "free" services currently offered on the Internet.

We're already seeing pressure on major business sites to get an AOL keyword associated with their site. For all I know, getting that keyword cost money. If it doesn't already, it soon will. When that starts to happen, I wonder if Slashdot will be pulling in enough revenue to maintain contact with it's AOL customers, or if Slashdot will become another site AOL subscribers have to jump through hoops (or pay and extra "access" fee) to access?

Will we see a day when on-line gamers will be required to use only the "service provider approved" gaming server, because ports to other servers are blocked? Isn't Microsoft doing something like this already on MSN requiring a Passport to access their Gaming server?

Will we soon see the day when trying to access any "terrorist" news site (like Al Jazerra) will be impossible, and accessing any "liberal" (read: non-corporate/administration partner) news site will be slow and unreliable at best? And if you're trying to get to the campaign web site of the non-incumbent candidate, well, you can just forget it.

There's more at work here than just simple economics. Without on open networking layer as we had with the PSN, there won't be the kind of telecommunications revolution we say after the AT&T breakup in 1984.

Re:More like the middle of the end

[boskone]

check out vonage. www.vonage.com they provide a box that plugs into your network, and has an rj-11 phone jack. really, really cool service, and it's cheap and good. Take the vonage box anywhere, plug it in, and your standard $10 telephone rings there as if it's local.

it's cool stuff. a friend of mine is using it and I'm signing up next week.

Re:More like the middle of the end

[thogard]

We fixed the VoIP problem. We built another network. It still has nasty hiss on the lines that I don't seem to remember from before the VoIP days.

The big place this will work is for the international compaines. Current VoIP voice rates from Australia to the US (or the UK) are AU$.019/min or about US$.68/hr. Compare that with full AT&T list price to someplace in central Africa. which can be several dollars per minute.

so does this mean

[cyrax777]

its going to cheaper in the long run for my telco service nah probly not they probly still charge the same amount and pocked the rest as always.

Re:so does this mean

[Anonymous Coward]

its going to cheaper in the long run for my telco service nah probly not they probly still charge the same amount and pocked the rest as always.

You mean those dirty rotten for-profit companies are going to make more money without charging you more? How can we allow such evil to exist?! We must stamp out all such evil. We must force companies to lose money. How dare they make a profit? How dare they worry about investors? Nay! We must do everything in our power to make sure companies never profit and never ever let investors see profits. I wonder what Kenneth Lay is doing. I'm sure he'll be happy to help us re-educate the corporate officers involved here.

Re:so does this mean

[randomErr]

If they just keep the rates the same I'll be happy.

With inflation and other operating cost increase that would be a mircle.

Re:so does this mean

[Servo]

In the short run I can almost guarentee they will stay the same, but as more competition heats up for "call anywhere in the US for a flat rate" plans (from both VoIP like vonage, or large telcos), they are going to have to become more price competitive or lose business.

Re:so does this mean

[tuba_dude]

They could always make more profit by lowering their prices slightly and getting more customers...Or, another crazy idea (probably too crazy), they could use the extra profit to upgrade infrastructure or increase R&D spending...crap! That can't work...Hmm...let's think of how this will really turn out...I can see it now...(Dreamy fade)

INT. 53rd story, Board Room - Day

12 Directors are sitting in their huge leather chairs at a gargantuan marble table. In the center of the table is a cake with money drawn on it in icing. Around the room are balloons full of money, and on each director's plate is a party hat made of money.

CEO
Do you like my hat? It's made of MONEY!

He laughs manically and all join in.

What a scene...

Not IP

[NeuroKoan]

The article didn't state that Sprint was switching to an IP based network, just a packet switched network. Is this actually going to use IP? A quick google search brought up no mention of IP (but I'm also lazy, so I only read the first page of links).

Re:Not IP

[hughk]

Modern switches can talk IP. They are essentially just computers with some specialised I/O. Switches can talk to each other locally via a LAN and they can send long distance traffic via a variety of WAN connections. IP6 has been preferred for a while between switching centres because of the QoS support. The lower layer is generally ATM.

A friend who used to work for Nortel (didn't many) mentioned this. Worldcom did most of their long distance stuff on top of IP6.

Re:Not IP

[Anonymous Coward]

Quite. I suspect the author is jumping the gun substantially when he says they're going to be using IP. Would have thought it would be massively inefficient for your standard 64kb voice stream.

Re:Not IP

[Heartz]

Worldcom did most of their long distance stuff on top of IP6. And look where that got them :P *dunks*

Re:Not IP

[dew-genen-ny]

As an ex-worldcom employee, I'm almost 100% they didn't use IPv6.....

Re:Not IP

[Anonymous Coward]

Modern switches can talk IP.

That would be a router, not a switch.

Re:Not IP

[c_g_hills]

Managed and Layer 3 switches *do* talk IP.

Re:Not IP

[hughk]

A friend was an engineering manager for them, the network was based around ATM as the transport and at least some of the stuff was IPv6. This was only seen at the deeper levels and the corporate network was IPv4 for data which would have been routed on top of the IPv6 connections.

Re:Not IP

[dew-genen-ny]

Perhaps in the US....

I was working as a Network Engineer for them, in amsterdam, and as far as I know, we only had a test IPv6 network.

We did a lot of discussing IPv6, and as per most of the big ISPs right now, decided that it isn't currently worth it.

Could insert obligatory US != world comment, but to be honest that's totally how Wcom worked....Europe? that in Texas ???

Re:Not IP

[hughk]

The friend was working in London and Frankfurt (mostly). AFAIK, it was only on the backbones and maybe only some of those. The idea was just to use QoS to separate traffic so that they could sell different service levels depending upon the nature of the data (i.e. video conferencing).

Re:Not IP

[cfulmer]

So, I'm a very recent ex-nortel person who did a lot of packet stuff. If I recall correctly, Sprint's network is an ATM one, either AAL1 or AAL2. AAL5 is IP-over-ATM, and isn't as common.

Note that in general, these are all behind-the-scenes private networks. You will still be circuit-switched to a point (inside your local office, typically). Then there'll be a TDM-to-packet gateway which converts your circuit-switched connection into ATM (or IP).

From an IP point of view, one of the side effects of this is that you don't need a seperate IP address for every phone, or even a seperate address for every house. All you need is an IP address/port number combination for each end of an active call in any given network. (And there are ways of getting around that restriction too.) Since these are all private networks cut off from the internet, IPV4 provides more than enough addresses.

Packet telephony all the way to the home, at least from the telcos, is some ways off. You'd either have to have a gateway inside your house to which you connect all your legacy phones, or you'd replace all the phones with IP phones. As you can probably see, there's a lot of inertia behind that *not* happening -- try convincing your great uncle Bert that he needs to replace all the phones in his house.

Re:Not IP

[vestus]

See UEMG9000
Networked over an OC3 ATM fiber link, with other options. 7900 POTS lines max, and can take DS1, xDSL, TDM, etc. VOIP is still in development afaik, but should be done soon.
I used to work on it, and its a pretty robust system. Anyone need an ex-Nortel embedded driver dev in RTP?

Re:Not IP

[windex]

But this was already done in some form.

How long has it been since you've seen a rotary telephone in use? ...

Re:Not IP

[Moskit]

> IP6 has been preferred for a while between
> switching centres because of the QoS support.

Ahem... this is not exactly true. Contrary to popular belief IPv6 does _not_ offer any particularly better QoS features. You can achieve practically the same results using IPv4.

Transmitting voice over IPv6 is also not very efficient - headers are quite long compared to v4.

Large operators usually run voice either over IPv4 or directly over ATM.

Re:Not IP

[Politburo]

How long has it been since you've seen a rotary telephone in use?

A long time, but if I flip the switch on my phone, I can still call anyone. Backwards compatibility doesn't exist in the IP phone switchover.

Re:Not IP

[RedStapler]

An even better question to ask is, should they choose to use IP, are they mixing data and voice? If you just decide to use IP for voice, it is not a huge deal provided you sufficiently overprovision the network. Right now, it probably isn't a big deal but it would be interesting to see what would happen come Mother's Day or other big call days of the year.

Re:Not IP

[warpSpeed]

Backwards compatibility doesn't exist in the IP phone switchover.

