AT&T Migrating Phone Network to IP

prostoalex writes "Following the lead of Sprint and Telus, who are moving their telephone networks to IP, AT&T will spend $3 billion to migrate to an IP-based network. By the end of 2005 about 270 legacy systems will be retired." The article also notes how the current ratio of packet traffic to voice is already 8:1.

Comments

Better be IPv6

[Thinkit3]

Well I suppose they're smart enough to go to something much more expandable. Just wonder how much legacy (ick) will still be stuck there.

What about VOIP

[rock_climbing_guy]

Is this any indication that with the proliferation of IP technology, even the phone companies will eventually start working with VOIP instead of trying so hard to kill it?

If so, maybe they should spread the good word to our frinds at the RIAA. ::/me wakes up::

Re:What about VOIP

[cmowire]

Not a chance.

Phone companies want to itemize and per-minute and allocation and whatnot anything to death. They will stop working with VOIP when they are forced to, not a second before.

Re:What about VOIP

[silentbozo]

Depends on which phone companies you're talking about. The incumbent local exchange carriers (ILECs) make their money by charging customers outrageous monthly fees, and by charging long distance companies for terminating their calls at the local subscriber's telephone. They hate VOIP.

Carriers like AT&T, which sell primarily long distance, like VOIP since it saves them money, and could eventually allow them to bypass the ILECs entirely, since it turns voice calls into another internet data stream. They like VOIP.

Re:What about VOIP

[Tmack]

And how does VOIP change this??

VOIP is just another protocol. Most people seem to not realize that by the time their phoneline reaches the edge of their neighborhood, it has become a digital signal. The transition to VOIP is just natural progression. It allows more flexability, but will still require routers and switchtes to operate. Through these switches and routers is how the phone companies will keep track of calls. VOIP does NOT mean an end to phone numbers, providers, etc... Remember that most of the internet is carried by the ILEC networks on the same loops used to carry voice, just reonfigured slightly to allow pure data traffic. VOIP providers merely use these loops in the data configuration with routers that convert the analog voice signals to packets closer to the customer end than normal voice lines. VOIP merely abstracts the traffic type from the physical layer more than current SS7 and other protocols. VOIP is not simply PC-PC calls placed by IP address. VOIP is only a different protocol, central switches are still used to route calls and keep track of things, they just run more efficently (ie: 1 VOIP switch about the size of a 10k cisco can handle the entire call volume for a decent sized city (or 2) where currently several switches are required by the ILECs). Per-minute rates and such will still be acounted for. Phone providers will switch to VOIP mainly due to the relative simplicity and flexability of its stucture. VOIP is NOT what alot of people percieve, it is simply a new method of routing voice traffic that does not eliminate the need for routers/switches/etc...

TM

Re:What about VOIP

[convolvatron]

you didn't live through networking 'revolutions' like ATM. at that time we were told repeatedly that IP may have been a nice playground, but the telcos, who really know how to run a service, were going to take over now. IP and TCP were never going to be suitable for large scale business deployment.

'at&t moves voice service to ip' would have been a hilarious gag article 10 years ago, now no one even blinks.

Bandwidth?

[lord_paladine]

As this article states, the bandwidth required for VoIP can be huge. I would seriously hope to see some more advanced algorithms or better yet, more bandwidth installed, before these systems become more heavily adopted.

Re:Bandwidth?

[Abcd1234]

Too bad your link refers to corporations using VoIP on their LAN/WAN as an alternative to traditional telephony. What we're talking about here is telephone operators using IP as a backbone transport (as opposed to voice over ATM VCs, etc). For telephone providers, VoIP has some excellent advantages, the most notable being consolidation of existing infrastructure (ie, being able to use the same lines for both voice and data).

Re:Bandwidth?

[phliar]

I used to work at a VoIP added-services provider. Received wisdom was that packet telephony was definitely the future, but 50-50 on whether or not IP was the right protocol for packet voice networks. One thing for sure is, in H.323 there's no point in using G.711 for voice -- a decent 10-12kbps codec will sound fine compared to the 64kbps that G.711 uses. I think that H.323 is sensitive to all kinds of parameters like comfort noise and silence suppression, you need to tune it to your network. In practice, it looks like well-tuned VoIP does take more bandwidth that good ol' PSTN, but the difference is not significant enough to justify running two kinds of networks.

