x
Cable/Video

Cable Gets a Vault from VOIP

This week there's more good and bad news for pure-play VOIP providers and CLECs. The good news is that residential VOIP services are still seen as a booming business. One industry forecaster -- JupiterResearch -- says as much as 10 percent of the U.S. population could be using VOIP phone services by the year 2009. The bad news? The cable companies look to be the ones positioned to benefit most of all, for all the usual reasons that big companies win in the communications world: the ability to bundle and bill for multiple services and the big budgets to afford aggressive marketing (see Report: VOIP Growth Won't Benefit All and Report: MSOs Scoring With VOIP). The JupiterResearch report, “Broadband Telephony: Leveraging Voice Over IP to Facilitate Competitive Voice Services,” states that about 1 million U.S. households will be using VOIP services by the end of next year, more than double the current total. That group will swell to 12.1 million by 2009.

”Next year will be important for continuing VOIP momentum as far as building customer awareness,” says Joe Laszlo, senior analyst at JupiterResearch. And he says cable companies are in the driver’s seat for dominating VOIP services:

“It’s the cable companies’ game for the next few years. That’s not to say that specialized VOIP companies like Vonage Holdings Corp. and Packet8 won’t compete, just that they don’t have the marketing muscle the cable companies do and must use a more focused approach. The cable companies are well-positioned for offering VOIP as a bundle with their existing video services.”

For wireline carriers, VOIP is a lower revenue service than a regular landline phone connection, but the services will be helpful when selling service bundles that include high-speed data connections. “Their pressing issue is not to have VOIP, but to add video services," Laszlo says. "They look at it that the customer doesn’t care if telephony services are IP-based or not but that they must have a bundle that competes with the cable guys.”

As noted at last week's UBS Investment Research media conference, VOIP services are gaining a fast following and often this comes at the expense of incumbent phone company access lines:
  • Cablevision Systems Corp. (NYSE: CVC) announced it had surpassed 250,000 VOIP customers after reporting just under 200,000 at the end of the third quarter. It is currently adding 1,000 customers a day, despite not offering local number portability. The company expects to add number portability next year, which should cause an increase in customers.
  • Vonage stated that it had 276,000 customers at the end of the third quarter, having added customers at a rate of about 27,000 per month.
  • Time Warner Cable estimates it will have more than 200,000 telephony customers by year-end and is adding roughly 10,000 customers per week. The UBS report notes that Time Warner Cable is now fully deployed in all its territories and has said that 75 percent of its new customers were porting numbers, indicating a switch of primary lines. UBS predicts that Time Warner Cable, which has 10.9 million basic cable subscribers, and a roughly 72 percent overlap with the Bells, will get more aggressive with marketing its VOIP bundle next year.


Of course, many of the wireline incumbent carriers are hedging the revenue they're losing to VOIP players with gains from increasing numbers of wireless subscribers. The JupiterResearch report found that younger consumers were least likely to turn to traditional carriers for service and the least likely to use landline telephony at all. Twelve percent of 18- to 24-year-olds say their wireless phone is their only phone. "For attracting young adults, VOIP’s biggest competitor may prove to be mobile operators, not the Bell companies," Laszlo says.

— Chris Somerville, Senior Editor, Next-Generation Services

Page 1 / 2   >   >>
OldPOTS 12/5/2012 | 12:59:11 AM
re: Cable Gets a Vault from VOIP RE: AlchemyGs comments:
Modern compression codecs are much better than ADPCM. G.729e at 11.8 k/sec does a very good job of reproducing voice and is OK at reproducing music. GǪandGǪ.
Eventually, the MSOs will have few media gateways since they'll be handing off the bulk of their off-net traffic using IP rather than TDM.

This concerns me. We tried some of the precursor compression techniques, like those mentioned, on Stratacom (Pre ATM) and TDM muxes to save significant money. But the company (Fortune 10) user reaction was almost a lynching. Then even when we went back to ADPCM, the financially conservative (tight) CEO could distinguish between PCM voice. So he got his own PCM lines, while we saved money on the workers with ADPCM. With my experience, I also admit to being a fan of the quality approach. But there are those that live on cell phones all daylong, with the not so understandable quality.
Idea: One might offer both with a price difference!? But my scares tell me the cheap guys are the squeaky wheels and end up costing the same (OPEX).