If you use a gateway to convert touch tone and rotary phones to IP, there is no reason why backwards compatability cannot be implemented for use with the current hardware. Backwards compatability for the phone numbers, and emergency service (911 here in the states), would require a software solution. That should not be to dificult to do (even for the phone companies)

How long has it been since you've seen a rotary telephone in use? This reminds me of a story (can't remember where I saw it) about a child going to the the school office who needed to call home. The school had an old rotary phone and the kid just started at it blankly, never having seen one before, he did not know how to use it.

Re:Not IP

[cotu]

Um no, this is completely misinformed. IPv6 is not being used, and IPv4 handles QoS just fine. The IPv6 flowlabel is just now being standardized, and is orthogonal to the DSCP which is in both IPv4 and IPv6 (something of a substitute for the normal 5-tuple flow classifier).

IPv6 will be good for VoIP, but to say that it is being used is just plain wrong. If only.

Re:Not IP

[Anonymous Coward]

AAL5 has nothing whatever to do with IP. It is merely an ATM Adaptation Layer used to transport packets in cells reliably. You can put frame relay packets in it or ethernet packets in it or MPLS packets in it or anything you want in it.

Re:Not IP

[hughk]

My apologies (note, I heard this from a manager rather than a techie). The idea was that they need the service class concept to separate and prioritise traffic. The other benefit from IPv6 was better separation between the many IPv4 links that would tunnel through the net. Voice was interesting but the major selling point was supposed to be video conferencing. Note that Worldcom was a major supplier of network services to the military (I guess thats why MCI were so quick into Iraq. they probably already have some infrastructure travelling with the HQ staff).

Re:Not IP

[hughk]

The main issue was to provide a very large address space. Worldcom had some very big ideas at the time and to keep the traffic separated. I think in those days (about 18 months before the fall) they had an incredible amount of international capacity. They were also a major provider of data links to the Pentagon amongst others. How much was just planned and how much was in execution, I don't know as my friend was a manager rather than a technical type.

Re:Not IP

[hughk]

I know that it usually doesn't go much beyond the Central Office, but do big private switches ever connect in via IP. The last installation I had my teeth on, we had fibre coming in to a box from the telco and separate lines coming out of the box for the switch and data. My area was just the data and we just saw an unswitched circuit (well, actually lots, but each delivered separately).

Re:Not IP

[PatJensen]

To add to that a bit, most carriers transport voice traffic between COs over ATM VCs (virtual circuits), which get switched and meshed onto a SONET network. SONET provides reliability, resiliency and redundacy and channelizes customer voice and data transport. SONET generally runs on single-mode fiber physical media and provides a "ring" that is self healing.

SONET provides the back end transport for ATM/DSL, DSx/T1 DACS (digital access cross connect - T1s just end up as virtual circuits on large ATM links) and Frame Relay as well as voice encapsulated on ATM.

Hope that helps.

-Pat (an optical SME for a telco)

Lets face it

[headbulb]

Some things are not ment to be packetized.. I say voice communiction is one of them. I like having a call having its own dedicated bandwidth, That Can't be doss'd to be hungup. Sure it will make things cheaper, but is it really the correct thing to do. Also with packet networks They are not inherently secure, Switched networks are. (once you get over the taps)

Re:Lets face it

[Alioth]

The network will still be switched at a local level, I suspect (even if the future telephone exchange, instead of switching analogue circuits, works more like an Ethernet switch). With so much copper going from the home/office to the exchange, it's likely to continue to be in use for the last mile for some time to come.

Trunk switching has been multiplexed for decades already. Previously, it might have been multiplexed by FDMA (frequency division), and now it looks like they are moving to IP based (or similar) to route calls through exchanges. The end user won't notice the difference. It's unlikely that the call routing will be done over the public Internet.

The trunk network can already run out of capacity - you do not now have dedicated bandwitdth and never had dedicated bandwidth over the trunk network (ever got the 'All circuits are busy' message?) A packet based trunk network is no less secure than the existing trunk networks. Packet switching != routing over the public internet.

Re:Lets face it

[pe1rxq]

You can have QoS on a packet switched network to...
(Being able to dynamicly allocate bandwidth to users can be really handy compared to the fixed switched networks.)
And you can even have a private network.... making it just as secure as a switched network (even more secure as you can add encryption to prevent clandestine wire taps).
Add to this that a traditional switched network gets really noise after a few switches and digital networks will be able bounce your signal around the world without adding additional loss....

Jeroen

Re:Lets face it

[headbulb]

Well I made my first comment just to the point accross that just using a packet network to save cost's, may not be the best idea. Things have to be done right. I really don't have a problem if they develop new standards. Instead of just using voip. Whatever they want to do, we need a bunch of telco's developing a new standerd. Instead of just hacking onto tcp/ip (or whatever)Another thought: Things work more smoothly when we work on it together and let it mature.

Re:Lets face it

[LarsG]

You can have QoS on a packet switched network to...

If the physical/link layer supports it.

The telco-inspired technologies typically have heavy QoS support (ATM, HiperLAN2, etc), while it is more or less absent in LAN/Internet technologies (Ethernet, 802.11).

The Internet is best-effort, and the technology isn't really suited to do end-to-end guaranteed bandwidth/latency.

Re:Lets face it

[Alioth]

Add to this that a traditional switched network gets really noise after a few switches and digital networks will....

This kind of switching hasn't been done for years. Electronic phone exchanges have existed for decades, and digital phone exchanges (at least where I live) have made up the entire network for over 10 years.

The electromechanical exchanges did manage to hang on into the early 1990s in many places though. Good old Strowger. (An excellent site about the phone network in days gone by is Light Straw. If you are ever in a position to visit the London Science Museum, they have a good-sized portion of a Strowger phone exchange that you can play with - makes lovely clattering noises!)

Re:Lets face it

[Himring]

My uncle is an old-timer, circuit-switch technician for at&t -- he keeps the things running and is the lead tech across several states. He tells me packet-switchin' ain't all that, and although at&t is "looking into" going packet-switch over circuit, for the volumes they do, it's just not ready to replace it yet. To back up that claim, he's taken me through "the node" he works at and the majority of it is housing the nortel, circuit-switch racks. Of course, no one has stated at&t is going packet-switched, but I found this story interesting based on the conversations I've had with him.

His blurb on the issue is that voice, unlike data, requires a dedicated connection, and packet-switched doesn't give ya that -- circuit-switching can/does (the packets can arrive out of sequence). Now, I'm prolly mutilating his take on this since this is me (who doesn't understand all of this) trying to regurgitate conversations in the past with him (and maybe my DIMMs have been polluted by searchnetworking.com and reading this forum too).

Re:lets face it

[Anonymous Coward]

Thats a cool site about the phone system in days gone by, got any good sites about the curent phone system?

Let me guess

[botzi]

..will make it cheaper for Sprint to grow its network..
, unfortunately, monthly fees will rise with 25% due to the *better* services that'll be provided....;o)...

IPV6

[termos]

Call me on 3ffe:0501:0008:0000:0260:97ff:fe40:efab, see you soon.

Re:IPV6

[Anonymous Coward]

hmm, so much for easy to remember phone numbers like 1-800-CARPETS.

Now it'll be 1-800-I-DONT-HAVE-A-FLOOR-U-INSENSITIVE-CLOD

Re:IPV6

[c_g_hills]

I tried calling you, but the host appears to be down..
Pinging 3ffe:501:8:0:260:97ff:fe40:efab from 2001:630:1c0:1:201:2ff:fea9:9ae0 with 32 bytes of data:

Destination host unreachable.
Destination host unreachable.
Destination host unreachable.
Destination host unreachable.

8 149 ms 149 ms 149 ms 2001:200:0:6c04::1
9 281 ms 279 ms 287 ms pc1.notemachi.wide.ad.jp [2001:200:0:6c01:290:27ff:fe3a:d8]
10 277 ms 276 ms 277 ms pc6.otemachi.wide.ad.jp [2001:200:0:1800::9c4:0]
11 Destination host unreachable.

Re:IPV6

[noselasd]

You know, most telecom companies that implement IP only implements it in their backbone net, that is you or your telephone will never see any ip packets. However the nearest switch or group swithch might convert your speech to travel over IP.

Re:IPV6

[AndroidCat]

Nah, they'll have DNS. Can you imagine the cybersquating wars over pizza.*?

Re:IPV6

[ebh]

They already do that 1-800-TOO-MANY-DIGITS thing.

The longest one I actualy saw was 1-800-333-DIAMONDS, and they did emphasize the S at the end (unlike 1-800-MATTRES, "leave off the last S for savings").

Do they do that idiocy outside the US too, or is this another example of our monopoly on stupidity?