However I worked at the software end, not VoIP network operations -- what do I know?

Re:Bandwidth?

[doogles]

As this article states, the bandwidth required for VoIP can be huge. I would seriously hope to see some more advanced algorithms or better yet, more bandwidth installed, before these systems become more heavily adopted.

Ah, but with packet telephony, we are only "burning up" bandwidth for active calls:

Take a traditional circuit-switch T1 carrying 24 DS0, sitting idle making no calls, and you still have a T1 that can be used for nothing else.

Take the same scenario in a packet-switched world, and you have a T1 100% usable for other data until such time as the circuit is needed. QoS (LLQ, or PQ/CBWFQ in Cisco-speak) ensures that when there IS a voice call it gets priority treatment.

Last note, on IP overhead: Enterprises with smaller links can leverage compressed RTP headers (cRTP) to reduce the 40 byte IP/UDP/RTP penalty down to 2 bytes across point-to-point links (Frame Relay PVC, leased lines, BRIs, etc). This concept doesn't really apply to a carrier because of the CPU impact header compression costs, but considering all carrier networks are currently severly underutilized I do not think this should be a reason to shy away from packet telephony.

You know

[loraksus]

I . . . loveusin. . . gIPtelephony. . . . Lagis . . . notreallyanissue.

Things can evolve

[Rosco P. Coltrane]

The article also notes how the current ratio of packet traffic to voice is already 8:1

The day the RIAA will be such a threat to peaceful P2P that people will start reading aloud file hex printouts byte by byte over the phone to share data, I think the trend will reverse. But I might be wrong ...

an easier way

[Anonymous Coward]

i've found a much easier way to do this. just set up the text-to-speech component of your favorite IM program. now all my friends sound like stephen hawking!!! : p

Phreaking

[Pro_Piracy_Guy]

Does this mean I am going to have to adjust the frequency of my Kaptain Krunch whisle when I use the pay phone?

This brings back memories

[Sevn]

A few years ago I was involved in a startup company. We had talked to qwest about buying a bunch of dark fiber. We had secured an insane amount of capital from Phoenix, E-Street, and MSDW surprisingly as they normally wouldn't be interested in a startup as they don't normally cater to incubator or angel type projects. So what were we selling? We had sat down and figured out that with some very expensive sycamore or juniper routers and DWDM and a bunch of dark fiber, we could roll out a nationwide flat rate VoIP long distance service for about 250 million dollars. We had an awesome business plan. A solid year of work. All the right buzzwords and an executive summary that would make the most hardened VC blush. We were a few months from starting. Qwest was excited. Everyone was excited. Then *poof*. All gone in an instant. It seems that AT&T had issued a statement to their stock holders that they would not be paying out dividends that year to anyone because they wanted to warchest that money in case someone like us came along. So the business plan was instantly invalidated. If things had gone the other way, I'd more than likely be selling a lot of you unlimited long distance service for 30 bucks a month, and expanding worldwide.

Re:This brings back memories

[Sevn]

Almost. We were going to lease it from qwest. They had already done the work. I think the way things ended up, we had a 20 year lease agreement that kinda would have screwed us down the road. But we would have had a decade or so to figure something out. Qwest was trying very hard to sell us a managed solution. They initially didn't like the idea of parting with any unlit fiber. You should see the difference in prices between a managed fiber solution and leasing dark fiber.

Re:This brings back memories

[Sevn]

You are overlooking the benefits of being first to market with something. That and the phone in the home isn't going anywhere anytime soon. Who is to say that we wouldn't have eventually been able to offer transport on our network to a t-mobile or similar company? That and advances in the past 3-4 years with fiber technology would have meant we could have easily doubled the traffic we would have been able to push across our fiber. It wouldn't have been hard at all to stay ahead. Granted, a great deal of our profit would have ended up going towards FCC lobby and other legal expenses because we knew the telcos would not take something like this sitting down. That's why we had provisions for such things in our business plan.