Remember Fgoldstein;
The nethead, GǪ will try to interoperate with other VoIP, handing off as much traffic as possible over IP, etc. The CLEC approach, in contrast, will try to provide the best possible dial tone experience, going straight for the ILEC jugular with top-quality phone service. GǪAndGǪ.
If, opportunistically, the cable guys do find a way to hand off their VoIP in IP format, directly to another carrier, that's fine too. But geeks squawking to geeks over cell phone-quality connections (Vonage on a good day) should not be the goal.

During my experience with designing core network long haul cross-country fiber optics networks (using Erlangs and CCSs), the price for those circuits was very expensive. The costs of those new inexpensive IP interfaces, OC48/GiGE, arenGt even enough to pay for the monthly or weekly cost of one circuit of the same size. For the "Class of '99 redux" nethead VoIP specialists, those long haul circuits are made up of more than one in-expensive SONET/DWDM fiber, repeater and equipment (for n>1, n x $ = $$$$). And then ask the land manager what Gright of wayG costs are. Access circuits are relatively cheap.

Once I had a super computer guy looking to connect to another one across country. So I quoted him a highly discounted monthly price for a circuit and he assumed that was good for all three years of his project. He just knew he could get one cheaper, just like IP marketers had told him. So he looked. Well he ended up with sharing the circuit and being subsidized by the CEO, the only person in the company that could authorize that large amount of money.

Since those long hauls are so expensive, you can bet the core networks eventually will contain eye pee with the header religion suppressed, with only good new compression techniques and rich separation/QoS. Especially when the big bandwidth consumer GvoiceG is added. (IGm strongly for IP/Ether at the edge aggregation point for flexibility/service) IGd bet on ATM long haul for the bottom line/operating profit. But POS with limited IP and no Ethernet headers is as efficient (All measured G no theory). Both designed for the task. GigE ainGt. It will happen because these efficiencies can save up to 20-30% on really big $$$ of core network without that IP & Ethernet header tax. But I digressed from edge/access discussions to long haul.

Later it will be discovered that one needs to add money to the bottom line as operating profit not just stock sales, something that the "Class of '99 redux" nethead VoIP specialists have not shown they can do.
I also agree with the comparison to Justice Stewart's definition of pornography statement. Netheads will eventually know it when they see it, after saving interface costs for a while until they discover someone else has more competitive operating profits and lower prices.

If the money is right for a larger Caribbean sailboat, I might consider coming out of retirement then. But IGll sail (bob along) for now.

OldPOTS
fgoldstein 12/5/2012 | 12:59:15 AM
re: Cable Gets a Vault from VOIP I didnGt realize that Cable VoIP was not the same as DSL VoIP

And my point is that there are nearly infinite variations on what can be calld "VoIP", if that's what the stockmongers want to hear. My FCC Comment in the VoIP docket spends about half of its time describing why they can't simply set up a rule about "VoIP" without understanding what is and what isn't. Justice Stewart's definition of pornography (roughly, "I can't tell you what it is but I know it when I see it") is apropos. VoIP ranges from NetMeeting to Skype to Vonage to what I call "dry Martini" VoIP, where you take a regular phone call and whisper "VoIP" over it, the way AT&T tried.

So you basic PacketCable rollout is indeed what alchemy describes, a CLEC with its own Class 5 switch and trunks to the ILEC and IXCs. Now you can take that and go two ways. The nethead, who worships eye pee and thinks the PSTN is a bunch of geezers selling buggywhips, will try to interoperate with other VoIP, handing off as much traffic as possible over IP, etc. The CLEC approach, in contrast, will try to provide the best possible dialtone experience, going straight for the ILEC jugular with top-quality phone service.

Now I admit to being a fan of the latter approach. Therefore I dismiss all compression, VAD, comfort noise, etc., as total crap that has absolutely no business in a "local" service. And puh-LEEZ don't tell me about the spiffy MOS scores that your favorite vendor's product got. It's useful for calling Botswana, maybe, since that's costly bandwidth. But a regular Bell phone doesn't do that crap to voice, and passes modems and fax just fine, so why should a competitor offer a lower quality service, just because it can? Yes, VoIP is often a "race to the bottom", but PacketCable doesn't have to play that way. The bandwidth is there to do it right, and do it well. That means uncompressed PCM with lossless, jitterless QoS.