Re:IPV6

[Anonymous Coward]

I havent seen any other country using this system.

new pipe

[hfastedge]

so they will be laying new lines for this?

Re:new pipe

[smartfart]

Probably not.

For one thing, there are bundles upon bundles of dark fiber (run but never utilized) stretched between the metropolitan centers that they've yet to find a use for.

Secondly, laying fiber is expensive. The cost of converting to VoIP from their regular gear has got to be enormous, and if they had to lay fiber there's now way they could afford it. The only way Sprint could justify the conversion is if they were able to use their existing lines.

***************

Speaking of Sprint, last year at my old company (hi, John!), Sprint quoted us $11,000 per month to link 5 offices in a 2-state area with single T1s carrying voice and data, with a black-box hub of some sort buried somewhere in their network. It was rather hard to keep a straight face during our meeting with them. My last project (I almost completed it before geting laid off, bleh) at my old company was to replace a $150/mo. 56k line running Citrix with a $50/mo. DSL line at one of our remote offices using FreeS/WAN IPSec.

Geeks = 1, Sprint = 0 :-)

Re:new pipe

[Anonymous Coward]

No, but I'll be laying some new bitches once they see my smokin' new IP phone.

Re:new pipe

[Anonymous Coward]

To Moderator: This word "Flamebait" - I don't think it means what you think it means.

IP?

[Mr2cents]

Where IP? Who IP? Just because you send data in packets doesn't mean you have to use IP.. The article doesn't say anything about the protocol they're using..

Re:IP?

[pe1rxq]

They will probably be switching to an isdn network over glassfiber as backbone. Most of the world did this years ago. (You will have a hard time to find an circuit switching telephone network in the Netherlands, although most actual connections are still analogue it gets digitized the moment you reach the central)

Jeroen

Re:IP?

[don.g]

You seem to be labouring under the mistaken impression that digital == packet-switched and circuit-switched == analogue.

Circuit-switched is things like ATM and TDM (over E1s, etc): you establish a "circuit" to who you want to communicate with (an expensive operation), and then talk over that.

Packet-switched is when you cut the data up into packets, each of which can be routed independently - they have a target address in a header, and you don't need to set up a circuit to send them. On the flipside, guaranteeing bandwidth for a stream of packets from one node to another is harder.

I knew there was a use for that CompSci degree :-)

Re:IP?

[LarsG]

You seem to be labouring under the mistaken impression that digital == packet-switched and circuit-switched == analogue.

Circuit-switched is things like ATM and TDM (over E1s, etc): you establish a "circuit" to who you want to communicate with (an expensive operation), and then talk over that.

Hmm. Doesn't ATM do "virtual circuits" over a packet switched medium? Not that different from a TCP session, except that an ATM VC can provide guaranteed bandwidth and latency.

In my book "circuit-switched" == dedicated line from end to end, doesn't matter whether the signal is analog or digital. With that definition, ATM is doing virtual circuit switching over a packet switched network.

Re:IP?

[pe1rxq]

Part of what you are describing is called virtual circuit switching in which a 'circuit' is established over a packet switched network. In real circuit switching there is a dedicated electrical circuit between the two endpoints, which indeed can be both digital and analogue.
TDM is sort of a strange thing in that there is no real electrical circuit but you do get a dedicated time slot on the line. ATM definitly is packet switched.
Guaranteeing bandwith (QoS) is not hard at all, the routers simply need a table of active circuits.
Only packets for those circuits and in only a certain amount get through.

Jeroen

Re:IP?

[don.g]

Er. It depends what you mean by circuit-switched. When you want to talk to someone via ATM, you need to establish a virtual circuit to them, a process that involves each ATM switch between you and the other VC endpoint. This sets up the path the ATM packets in that VC will take through the network (and ATM packets are *small* - 53 bytes).

Compare this with IP, etc - the routers don't (or at least shouldn't have to) care about anything beyond the destination header in the packet, something that shouldn't be modified en-route. TCP doesn't require the intervening routers to know about it.

Re:IP?

[LarsG]

Er. It depends what you mean by circuit-switched. When you want to talk to someone via ATM, you need to establish a virtual circuit to them, a process that involves each ATM switch between you and the other VC endpoint. This sets up the path the ATM packets in that VC will take through the network (and ATM packets are *small* - 53 bytes).

True. I guess we're just arguing semantics. :-)

Anyway, in my book a virtual circuit running over a packet switched network is an emulated circuit over a packet network.

ATM cells are small, and ATM is specifically designed from the ground up to facilitate virtual circuits. No argument there. But it is still a packet switched network, imho.

Compare this with IP, etc - the routers don't (or at least shouldn't have to) care about anything beyond the destination header in the packet, something that shouldn't be modified en-route. TCP doesn't require the intervening routers to know about it.

IP is best-effort, TCP is a session protocol on top of IP. None of them have any comprehensive support for the kind of QoS you need for voice connections.

However, you can do pretty much the same thing for IP traffic by using MPLS, 802.1p and 802.1Q.

Last-mile

[benntop]

It is nice to hear that Sprint will be moving long distance telephony into the next generation. It sounds like it will save them a bundle and help them manage growth. Still, I am languishing with an analog copper-pair connecton to my home phone. Will someone (anyone) drop a fiber into my backyard?

Re:Last-mile

[netean]

nah you don't want fibre running to your door.
Otherwise you won't be able to get DSL services (well at least that's the pathetic excuse the use in the UK when you request aDSL and you have fibre to your door!)

Re:Last-mile

[amorsen]

That is not an excuse, it is simply the truth. *DSL needs a pair of metal wires. If you do not have that, no *DSL for you. You may have to make do with STM4. The US equivalent of STM4 is OC12, by the way.

Circuit Switching will still be around...

[krystal_blade]

I really can't see an end to circuit switched networks any time soon. The switch to Packeted data is fine for most commercial traffic, but there are a few areas that will continue to require "locked in" circuits as opposed to packet buffering systems.

There is a lot of value in the use of packetized data. More "lines" over fewer trunks is just one of them, and for your average, everyday user, they will not notice the difference.

On the other hand, certain timing based encryption schemes will have to remain on locked in circuits to function. The latency caused through the use of packet buffering regardless of how slight, may be enough to cause a "handshake" failure, or just spew unintelligable garbage.

Of course, as encryption systems become more and more robust the need for "hard lines" will start to dissipate.

I for one welcome our new packetized telephone overlords...

krystal_blade

Re:Circuit Switching will still be around...

[Cynikal]

"Of course, as encryption systems become more and more robust the need for "hard lines" will start to dissipate."

but if they cut the hard lines, we'll be trapped!

Already going on .... elsewhere

[Anonymous Coward]

there is nothing new to this kind of thinking.
there is a paradigm shift going on through out the telecom voice/data industry ... moving from pay-per-connect (time) to pay-per-use (bandwidth)

such a move is already being made in the "test-bed " countries like CHINA and INDIA where the telco network is newly gaining great importance in infrastructure.

already in INDIA the BhartiGroup and RELAINCE (2 main telco operators) have their backbones as IP based traffic.

it seems strange that such a move made in such "developing " contries is only being taken up in the US.

That's cool but...

[GreggyBUIUC]

... I still doubt I'll be able to make calls on my Sprint PCS phone between 9-10pm. Once 'nights' kicks in in this town the network is perpetually busy.

Quality

[Anonymous Coward]

Won't this be a loss of Quality? Bandwidth is money is also true for telephone companies. I used some cheap numbers for inernational calls an always noticed a few lost packets. Thrifty compression algorithms will do the rest...

Damn!

[HornyBastard77]

If they get rid of dedicated circuits then how am I going to get out of the Matrix anymore?

Re:Damn!

[Anonymous Coward]

in packets

Re:Damn!

[Shinobi]

In bits and bytes, I guess......

beginning of the end for circuit-switched networks

[evilviper]

It seems that technology moves in cycles...

"Serial is slow, let's move to parallel."
"Now parallel is slow, let's go to serial."

It all started with central, time-sharing systems, then switched to distributed computing when the technology permitted, and there now a trend torwards centralized administration again.

Batch processing was popular, then on-line processing replaced it, now many things are going back to batch processing because of the time/cost advantages it provides.

It seems that as technologies disappear, even newer technologies come along that remind everyone of the (still) very valid why they were using the older technologies in the first place.

Just wait, in 5-10 years, CRTs will be popular once again, and I suppose circuit switching will probably find a new foothold as well.

bandwidth?