IP telephone service has come a long way

[joel8x]

A few years ago I was part of a deployment of an all IP Siemens phone system at this place I used to work at. The voicemeail was actually part of an MS exchange server, and you recieved all of your voicemail in you inbox in Outlook. I loved it because I could set up a PST and easily archive phone messages on my hard drive. Unfortunately the system suffered from horible sound quality (there was a lag when you talked to people and it echoed like crazy) and was just not ready for prime time. I got a great taste of the future of business IP phone systems, though.

Outdated infrastructure?

[RobertB-DC]

From the article:
AT&T plans to retire 270 legacy systems across the world by the end of 2005. Approximately 130 legacy systems were retired over the past 18 months, with another 140 systems slated for phase out over the next two years.

The article didn't define exactly what "legacy systems" were (switches? entire local networks?), but that sure sounds like a lot of high technology that's heading for the dustbin. We're talking technology that's currently in use creating a mobile communications system that would have been unimaginable just 15 years ago.

Will it all be scrapped out? Will barges full of misc parts be shipped to third-world scavenging companies to recover the precious metals? Or is there some way to move the equipment to areas that need it -- Afghanistan and Iraq come to mind right away, but I'd think that under-served (and under-reported) countries like Somalia and the rest of Africa could make use of this supposedly outdated hardware.

Of course, we're back to the same old question -- when it costs more to recycle than to dump, how do you justify doing the Right Thing to shareholders whose only interest is in doing the Profitable Thing?

Re:Outdated infrastructure?

[Kallahar]

The problem with POTS systems is that you have to have the entire country wired with copper. A place that doesn't have the same huge investment in infrastructure that the US has is probably better off buying a used cell phone system and just running fiber between the towers. We may think of cell phones as being a luxury but that's only because we have 70 years of investment into copper to every home.

I hope this doesn't mean what I think it will

[donutello]

I make a lot of international calls and I've experimented with Sprint, MCI, AT&T and a whole bunch of phone card providers.

Without question, AT&T has been miles better than the rest. The other providers obviously use packet switching as evidenced by the intermittent delays as much as a couple of seconds. Sometimes you can get half-way through a sentence when you hear the other guy starting a sentence that he did when there was silence - it gets very annoying because both of you have to practice random backoff which can either result in empty silence or both of you speaking over each other.

I hope AT&Ts service doesn't go that way.

Dialup Users

[jasontwarnock]

Hmm, I wonder what this will do to dialup users. TCP/IP over PPP over VOIP.

Oh, great.

[AJWM]

Now when the next sobig.f or whatever hits, we'll lose the phone service as well as the electicity.

Good!!

[moehoward]

I, for one, welcome our new packet-based overlords.

Now. About this encryption thing...

Does AT&T get to avoid regulations now?

[geekee]

According to this slashdot post VoIP should remain unregulated. Now that AT&T is using VoIP, do they get the same treatment?

Skype

[westlake]

CNET has posted a Sept 11 newsmakers interview with Kazaa's Janus Friis promoting his P2P telephony app, Skype: Why VoIP is music to Kazaa's ear. The download (for Windows 2000 and XP only) can be found here: Skype beta.

Re:Dumb Dialup Question

[cfulmer]

This is actually one of the funny problems with VoIP. If you just take a standard phone call, you can packetize it and if you haven't tried to gather too much audio into one packet, the additional latency won't even be noticable on the other end. Heck, you can even drop an occasional packet and chances are nobody will notice.

Modems, however, do not handle either latency or packet loss well -- part of the initial V.(90??) standards take a latency measurement at the beginning, expecting it to be some small number that doesn't change. In VoIP, not only isn't the number small (closer to 100ms than 20ms), but it can vary over the life of the call.

So, what ends up happening is that your local gateway (the thing that converts between traditional phone and packet communications) listens for your modem tones and kicks in a V./G.whatever codec to convert it into packet. Then, at the far end, the same thing happens.