But hey, even though they can compress the eye pee headers out of the packet, it's still possible to tell The Street that it's VoIP, so the MSOs can keep their stock multiples up. And make money at the same time, operating profit not just stock sales, something that the "Class of '99 redux" nethead VoIP specialists have not shown they can do.

And if, opportunistically, the cable guys do find a way to hand off their VoIP in IP format, directly to another carrier, that's fine too. But geeks squawking to geeks over cellphone-quality connections (Vonage on a good day) should not be the goal.

Insofar as replacing the CMTSs is concerned, yes, that's an issue. My Tellabs 2300-based phone service is a lot more reliable than my cable modem. Cisco just doesn't get it. The next generation of CMTS will need to be real carrier grade stuff. It's available.
alchemy 12/5/2012 | 12:59:15 AM
re: Cable Gets a Vault from VOIP OldPOTS writes:
I didnGt realize that Cable VoIP was not the same as DSL VoIP. IGve compressed voice > 3:1 on Stratacom fast packet muxes (ADPCM & VAD) many years ago (PreATM), so good 4:1 compression should be obtainable by now.

Cable VoIP can interoperate with other VoIP. The lowest common denomonator is uncompressed G.711 uLaw and aLaw in 20 mSec frames over IP/UDP/RTP. Cable VoIP layers a nonstandard security model on top of this that won't interoperate but you can disable this and encrypt within layer 2 DOCSIS using BPI+. Eventually, the MSOs will have few media gateways since they'll be handing off the bulk of their off-net traffic using IP rather than TDM. Today, the model looks just like a CLEC operating their own class 5 switch where they have trunks to the ILEC and trunks to one or more IXCs.

Modern compression codecs are much better than ADPCM. G.729e at 11.8 k/sec does a very good job of reproducing voice and is OK at reproducing music. If you layer voice activity detection and comfort noise generation on top, it's quite network-efficient. 3G wireless has codecs that compress even more with similar audio quality that use the Qualcomm variable bit rate scheme.

Sadly, the MSOs latched onto a (claimed) royalty-free codec called iLBC from a Swedish shop called Global IP Sound. They claim that since they've made the algorithm public domain, it's royalty-free. It's likely that it steps on quite a few patents if anyone cares to go after them once the codec gets widely deployed. Unlike Linux, there are lots of companies to sue here since the MTA and media gateway companies are all incorporating it and the MSOs are buying product that uses it. The pathetic thing is that iLBC is a lousy codec when you bake it off against its competition. It sounds very tinny and it's poor at reproducing women.
OldPOTS 12/5/2012 | 12:59:17 AM
re: Cable Gets a Vault from VOIP I didnGt realize that Cable VoIP was not the same as DSL VoIP. IGve compressed voice > 3:1 on Stratacom fast packet muxes (ADPCM & VAD) many years ago (PreATM), so good 4:1 compression should be obtainable by now.

So letGs see, itGs Cable DOCSIS and itGs upgrades that contain eye pee with the header religion suppressed, with good new compression techniques and rich separation/QoS. MSOs need to forklift upgrade their legacy CMTSs and replace them with carrier-class CMTSs to get these features and achieve telco-grade availability @ ~$200/customer. But will it still handle my old POTS crank phone?

Sounds like the Cable operators are one technology step ahead of the RBOCs with ethernet DSL and FTTH @ ~$300-800/customer.

And I thought I was the only one left that knew how to calculate capacity with CCS and Erlangs. I thought busy hours were when you went to the bars to avoid the traffic congestion..... IGd bet it is not only Vonage and my spell checker that canGt spell those terms either, much less understand and use them. How many marketers know the VoIP buzz words, but not Erlangs (Just old switch marketers?) But like my cell phone company, theyGd all give it their best effort to distort them.