[nich37ways]

Assuming at some point in the future voice calls will be via "packets" from end to end, instead of just from the exchange to exchange, what kind/amount of bandwidth would/does reliable conversatinons require.

Anyone using this type of service already, how good is the quality on it? Also how does the phone connect to the providers server?

Looks very promising and should hopefully lead to at least a freeze in the cost of phone calls, and hopefully a steady decrease.

Re:bandwidth?

[pe1rxq]

You can compress voice down below 20kbits and get a quality comparable with analogue phone lines.
I believe GSM uses 13kbits (or in that neighbourhood) and I have used the speex codec in 8kbits mode recently and it give good enough quality to make conversations.
If you use uncompressed audio you need about 32kbits.

Jeroen

Re:bandwidth?

[ZenJabba1]

Uncompressed Audio (g.711) = 64k channel total for transmit and recieve

Re:bandwidth?

[Anonymous Coward]

If they do not compress it (which hopefully they do not) 64Kbps is the number -- but in order for it to sound right, the 64Kbps has to be delivered with an end-to-end delay of less than 250ms. Preferably *much* less. That's the point at which it starts sounding like you are talking to Neil Armstrong.

With compression in the mix it is very hard for IP based networks to efficiently transmit voice because the packets have to be too small in order to arrive on time. When the packets get too small IP based networks suffer a lot of overhead. This is why a lot of networks are frame relay and ATM. (For voice ATM has a constant 10% overhead using AAL0, and frame relay can get slightly lower even than that.) Both ATM and some Frame Relay systems allow very small slices of both data and voice to be interleaved (for ATM the payload is 48bytes/cell, so the latency is the transit time + buffering time + 6ms for uncompressed voice, and in correctly built ATM networks the buffering time is very low as well.)

Personally I do not look foward to voice-over-anything, with the possible exception of ATM (which incidentally should *not* be referred to as a packet based technology). I can easily see huge meltdowns of the telephony core -- I'm sorry but even as a data guy I can see that packet systems aren't up to snuff with circuit systems for reliability. Plus it is too tempting to the telcos to use compression, which can mess up modem and telecommunications devices for the deaf. The whole business of fitting more calls over less trunks is better addressed by making the trunks wider using modern optics.

Yeah but the question .on every /.er's mind is ...

[R33MSpec]

Will Sprint use the Evil Bit?

apt-get is better than packet switched telephones

[Debian Troll]

Sprint announced that it has 'begun transforming its telephone network so voice calls are transmitted in packets

This is a very interesting idea indeed. There are many skills and lessons which have been learned in the computer-related packet networking field which will be able to be translated to achieve enterprise-level synergies and produce a best-of-breed telecommunications system. I would like to point out, however, that a packet switched telephone network will not be able to be upgraded using apt-get. This is a serious problem. If we use Debian systems to run the network, just imagine how easy it would be to upgrade to video phones when they become available?!?! Simple...just apt-get install videophones. You couldn't do that with an RPM based telephone network.

Ever hear of apt-rpm?

[Anonymous Coward]

Maybe you should apt-get your trolling update.

About Freaking time

[zakezuke]

Part of my major complaint in the 21century is the fact that modem technology has not really improved a whole hell of alot. Sure we have Cable and xDSL service if you are lucky enough to live near a place where they saw fit to actually upgrade.

When I see stuff like this, I get this warm happy feeling inside when it seems like it's actually a *good* idea to actually upgrade from our old vintage phone system to something that can do a hell of alot more useful things. Datapackets can be uniquely identified as "voice" "fax" or "data", which could in theory make a whole slew of things possible...

Though it makes me wonder, if the telcos are going for packet based voice communications, why the hell would I bother placing a long distance call through them when I can use VoIP software. Don't get me wrong, i'll all for the idea digital packet based phone service, if for nothing else but making all phones with that service high speed internet ready.

Re:About Freaking time

[Drakonian]

Just because it's IP (which it may not even be in this case, I think it's ATM) does not mean it's connected to the Internet. It's a bunch of private lines owned by the Telco.

But...

[Anonymous Coward]

does it run linux?

But will this benefit the consumer ?

[Anonymous Coward]

Nice idea and all that but how will the consumer benefit from this ? will we get lower call charges or will the CEO just get another 5million on his paypacket ?

Re:But will this benefit the consumer ?

[zakezuke]

Consumer benifits....

1. One phone number for multi devices (I think this was covered in the article).

2. Phone numbers not tied to physical location, but rather device or authentication. Would be most nifty for mobiles to go landline. (this was covered)

3. Multi communication... end users could in theory have two telephones, and place two calls on the same line. No further need for an alarm wire from your telco.

4. No D/A loss when you copy your CD over your phone.

5. Everyone is highspeed internet ready... in theory you need 32Kbit for decent voice, perhaps 64K / 128K bit just to be safe. Pay more money to throttle you up to internet speeds... no more waiting for low paid installers.

6. Networked appliances no longer need "internet access" but rather phone access, and no gay ass 300 baud modems in your digital cable box.

Re:But will this benefit the consumer ?

[blastedtokyo]

Not so fast. This is Sprint's long distance network. If your local telco's still using tin cans, it's not going to enable any of this. What it does mean is that the next time there's a SQL Slammer or other bug clogging the web, you're phone service is down too. I bet they're getting funded by corporate IT and helpdesk staff.

Re:But will this benefit the consumer ?

[241comp]

No, this isn't Sprint's long distance network. It is Sprint LTD (local telecommunications division). It also doesn't mean that any bug clogging the web will bring down service - this is not voice over the Internet. In fact, it is not even voice over IP. It's voice over ATM.

Re:But will this benefit the consumer ?

[ptbarnett]

3. Multi communication... end users could in theory have two telephones, and place two calls on the same line.

This was Sprint ION. DSL and up to four voice phone lines over the same wire.

Sprint pulled the plug on it last year, after losing a lot of money. They had some deployment problems, but users able to get it were reportedly happy with it. They shut it down just before it was scheduled to be deployed at my CO.

From what I've been able to gather, a large part of the problem was Sprint's CLEC status in most areas. The ILEC would tell an interested consumer that the line wouldn't support DSL, and cancel the order. Then, they would offer DSL to the customer a few weeks later.

Re:But will this benefit the consumer ?

[WaysideWeasle]

Actually, most of the telcos today utilize a packet-switched technology for their long-distance networks. What Sprint is doing is extending that to the local networks to the switch level. Sure...from the local office to the customer is still copper and circuit-switched to a degree, but it creates more efficiencies in network management and paves the way to bigger and better things. Clarity is improved greatly.

It'll never work

[cordsie]

What happens if voice IP traffic gets mixed with, say, a few Quake deathmatch packets? What happens if a bot starts taking railgun shots at bits of your conversation with your Mother? Or if a L33T D00D pulls down the grid for an entire city with a strategically placed rocket? I want answers, damn it!

Re:It'll never work

[noselasd]

Very simple. It doesn't get mixed in. These networks are (ofcourse) private and the traffic on it are controlled. Meaning yoo don't get access to it unless you dig up the fibre yourself and use fibre splitters on them.

Re:It'll never work

[Anonymous Coward]

Very simple. It doesn't get mixed in. These networks are (ofcourse) private and the traffic on it are controlled.

You mean there are big private networks like the Inter-web? Why don't they connect to the Inter-web and then just use that, it is free! Sprint could run their traffic through a Roadrunner cable modem or something.

Re:It'll never work

[Kevin Stevens]

I would be more concerned if bits of typical Kazaa traffic got mixed in your conversation with your mother ;).

it's confusing

[Sarin]

So when you use a dail-in account with a sprint-network, everything will be converted to packages first, then send to your provider over the internet, converted back to analog, converted to digital and then you have your connection on the internet?

Re:it's confusing

[LarsG]

So when you use a dail-in account with a sprint-network, everything will be converted to packages first, then send to your provider over the internet, converted back to analog, converted to digital and then you have your connection on the internet?

Yup. All reasonably modern telcos today have digital exchanges. Telenor, the old telco monopoly in Norway, started the change in 1986 and the last analog exchange was replaced in 1997.

Modem - Analog line - Local telco exchange - digital through telco network(s) - Local telco exchange - Analog line - Modem

Ah, but what else is out there?

[ites]

Unlikely that this will do anything except increase their margins. Established players do not generally like to change the rules of their markets. So, this is like saying "Sprint move from copper to fibre". Yeah,... international calls going to be any cheaper? Forget it.

Somewhere out there, someone is building systems that _will_ have a significant impact on telephony. Find these guys - that will be news.

Intellectual property, or real property?

[User 956]

Sprint Moves Phone Network to IP

Why, was property in the real world too expensive?

[rimshot]

Thanks, I'll be here all the week.

say what?

[djupedal]

Sprint who?

phonIE payper liesense felons move..

[Anonymous Coward]

hole FUDgePacking(tm) festival to /./va lairIE's decaying pourtoll.

lookout bullow, the daze of the Godless hostage ransom stock markup frauds, is WANing into coolapps.

consult with yOUR creator.

IP?! Or ATM? Or something else?

[nmg196]

I presume the insertion of "IP" in the title of this article was a mistake or assumption made by a naeive author? You don't use IP to carry telephone calls on a phone network. Ever. IP is no good for carrying voice data and there are many better protocols around which were designed for this purpose. I presume that they really mean ATM or some other voice protocol? You need a small packet, circuit based protocol to handle large numbers of voice calls efficiently. Although IP could be made to work, it would be pretty difficult and is essentially rather like trying to put a sealed bus in a sea to try and make a ferry - why not just use a boat?

Nick...

Re:IP?! Or ATM? Or something else?

[thebigmacd]

You don't use IP to carry telephone calls on a phone network. Ever. YOU don't but Telus and Sprint do. Where have you been for the last few years? VoIP.

Re:IP?! Or ATM? Or something else?

[virtual_mps]

Dude! The 20th century called and it wants its ATM back!

Re:IP?! Or ATM? Or something else?

[richard_willey]

I beg to differ.

Voice over IP is already standard part of corporate IT. It is rapidly leaking into the consumer space.

Historically, Big Dumb Pipes have continually displaced managed bandwidth type systems. Voice over IP is just the latest example.

I did a consulting project for Qualcomm as part of my classwork last semester studying whether 802.11b has the potential to disrupt CDMA networks. People might find the paper interesting, since it indirectly addresses many of the same issues.

http://web.mit.edu/~rwilley/www/Qualcomm.pdf

Re:IP?! Or ATM? Or something else?

[noselasd]

And the quality of VoIP ? Unless the person you call sit in the office next to you. Have you doene som qos testing on voip ? it utterly sucks.

Re:IP?! Or ATM? Or something else?

[dolt]

It works for the Italians,
"Telecom Italia to use IP network for 50% of European calls".

There is also VoIP-over-cable, and VoIP-over-cellular, and VoIP-over-Ethernet.

The fundamental problem with ATM is that it doesn't go anywhere -- you can't get ATM over Ethernet, over cable, over satellite, over 802.11, and so on.

So, with ATM, when all is said and done and you have built a big Voice-over-ATM network the end result is the same closed, designed-only-for-voice network that is the PSTN today.

I, for one, would like any-to-any TV-quality videoconferencing, exactly like I have with any-to-any voice today -- I can call anyone in the world with a phone. I can't call anyone in the world with video. From there, there are lots of interesting services - read some scifi to get a fill of cool ideas.

Re:IP?! Or ATM? Or something else?

[Anonymous Coward]

Nope. No mistake. Most of the big telco switches talk IP/ethernet already in addition to ATM/SONET.

We're not talking about routing over the public internet, but over private WANs. The local connections are usually ethernet, the long-haul stuff is usually fiber. Of course this allows for standard ATM, IP-over-ATM, or even IP-over-SONET.

I work on a compact VoIP softswitch that only talks IP/ethernet (although we have another box to translate to ATM/SONET for legacy interworking).

Re:IP?! Or ATM? Or something else?

[Abcd1234]

I presume the insertion of "IP" in the title of this article was a mistake or assumption made by a naeive author? You don't use IP to carry telephone calls on a phone network. Ever.

Umm... no. Many telephone companies are moving AWAY from circuit switched networks to packet switched ones (whether that be IP or some other protocol). Why? First, it's cheaper (you can use more commodity hardware, especially if you're using IP). Second, it can be more efficient, since traffic can be re-routed to make better use of your network resources (ie, real-time QoS and traffic engineering). Third, a packet-switched network can be more reliable, since there are no single points of failure... if a node goes down, just route around it (which also means less need for incredibly expensive, massively redundant network nodes). Fourth, since data (as opposed to voice) is becoming more and more important, a packet switched network makes it FAR easier to move data AND voice through the same backhaul network. Fifth, if there really are scenarious where you want a "circuit-switched-style" transport (for whatever reason... perhaps to provide a guaranteed pipe for, say, telesurgery), you can achieve this using things like MPLS, meaning you can get the best of both worlds.

Re:IP?! Or ATM? Or something else?

[hamanu]

parallel analogy: You can't fit a C130 cargo aircraft in the back seat of a car, so having a C130 doesn't get you anywhere.

criticism: you CAN fit your car in the cargo space of a C130, so it DOES get you somewhere.

ATM is a great protocol and technology. It missed it's ideal adoption time, but you can expect to see it re-incarnated in later tech. You can send IP over ATM, voice over ATM, ethernet over ATM, et cetera. ATM is the best "bearer channel" solution we have today. You don't need end to end ATM to get a benefit from it, because it's great at interfacing to other kinds of protocols, which is why sprint (a big company) has decided to build their core network out of ATM, and you (an individual slashdot poster) probably don't understand it enough to criticize that decision.

Re:IP?! Or ATM? Or something else?

[thebigmacd]

I never said it didnt suck. The parent said it ISN'T USED, which is DEAD WRONG. Many networks partially use VoIP and obviously Telus and Sprint are moving to full VoIP. IT isnt used? Do you even read the threads?

Re:IP?! Or ATM? Or something else?

[Anonymous Coward]

All I can say is god help me if I ever have to receive remote telesurgury over MPLS.

VoIP technology is having a heyday because the systems it is running on have yet to become old, oversubscribed, and underbudgeted. When that happens, there will by much wailing and gnashing of teeth.

Wiretapping

[Anonymous Coward]

The really fun thing about this means that any router can be told to simply copy every packet in a particular conversation to law enforcement.

Re:Wiretapping

[miu]

The really fun thing about this means that any router can be told to simply copy every packet in a particular conversation to law enforcement.

That is sort of the point of the Communications Assistance for Law Enforcement Act .

That particular battle was fought and lost 9 years ago.

Re:Wiretapping

[zwoelfk]

Possibly. But VoIP also makes it a simplier task for the end users to install encryption. VoIP over SSH. That would be useful.

Re:Wiretapping

[miu]

But VoIP also makes it a simplier task for the end users to install encryption.

But probably not in this case. To "leverage existing infrastructure" this will not be a user visible voip application. One or more endpoints and transport networks involved will use existing circuits for quite some time.

Re:Wiretapping

[barbazoo]

I don't think that packet relay is harder nor simpler to tap than circuit switched. It's all digital anyhow.

DSLers 'dedicated' line just got clipped?

[Anonymous Coward]

So does this mean now that DSLers can't claim that they are on a dedicated line instead of sharing bandwidth with the neighborhood? It's basically their favorite jab at cable users and i'm sure they will be in denial over this one...

Re:DSLers 'dedicated' line just got clipped?

[djnichol]

All bandwidth is shared.

Re:DSLers 'dedicated' line just got clipped?

[Kredal]

DSL shares bandwidth too, before it gets to the final switch and out to their house. Same effect, just different way of doing it.

Question for Germans !

[Anonymous Coward]

I am French and I am wondering how you can bear that lot of American basis are still on your soil ? Are you really independants ? Why don't you kick the ass of the Americans like us thanks to De Gaulle ? Do you scare the Americans ?

Packet Switched Voice is not the Internet

[Detritus]

Although they will probably use IP to transport voice data between their switches, that does not mean that any of the data will travel over the public Internet or that the end users will use IP. All this does is change the design of the subnet used to transport voice data in between toll switches, not the interface between the toll switch and the end user.

Re:Packet Switched Voice is not the Internet

[4of12]

their switches

That's got to a key ingredient. The ingredient that probably caused other posters to think - "No Way can this work!"

If voice service were conducted over some of the public pieces of the internet that I use, the latency chop effect would make Max Headroom sound smooth.