Great responses!
OldPOTS
fgoldstein 12/5/2012 | 12:59:22 AM
re: Cable Gets a Vault from VOIP Again, the pundits, banAnalysts, and imbeciles who (mis)manage large sums of money are allowing a buzzword (VoIP) to confuse the issue. Here's a klew we should bash them with: Not all "VoIP" is the same thing.

Cable companies don't need VoIP to do triple play. The old Arris and Tellabs stuff works very well, probably better than any VoIP ever can. But the price is higher. It's higher because there's no standard. You buy a Tellabs 2300 system, you buy Tellabs 2300 terminals. You buy Arris Cornerstone at the CO, you buy Arris Cornerstone at the house. Forever. Vendor lock in protects them in the short term. But buyers never like that. They like commodities. That's exactly the same reason why Bellcore began the SONET push in 1985, so the RBOCs weren't stuck with Fujitsu, Nortel, AT&T, or any other vendor's proprietary fiber networks.

So CableLabs did a standard for voice over cable, PacketCable. It is 100% buzzword compliant. It uses eye pee headers, so they can tell the idiots on Wall Street that it's VoIP. That opens up what Lucent once described as the main benefits of VoIP, "Access to Capital" and "Wall Street Image".

But PacketCable isn't Vonage, not by a long shot. The bandwidth needed for a call is reserved, and the voice traffic is segregated from data at the CMTS. The IP headers are largely gratuitous, though they could be used to aggregate multiple CMTSs to one switch/media gateway. (Of course ATM could do the same thing with less overhead.) So you end up with Real Telephony, or something quite close. And because it's a standard, it's dirt cheap -- the price of an "EMTA" (your choice of vendor) is a lot less than that of a Cornerstone subscriber box (your choice of Arris).

Bottom line is that an MSO can roll out good-quality telephony for an incremental capex <$200/line, plus switching (which is pretty cheap too, a couple hundred dollars per active call of capacity).

Is the bandwidth there? On some of the crappy old ex-TCI or Adelphia systems, if they still have 1000 homes/node, no. But on any decent system, sure. DOCSIS 2.0 triples upstream bandwidth, after all, so long as everyone on a given channel uses it. Figure a voice call takes about 100 kbps in each direction (no compression, no VAD; this is real local telephony). Figure average traffic of 5CCS. Figure DOCSIS running with 10 up, 30 down, and 200 homes/node with a 50% (optimistic!) telephony take rate. 100 PacketCable homes * 5CCS = 500 CCS or 14 Erlangs, which needs 23 "trunks" or 2.3 Mbps peak and 1.4 Mbps busy hour average. No problem. (And most CMTSs have multiple upstream channels per downstream anyway.) Business lines are a different story; the MSO probably needs to drop off fiber in order to have capacity to serve multiline business. But they probably have it nearby if they decide to serve the business area.

You don't see Vonage making those calculations. That works when it works. Not quite the same thing.
alchemy 12/5/2012 | 12:59:23 AM
re: Cable Gets a Vault from VOIP OldPOTS writes:
Also I wonder if cable has the capacity for VoIP. Please note that VoIP carries a header of 20-30 bytes for each ~15 bytes of voice. My Comcast IP link already has troubles with the heavy users/gamers in my projected neighborhood. Comcast left me with a capacity tester and capacity really varies. At times it becomes less than 40%. Not good for 911 calls.

Today, VoIP over the MSO network tends to use G.711 (uLaw) at a 20 mSec frame size. That's 160 bytes of payload. The header overhead for RTP/UDP/IP?MAC is roughly 40 bytes. As the take rate increases, the MSOs have two things in their bag of tricks: compression codecs and packet header supression. You can get 4:1 or better improvement by using a modern compression codec. DOCSIS supports packet header supression so much of that 40-byte header can vanish. Part of the activity for DOCSIS 3.0 is to make this header supression work better and ensure that cable modems and CMTSs interwork in an optimal way. With vanilla G.711 and no packet header supression, MSO IP network capacity is certainly a concern as the MSOs scale their product. When all this stuff is turned on, VoIP is quite efficient on the MSO IP network and the impact on the network is minimized. The main cost impact is that MSOs really need to forklift upgrade their legacy Cisco CMTSs and replace them with carrier-class CMTSs that typically come from Arris or Motorola. You can't achieve telco-grade availability numbers when your CMTS isn't redundant and has to be taken down to do software or hardware upgrades. Cisco, of course, is in denial about all of this and has an army of suits with decks of PowerPoint slides trying to refute the need to dump their CMTSs.