And in the UK

[Zog The Undeniable]

British Telecom's main fibre optic backbone is a packet-switched network. It's only the "last mile" that's truly circuit switched these days. We have a fairly modern telephone system in the UK, only hampered by stupid area codes based on centres of population rather than numbers of people as in the US (so the big towns such as London, Bristol, Reading and Leicester ran out of numbers quickly and had to have their codes changed). To be fair, no-one anticipated fax machines and data connections when the coding system was decided.

VOIP end to End

[slashnik]

All very nice but what happens when one VoIP office needs to talk to another VoIP office over the packetized carrier. The analog voice data will be packetized in the first office to pass over there LAN converted to analog for the connection into the carrier who convert back to packetized data to cross this carrier, converting back to analog to interface with the second office who the re packetize to cross their LAN. With all the buffering required to do this what's the quality going to be like at the far end and will it sound like I'm calling from the moon?

Nortel vs. Cisco

[Disco2k]

Wow! I was just about to post to the now inactive Telus using VOIP discussion since I spotted this article from The Globe and Mail about Sprint.

I found it interesting that Telus, a Canadian telco, will use equipment from American companies Cisco Systems and Juniper Networks whereas Sprint, an American telco will use equipment from Canada's Nortel.

I have nothing particularly insightful to say right now. Talk amongst yourselves ;-)

hope they have echo cancellation

[blastedtokyo]

Sprint with packet switching. So clear, you can hear a pin drop...twice.

No, the question .on every /.er's mind is ...

[hummassa]

Can you imagine a Beowulf cluster of these?

Not IP but ATM

[geirt]

The Sprint Press Release states that they are going to use ATM, not IP.

Re:Not IP but ATM

[dmccarty]

Yeah! That's, like, great, dude! Now we can use our phones to withdraw cash!

Some questions/observations

[whovian]

1. This is only going to help drive down cell phone prices, helping to make land lines obsolete once again. (This was likely to happen eventually anyway, right?) If the article is right and you can forward your conversations to any IP address (at an extra cost), the primary advantage cells have are mobility, such is the case now. Until, that is, when you think about wireless solutions and VoIP. Mmmmm...DHCP + VoIP :)

2. A brief search of the web suggests VoIP can be more secure than traditional telephony. To what extent will government fight this? Effectively having an SSH tunnel to the other caller wouldn't be appreciated by the gov't given the present modus operandi of the US.

3. VoIP is certainly a logical progression, and I don't see the big telcos going out of business soon. Where I live, there are just a few DSL providers but only one company (SBC) owns all the wires into the area. Their only real competitor is cable TV whom they are fighting tooth and nail to gain marketshare. I imagine access to wireless frequencies has very little competition (think: 802.11), but will there need to be legislation to keeping it open?

Re:beginning of the end for circuit-switched netwo

[w1r3sp33d]

The first telco wiring was bundles of copper lines, one pair per phone call (anyone see that picture of San Fran circa 1930 where you nearly couldn't see the water through the lines?) The second phase of telco life was TDM, one wire pair=one T1 with 24 paths and a path was created on the fly across one of these "channels." Now packet switching city to city is really the current answer, I don't see us going back to installing thousands of miles of copper in a city or installing TDM technology anywhere but to the doorstep.

I have been installing VOIP, VOFR, and IP Telephony for years now for many businesses, I have lots of 99.999% uptime systems, no complaints in almost two years for quality of voice, I can't believe /.ers are amazed and puzzled by such simple things as a forty year old idea being used by a carrier. I guess /. isn't what it used to be.

Plenty of voice communication is already...

[AnEmbodiedMind]

Have you ever used a cell-phone on a modern cell network? It is 'packetized' Also, packet networks are slightly more secure as it is more difficult for even the cell company to bug you... They have to find and reassemble from the network all packets comming and going from your phone.

Re:Plenty of voice communication is already...

[amorsen]

Yes, it is really hard for the cell phone company to find and reassemble all the packets in your conversations. It is not like they have to do it anyway in order to be able to connect your call to the intended recipient. Or something.

Re:Plenty of voice communication is already...

[AnEmbodiedMind]

Sure, at the final switch before the cell site all the packets are in order and about to go to the phone.

That doesn't mean that they necessarily have built in the ability to tap into arbitrary calls from their central offices.

The fact that the US government is funding carriers to implement the required changes to their networks to ALLOW wiretapping, makes it pretty clear that it can be difficult for cell phone companies to eavesdrop on calls if they didn't build it in in the first place.

Check out this for one article discussing this...

Quote:

Despite the somewhat torturous path CALEA has taken toward final implementation, it isn't supposed to be an unfunded mandate on carriers -- the U.S. Department of Justice is supposed to pay for some of the required upgrades out of a US$500 million fund.

Re:Plenty of voice communication is already...

[amorsen]

It seems from the article that they are talking about conversations wrapped in VoIP or going through long distance services. I really cannot imagine that tapping your very own packet network such as GSM or CDMA presents a problem. It gets more challenging when customers use data calls to carry voice. Anyway, GSM is easily tappable here.

Also, cell phone companies keep track of which antennas a phone is in contact with at which signal level all the time. So if you carry a cell phone that is turned on, the phone company knows where you are.

How very 1960s...

[Simon Brooke]

The French telephone network has been entirely packet switched since at least 1968. Oh, well, America gets there in the end...

Re:How very 1960s...

[noselasd]

No, it hasn't. They use(at least when we were there with ISUP monitors and simulators) standard equipment on rather standard 2mb/s systems. The signalling is ofcourse packet switches, as in all SS7 networks.

Re:apt-get is better than packet switched telephon

[djnichol]

Enterprise-level synergies ... best-of-breed ... [Puke]

Re:beginning of the end for circuit-switched netwo

[OmniVector]

CRTs will be popular again? Why?
Newer LCDs are cheaper to manufactur, offer color ranges just as good (and in some cases, i.e. the apple 23" cinema displays, better color ranges), lower power consumption, brighter backlights, and smaller desk space consumption. About the only problem LCDs haven't tackled is latency but give it time.
Of course hopefully OLEDs will get far enough that LCDs will just drop off the planet.

Re:Already going on .... elsewhere

[uspsguy]

Gee! Isn't it amazing that countries that don't have an infrastructure would be building one using current technology and a country that has a huge, solid, working one would be a little slower to convert to something new. Because the US was an "early addopter" of telephone technology, we're a little slower upgrading but we've been talking all over the country for a loooong time.

PCS laptop cards

[bmongar]

I wonder if this is related to the comercial I saw this morning with sprint now selling pcs laptop cards for wireless internet connectivity.

In Soviet Russia...

[Anonymous Coward]

Beowulf clusters YOU!!!

Re:beginning of the end for circuit-switched netwo

[Citizen of Earth]

"Politics is a pendulum whose swings between anarchy and tyranny are fueled by perpetually rejuvenated illusions." -- Albert Einstein

OH CRAP!

[subreality]

THE'YRE CUTTING THE HARD LINE! Get out now or you'll be trapped in the matrix!

Uh, Yeah

[Anonymous Coward]

AT&T outright said that they won't be buying any more class 5 gateways - They said that 3 years ago, instead moving to packet switched networks.

Slashdot seems to think that this move is new. It's been going on for 3-4 years.

Friends?

[Anonymous Coward]

Can we put our friends and family on LAN-specific addresses? It would be nice to have:

MOM 192.168.0.10
GF 192.168.0.15

etc.

Hidden Agenda

[gr8_phk]

For you conspiracy buffs. Short term, VOIP requires more bandwidth for the same voice quality due to the overhead. What it does allow is variable bit rate compression instead of a fixed 64kbps. Once they start compressing close enough to the home, those with dialup will be screwed and switch to DSL or cable. Or, perhaps something else ;-)

reading /. tends to make me think in this way :-b

VoIP

[primal_rage]

I wonder if they will use any microsoft software. I would really hate for my telephone to blue-screen. My phone lacks the buttons.

good, now I won't get all circuts busy

[the_2nd_coming]

when I try calling some one.

Thanks, Communicatiosn Conglomerates

[Anonymous Coward]

Ah, its so nice to see major telcos moving to VoIP so they can save money, after they have worked so hard to prevent us from using VoIP inless they are the ones getting the savings.

booo.

Doesn't say IP

[ca1v1n]

Nowhere in this article does it say they're doing anything over IP. If I had to take a shot in the dark, I'd guess they were using something like ATM, which has all sorts of wonderful properties for real-time communication, and more closely resembles circuit switching (in the good ways) than any kind of IP connection. I've heard rumors to the effect that IPv6 allows for some of these properties, but no form of IP will ever do what ATM does, for lots of very good reasons.

I think Telus is nuts to use IP. I hope they succeed, but I still think they're nuts.

Re:Doesn't say IP

[Abcd1234]

more closely resembles circuit switching (in the good ways) than any kind of IP connection.

You haven't heard of MPLS I take it? It's pretty cool technology. Basically, if you set up an IP cloud with MPLS-aware routers, then you can use this network for standard packet-switched communication, or you can establish (in real-time) "virtual circuits" through the network for high speed, low-latency communication. Thus, you get the best of both the packet switch and circuit switched worlds.

I won't bother debunking your assumption that telcos aren't using IP for their backhaul network, since there are enough comments in this forum to disprove that erroneous assumption.

Background info

[vestus]

See Nortel UEMG9000 for info.
I used to work on this, and can say that its quite a robust system. Runs about 8000 POTS lines or 2000 xDSL, and also supports DS1 and TDM lines. Backbone is OC3 ATM with other options available. VOIP should be done now/soon but I don't believe Sprint went that route. The system has Echo Cancellation and all the other required perks to ensure good quality.
Used to.. Anyone need an embedded driver dev in RTP?

No, I haven't seen the picture.

[Gallowglass]

Do you have a link, by any chance?

please....AT&T doing this for years

[rdejean]

AT&T and MCI have been doing this for years (using IP on their voice backbone) using "SoftSwitch" technology from 3com Commworks. It wasn't marketed to the general public since it gave the big boys a 'competitive advantage', saving them millions per month in tariffs.

Now that Commworks has been sold off, 3com still owns the technology and plans to market it. It has been in production over 5 years, and i'm hearing that it puts anything from Nortel or Cisco to shame. And it runs on Solaris, hehe..

I'm bored.

[windex]

>tracert 2001:630:1c0:1:201:2ff:fea9:9ae0

Tracing route to chaz.ws.ipv6.ne-worcs.ac.uk [2001:630:1c0:1:201:2ff:fea9:9ae0] over a maximum of 30 hops:

1 3 ms 2 ms 3 ms 3ffe:b80:1c5d:1:200:cff:fe3e:d15a
2 72 ms 137 ms 126 ms 3ffe:b80:3:5d89::1
3 97 ms 77 ms 66 ms tu-viagenie.ipv6.noris.de [2001:780::b]
4 157 ms 179 ms 197 ms 3ffe:b00:c18:1017::2
5 467 ms 465 ms 454 ms 3ffe:2100:1:9::c13f:5e06
6 * 529 ms 516 ms 2001:630:0:f006::2
7 463 ms 472 ms 464 ms int-ipv6-gw.ne-worcs.ac.uk [2001:630:1c0::3]
8 451 ms 456 ms 449 ms chaz.ws.ipv6.ne-worcs.ac.uk [2001:630:1c0:1:201:2ff:fea9:9ae0]

Trace complete.

SCO...

[Cobralisk]

Is suing Sprint for violation of its IP. ;)

I hate IP phone calls

[ooglek]

I have an issue with IP based calls. Now granted, Sprint may not stop/reduce the number of packets sent when I'm not talking, or there is a pause in the conversation, but I HATE it when I can't hear the background noise -- the ambient noise that always occurs on the other end. Stuff like another person saying something. Analog? I can hear them. Digital/IP? Most times either I don't know the call is being IP packetized, or I hear absolute silence and then maybe it cuts in. It's like a friggin' speakerphone.

If Sprint goes IP (and maybe a lot of other providers already have too) I sure hope they don't do compression of the signal to the point where I hear nothing (not even ambient noise).

I still think analog is better -- audiophiles will agree.

CISCO VOIP phone

[SuperKendall]

In my office for the last few years I've had a Cisco VOIP phone - the only connections it has are power and network. It works pretty well, and that's an older phone.

Voice over IP is the way voice traffice will be handled in the future. The article talks about sending IP traffic over ATM, at least for now - it's more expensive that way but the cost of new switches is also quite high.

You might want to read more on the "Martini Draft" and MPLS to get a sense of how ATM will be replaced by IP technologies.

Packet based

[SuperKendall]

But that is packet based (IP like) traffic over ATM. They are just using ATM because it will take them a while to replace the ATM switches - elsewhere they say about 14 years until they have a "native" packet based network!

Re:Packet based

[Anonymous Coward]

SuperKendall wrote:

But that is packet based (IP like) traffic over ATM. They are just using ATM because it will take them a while to replace the ATM switches - elsewhere they say about 14 years until they have a "native" packet based network!

The subject of this slashdot story is: Sprint Moves Phone Network to IP

the internet IS circuit-switched!

[psycho]

> Is this the beginning of the end for circuit-switched networks?
This shows a deep lack of understanding of the real nature of the internet (yeah, well, its slashdot, i know). Most core backbone IP networks are layered over circuit switched networks, thanks to high-speed ATM technology. In fact, there is some work by McKeown et. al, which i am too lazy to cite now, that shows the backbone core of the Internet is (underneath) primarily circuit-switched.

Re:the internet IS circuit-switched!

[Anonymous Coward]

If true, and I have heard similar claims regarding L3's so-called IP network being based on ATM, then it underwrites a big problem. ATM uses frames, IP uses packets. Bit sawdust (the percentage of frames unoccupied when packets are carried over ATM baselines) runs about 56%.

Time to look at this in greater detail and check up on some emerging resilient Ethernet tech. Have fun!

/.ing

[dark-br]

Cool... I suppose now we can slashdot the phone network!

My sister have been trying that for years but still no effect :P

Wired in 1996 on ATM and IP

[Jasn]

Boy, does this take me back. One of the better articles in Wired magazine's history, in my opinion, was the one on "Netheads vs. Bellheads" highlighting an internal battle at Sprint as a microcosm of a bigger battle of packet-switching vs. circuit-switching. It's long but entertaining and worth it, talking about philosophical differences and ATM and IP.

... could this news be more old? ...

[ninjagin]

Packet-switched telephony is old hat. It hasn't had the kind of broad-based consumer support that it probably could have, but businesses have been using enterprise-scale packet-switched telephony for more than a decade. Businesses like Clarent, Sonus and Siemens have already carved out the space for current VOIP and packet-switched technologies. Some countries with developing communications infrastructures are skipping circuit-switched telephone systems entirely and moving straight to packet-switched backbones.

They tried to offer consumer VOIP over CATV a couple years ago here in the Denver area, but this was also back when Qwest was "riding the light". Consumers liked the idea of one bill, but didn't like having to depend on a single incoming cable, and a single company (AT&T) for everything. There also wasn't much savings. (It all smacked a little of the time before the Bell breakup.) Given that the local cable company has changed hands about every five years, consistency in service may have also been a concern.

Has anyone around here got/tried VOIP over cable? I'd be interested in knowing what the experience was.

Time Warner telecom made a handy bundle (and probably still does) on installing SONET rings for businesses to use for putting their phone calls on packet-switched networks -- all to cut the cost of long distance down to a fraction. As someone mentioned in another post, you can get the same 99.99 quality from packet-switched telephone systems as you get from circuits, and two stage dialing (not to mention billing metrics) is much easier to manage.

Clarent, the early leader in VOIP gateways (now under new ownership, btw), has got some great softswitch technologies (Class IV and V call managers) that do SS7, H323, and a bunch of other protocols. Having spent more than my share of time prodding 5ESS and Datakit, it's comforting to see softswitches taking more of the burden, and circuits being freed for other uses.

I like reading /. just as much as anyone, but this post wasn't terribly exciting and didn't say anything new. There's so much that really is exciting in telecom that I'm led to believe that this one must have snuck past the front line. Hopefully we won't see forthcoming stories about how "Graphite powder reportedly makes Abacus calculations faster" or "Fifty ways to soup up your slide rule".

-- ninjagin
"640K ought to be enough for anybody" - Bill Gates

It will take 15 years to upgrade network...

[Anonymous Coward]

So by then no one will use "old" phone networks like that. Everyone will have a cellphone, and all "land lines" will be just for net access. :P

Re:It will take 15 years to upgrade network...

[interstellar_donkey]

all "land lines" will be just for net access

Naw. In 15 years all net access will be wireless over 802.11r. Though service will be sketchy since no vendor will be able to agree on a set standard.

VoIP != ATM

[RedStapler]

Ummm, this has nothing to do with VoIP. VoIP is not ATM. ATM can be used to transport VoIP but the two are entirely separate technologies.

FYI, ATM is about as close as you can get to circuit-switching and really is not a phenomenal leap forward (for the telcos maybe but in terms of network technology, not really). While ATM does virtualize the link to allow for packets, it still has the notion of virtual circuits/virtual paths. ATM very much follows the circuit-switched paradigm of yesteryear, just with a different flavor.

ATM was all the rage about 5-8 years ago, best of luck trying to do any cutting edge work with ATM right now.

IP Phone...hmm...

[Cynikal]

dial 555-3175
555-3175 contacted, waiting for reply
Ring... Ring... ... Request timed out.. Ring..
We're sorry, the customer you dialed connot be reached offline, would you like to connect or remain offline?

----
my question is if your phones crap out and you need to reinstall tcp/ip on it, how do you call tech support?

the last 1/2"

[interstellar_donkey]

First the backbone goes to a digital packet switched network, with A/D conversion going on when the local telco tries to go outside. Then the local telco goes all digital, with the A/D going on when the copper meets the CO. Next we are going to see A/D happen right at the telephone (already happening with some systems).

The only logical progression is to pull analog out of the loop completely for the last 1/2" between the phone and the user, bypassing the soon to be obsolete analog audio output "mouth". The only technological hurdle for doing this is to figure out how to install the interface into the antiquated standard form-factor "skull" case most end users still lug around with them everywhere they go. Perhaps we should look into a new IEEE standard for that and start adopting it now.

GSM

[yerricde]

I don't see us going back to installing thousands of miles of copper in a city or installing TDM technology anywhere but to the doorstep.

What about putting up huge antenna towers and running TDM through those? Some digital mobile phones are based on time-domain multiplex technology. As far as I know, GSM is like this, carrying voice and data in burst signals. (Other digital cellphones use code-division multiple access, or CDMA.)

LPC vs. background noise

[yerricde]

You can compress voice down below 20kbits and get a quality comparable with analogue phone lines.

Voice typically runs at 8000 samples per second. Compression of such a signal below about 24 kbps is typically based on a complicated linear-prediction model. Linear prediction, especially at GSM bitrates (13 kbps), has trouble with background noise.

Re:Not IP (Sprint did the gateway in the house)

[fazzumar]

Anyone remember Sprint's attempt at merged services, called ION? They installed a router in your house and all of your communication needs ran out of that box. The router connected to Sprint's network and ran voice and data over ATM. It was a very neat idea and might have worked if there were more early adopters (and if they could have resolved all of the installation issues... I worked on part of the software that brought together the connection allocation, tracking and billing systems, and it was extremely complex)
I didn't subscribe to the service because the cost was prohibitive for personal use, but it seemed pretty cost effective for business use (4 phone lines and a T1 downstream, 256kbit upstream for $200 a month [I think] and that included several hundred minutes of free long distance per month).
I don't believe the subscribers had to change their phones to allow that to work, they just had to have Sprint's router in their house that all of their phones tied in to.

Re:Already going on .... elsewhere - in the US

[Skleed]

Yes - it is going on elsewhere - try right in the good old US. I do some work for a rural phone company (30,000 lines) that has an entirely packet switched network. They did not make any big announcements, and I bet that 99.99% of their customers don't know or care. Because the telco can control the QoS, the quality is at least equal to the traditional infrastructure.

Many of the rural telcos adopt these types of technology more quickly than the baby bells. How? You and I are paying for it every time we pay our phone bill - Universal Service Fees have subsidized a tremendous improvement to rural telephone service.

Yup.

[mindstrm]

And it should be.. why should they be forced to use ancient and expensive technologies on the back end?

This isn't even super high tech or anything.. it's just that phone companies have a level of service to provide, and making big changes takes big planning.

WE waste way too much bandwidth on voice.

Re:beginning of the end for circuit-switched netwo

[cpeterso]

Circuit switching will probably return under the guise of "guaranteed quality of service" over the unreliable packet switched network. In fact, ATM does virtual circuit switching already.

voice over ip? Why am I paying for long distance?

[bryan314]

If I can send an email across town and across the world for the same price. A local phone call. Why is it a local call and a long distance call cost different?

Voice over ip. Charge by bandwidth usage. A call across town should cost the same as a call across the world.

Anybody?

Bryan :)

Re:beginning of the end for circuit-switched netwo

[evilviper]

I can't believe /.ers are amazed and puzzled by such simple things as a forty year old idea being used by a carrier. I guess /. isn't what it used to be.

Amazed and puzzled? I think not.

All I'm trying to point out is that they will no doubt discover some purpose in the future where switched-circuit systems are better suited than packet switching.

It's quite possible the more important systems will go back to being circuit switched because of the better reliablity... With no other devices sharing the line, there's no chance of resource exhaustion. Just ask Bank of America how well their VPN worked when SQL Slammer hit. I'm not sure that's the main advantage of circut switching, but I have no doubt technology will turn around, and make circuit switching in-vogue again.

I have lots of 99.999% uptime systems

Thanks, but I'll stick with my 100% uptime phone system for now. Not that VoIP is terribly useful for residential situations right now.

Re:beginning of the end for circuit-switched netwo

[grumling]

Thanks, but I'll stick with my 100% uptime phone system for now. Not that VoIP is terribly useful for residential situations right now.

Your phone system is only 99.999% up. This equates to about 10 minutes unscheduled downtime per year, per line. Of course this is an agragate number for all phones on the network.

Wrong!

[Anonymous Coward]

VOIP does not have require the same bandwidth for identical quality. The above statement is FLAT OUT WRONG. VAD (Voice Activity Detection) alone can dramatically drop bandwidth requirements for even a G.711 (no compression) codec. (Most of a phone call is silence, as speakers pause between words and phrases. If no data is transmitted after a tiny "silence start" packet, then bandwidth is immediatly saved).

Acceptable PSQM (PSQM, perceived voice quality, ITU standard testing) scores are attainable using 30 millisecond codecs, and even multiple voice frames per IP/UDP/RTP (RTP = Real Time Protocol).

So when you take say, 30 milliseconds of G.711 (uncompressed) data, tack 42 bytes of Ethernet/IP/UDP/RTP header, the header starts to diminish when compared to 40 bytes of uncompressed voice data per 10 millisecond period (120 bytes per 30 ms packet).

Still not terribly efficient, but PSQM scores with these codec configurations have IDENTICAL scores in well designed systems/networks to conventional PSTN telephony. Throw in even the simple VAD/DTX and bandwith is saved with a zero-compression codec. Add in some of the common "real" codecs that do compression, (G.729, G.726, G.723) and significant bandwidth is saved.

There are single chips that will do the conversion from traditional telephone networks to packet networks (either ATM or VOIP) and do hundreds of calls per chip. These devices are being used to build the gateways being used by the telco's in every majior country to build the next networks. In countries like China, they are building the infastructure from the start using VOIP networks.

Slashdot is not what it used to be...

Re:beginning of the end for circuit-switched netwo

[Ex-MislTech]

One point to make here .

As someone that worked in the Telco industry there
are -->> ** NO ** -- 100% uptime systems .

At some point, the system will error, busy out,
or a fiber seeking backhoe will feed itself .

The last major california earthquake showed them
that they needed a system with dynamic routing
capability on the fly .

When the World Trade Center fell phone service was
in big trouble in that area for some time .

This has been played out time and time again ,
not only nationwide, but worldwide .

VoIp can change some of it, but not all of it .

Nothing man can make is 100% .

The US phone system is geared around its heaviest
call voume day, mother's day, That is their loose
benchmark .

The reason he quotes 99.999% is that is the five 9's
reliability quotient set by the FCC .

VoIp Dial offload is providing 5 cents a minute any
time night or day , 365 days a year , and you do not
even know you are using it most of the time .

It is transparent to the end user if done correctly .

Companies like Sonus and Genuity made millions off
their VoIp solutions, and so did many others .

I worked at a now defunct Cisco Systems VoIp lab in
Herndon Virginia that used MGC , and it was steamrolled
by SIP internally , and other solutions externally .

When the other solutions beat us out, Cisco nearly
closed the facility they laid off so many people .

Over 50% of the office I worked at was let go, and a virginia law
came into effect for letting so large a number go .

I learned alot there, and one thing for sure is that
there are no 100% uptimes .

Only multiple redundancy can give you 100% uptime, and then
you are just building a 2nd parallel network .

Peace,
Ex-MislTech