DOCSIS has a rich QoS mechanism that ensures VoIP packets get delivered with no packet loss and minimal jitter. Some amount of bandwidth is reserved for '911' by requesting "priority voice" flow specs.

The CMTS vendors are all working on much higher capacity line interface cards so the price of IP will continue to drop. They're also re-partitioning the architecture somewhat in an attempt to drive down costs. As the MSO network migrates to all-digital, a huge amount of spectrum will be reclaimed that can be used for yet more IP-based applications.
spelurker 12/5/2012 | 12:59:24 AM
re: Cable Gets a Vault from VOIP > Also I wonder if cable has the capacity for VoIP.
[...]
> My Comcast IP link already has troubles with the
> heavy users/gamers in my projected neighborhood.
> Comcast left me with a capacity tester and
> capacity really varies. At times it becomes less
> than 40%. Not good for 911 calls.

Cable does have the capacity. A cable plant with operator-provided Voip (e.g. Comcast, NOT Vonage) will always have extra capacity available for 911 -- similar to how a full POTS switch will always allow a 911 call through.

The problem has a lot more to do with the capacity planning of a particular network. You'll have that problem with any provider (what exactly is behind the DSLAM/CMTS & how many users are on it, etc.)
OldPOTS 12/5/2012 | 12:59:25 AM
re: Cable Gets a Vault from VOIP One of the RBOCs, SBC, is going to counter with DSL, at least for existing customers. This may be at a gradual rate (evolution) as they trial in 16/28MB DSL. This action is to conserve spending up front money and realizing that not everyone is going to take all the services immediately. They can put different DSL types in the same DSLAM and meet each customers need. Even in green field FTTH installations the take rate for all triple play services is lower than expected. And these green fields are usually where the projected customers live

Also I wonder if cable has the capacity for VoIP. Please note that VoIP carries a header of 20-30 bytes for each ~15 bytes of voice. My Comcast IP link already has troubles with the heavy users/gamers in my projected neighborhood. Comcast left me with a capacity tester and capacity really varies. At times it becomes less than 40%. Not good for 911 calls.

I currently have extended Analog Cable vs. the Digital cable that they have tried to GslamG me to twice. They are slowly removing channels off of Analog (where most of their subscribers are) to get subscribers to move to the more expensive Digital, even with the steep increases (~30%) in Analog service. Digital cable pricing structure is much like satellite. Digital also gets them more bandwidth. I have been advised that because of a 1996 federal law, the city (franchise) will have no control over cable. This is because there is a greater than 15% take rate of satellite (DirecTV and Dish).

So I expect to be forced to satellite, before SBC or VZ gets to my house. Hopefully providing an RBOC choice will keep the cable providers honest. But the really big issue is where they get, or how much the RBOCs pay for content!!!

OldPOTS

keelhaul42 12/5/2012 | 12:59:28 AM
re: Cable Gets a Vault from VOIP I remember reading about that not so long ago (FTTP). I also remember thinking (sic) "talk is cheap".
I wasn't sure that the RBOC's were going to counter with anything given the mindset they've had.
Ok, I'll lighten up. If they're really going to do it, well, more power to them.
Hopefully providing a choice will keep the cable providers honest.



-kh
bored_lurker 12/5/2012 | 12:59:29 AM
re: Cable Gets a Vault from VOIP I'm not sure why you all think that the RBOCs are going to counter with DSL. All the way up to 8MB - wow! Not. The joint FTTP proposal meeting happened less than a year ago and I expect to have fiber up and running on my house in the next few weeks from VZ.

That pipe gives me 15MB down and 2MB up plus up to 4 POTS lines today. As I hear it VZ will offer me TV service next year. And believe me, VOIP will run much better over VZs 15 and 2 service than it will over my current 3MB and 256K cable service.

Yes, if the RBOCs don't change to something like this they are done for - but I think they get that. Fiber brings the triple play well within their reach.
Page 1 / 2   >   >>
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE