How ISPs May Quietly Kill VoIP

ravenII writes "PBS's i'Cringley's informative piece gives an eye-opening look at the anticompetitive behavior of some ISPs who are showing up late to the VoIP game. This is not something that could be easily mandated, and the beauty of this approach is that they're not explicitly doing anything to the 3rd party service applications. They're just identifying and tagging their own services, which is within their rights."

Comments

Not fair

[turtled]

That's not fair. A new innovation comes and is sucessful, and people have to squash it wrather than create compition, which would in turn create better products and lower prices for consumers, yet possible revenue for the best player. I have vonage. I love it. $25 a month, it kills the same bill from SBC ($73/month, everything the same) and Verizon($93/month, everything the same)

Re:Verizon

[Paska]

Read the damn article, this is different to the case you mention. They are not blocking access to other VoIP providers. However they are tagging (or will?) be tagging their own VoIP traffic which will force all other VoIP provider's traffic to run as non-guaranteed traffic and thus, could lead to dropouts or all round crap service.

It's going to be bad, in theory

[Dancin_Santa]

So the main point seems to be that there will be a preferential class of packets that will be guaranteed to have some level of service such that the packets arrive quickly and in order. The bad part is that all other traffic will remain at the same old unguaranteed service level.

Well, that's what we have now.

Face it, the reason people use VoIP is because it is cheap/free, not because it has superior QoS than POTS. Throw in compression and encryption and you're talking about some pretty serious degradation of service.

So, in summary, nothing to see here.

Re:It's going to be bad, in theory

[m0rningstar]

Well... VoIP technology is inherently extremely sensitive to both latency and jitter; this is why Cisco is trying to work with ISPs (their 'V3PN program', which always sounds like a Star Wars driod every time I talk about it) to get them to listen to QoS/DSCP values as set by the customer in their network. (Or to route DSCP tagged traffic into the appropriate MPLS TE 'VPN', or whatever you choose as a methodology)

This, of course, raises huge issues for the general consumer, since those willing to pay what's probably a premium to NOT have their DSCP values stripped off at the edge of the network get further stomped, even without any form of 'anti-competitive' prioritisation -- the end users get squished first as they don't have a 'business class' service and the only real way for a backbone provider to make money is to over-subscribe their backbone and rely on the bursty nature of IP traffic to handle it. (At least, that was the plan when I was working with VERIO engineering a few years back; now I'm just a conslutant on the Cisco side... )

Re:It's going to be bad, in theory

[Mammothrept]

Call bullshit.

There is something to see here and you are averting your eyes. The throttling scam works like this:

Assume the total amount of VOIP traffic that wants to move across a telco's network is some number. Let's call that number 11 (think Spinal Tap). Now, of that 11, 3 is VOIP traffic from the telco's own service. The remaining 8 is Vonage, Skype and all the rest. Rather than fuck with the rest directly (illegal), the telco throttles total available VOIP bandwidth to 10 but assigns preferential QOS headers to the 3 that it profits from. Vonage and company now have to share the remaining 7 even though they need 8. Their quality suffers and they shed customers to the telco's VOIP service. As long as the telco tweaks the throttle correctly, they can bleed Vonage without breaking the law as currently written.

Re:It's going to be bad, in theory

[Barnoid]

now I'm just a conslutant on the Cisco side...

is that the female form of consultant? ;)

Re:It's going to be bad, in theory

[Anonymous Coward]

No, slut applies to both genders of consultant.

Re:It's going to be bad, in theory

[dgatwood]

Latency? A typical cell phone call can have more than a half second round trip. Try it some time. Have the person on the other end listen and start counting along with you. You're not going to even -approach- 250ms latency on the public internet unless you're doing transcontinental satellite hops.

As for packet loss, for telephone conversations, most of the time, people will barely even notice a single packet being lost if you're doing things right. I mean, do you change phone companies every time your cell phone drops a packet? I didn't think so. It's par for the course, and you're used to it and probably don't even remember the last time it happened to you (which was probably some time today).

This seems like much ado about nothing. Even on hops clear across the country without any QoS, iChat AV can shove freaking video streams. Compared to that, audio is a tiny drop of bandwidth. I just don't see how we'll get anywhere close to the limits of the backbones unless they put the priority for VoIP traffic lower than standard data traffic.... The mere notion just doesn't make any sense.

QoS, like MS isn't the answer. It's the question. No is the answer.

I use VoIP for business.

[hot_Karls_bad_cavern]

When our call service can't reach me in order to help the world's largest retailer (you figure it out), then we'll see what ISP gets what heated phone call from whom. Hint: it won't be me, rather someone at bit less friendly with a much bigger bat.

Re:I use VoIP for business.

[MBCook]

Yep. A friend's boss (who controlled one of the largest cell accounts in town) had his signal get dropped on the way to work every day, which caused him problems. So he called them up and said "fix it or we change providers." They put up a cell tower, "just for him." If you control money, they'll fix it for you.

The problem is, you have to control money. They won't screw with "world's largest retailer", or if they are dumb enough to do it, they'll learn the lesson and from then on make sure their computers are nice to "world's largest retailer's" traffic. The problem is that when it's just grandma, they'll say "Hmm. That's too bad." or "We'll look into it" and nothing will ever happen.

PS: As a side note, I've heard of the new boom business for VoIP: telemarketers. No long distance to anywhere, and you could call from your call center in India to Seattle for the same price you'd pay if your call center was in Wala Wala. At least the national Do Not Call list works (for the most part).

Encryption is the simple answer...

[datastalker]

...and just as you can tunnel just about any traffic you want through port 443 assuming you know what you're doing, you can encrypt traffic between networks. Granted, that will make things more difficult at first, but it will allow people to get around things like this.

Re:Encryption is the simple answer...

[m0rningstar]

No, it won't. You just de-prioritise ANY traffic other than your VoIP traffic.

And without some form of prioritisation across a public network, VoIP becomes a flaky proposition at best. You have a 250ms round trip latency budget, and encryption adds to the serialisation delay on both ends and impacts this. Plus any out of order packet delivery or jitter will further impact voice quality, along with compression.

And people expect their phone to work. All the time. Early adopters will tolerate the impact, but the money is in the commoditisation of the service and deploying it to everyone -- and everyone will not be willing to deal with a flakey phone.

If we had control of our politicians....

[Cryofan]

They would not DARE do this.
But we do not have control of our politicians, our public servants. Why not?

Fortunatly there is a choice.

[bluGill]

I get my Internet from wifi. There is also cable and DSL at my house. The electric company is talking about the IP over powerline stuff. I can go to someone else if they mess with my connection. Even if it isn't intentional, if the service isn't up to the level I want, I will go to someone else.

Remember people, vote with your feet.

Congress won't interfere unless it means taxation.

[Alpha_Traveller]

And on taxation alone with Congress enter the fray. Basically you'll be looking at a situation where Congress will step in, if only to provide a "regulating influence to ensure competition". And to make sure they finally get their hands permanently into the net and "free enterprise".

You can bet they'll weigh in on this issue shortly, if the proceedings and back room deals haven't begun already.

Companies like Vonage will be fine, but it won't be long before things like "Federal Subscriber Line Charge" and garbage like that begin sweeping in to cut profits and make it much harder for Vonage to conduct business.

Be prepared to be taxed if the business is within the US, or is conducted in any way within US territory. It's coming regardless of your desire to see it or not. It's too big a honey pot to ignore.

Encryption for VoIP traffic

[SamMichaels]

People have mentioned encryption, tunneling, etc in the past...my question is: why wasn't this implemented from the start? Nothing to do with beating ISPs being meanie heads...but simple security for a private phone conversation?

This looks like a MAJOR oversight here...a key-based/challenge scheme on negotiation and then compress the encrypted stream. Oh wait. I just described GSM (cell phone).

Grant it, the ISP can tag packets destined for the VoIP servers...that'll take something else. Perhaps off topic, but this encryption oversight makes me wonder.

Robert X. Cringely is a sick, sick man

[pyrrhonist]

First he says:

establishes a virtual voice circuit to your girlfriend in Bulgaria so you can listen to each other's heavy breathing.

Then he says:

Your grandmother wouldn't understand. Or she might if she's Bulgarian.

Just what the hell is he suggesting here?

Re:Robert X. Cringely is a sick, sick man

[JudeanPeople'sFront]

He-he, actually "vulgar" comes from Bulgar and used to mean "common people", "citizen". The Vulgate Bible was written in the commonly spoken language, the vulgar language. Germanic languages have probably gotten "volk" and "folk" from this word, also.

There were mass settlements of Bulgars in the Apennines at various times trough the Middle Ages. There are a lot of Bulgarian toponyms in Italy, for example Monte Bulgheria. Bulgarian derived family names are common in some places - Bulgari and Borgii are famous examples. The Latin language itself was "vulgarized" due to a large number of non-native speakers (mostly Germanic people, like the Langobards).

"Vulgar" acquired a negative meaning later: plain, plebeian, unrefined, coarse and rude. It is quite a common linguistic phenomenon. Actually, in "Genealogy of Morals" Nietzsche claims that the words for "bad" in all languages have evolved likewise.

Way off-topic, I know. Whatever.

Here Come the Commies...

[Nimrangul]

Ok, I know you folk out there in will strongly dislike this idea, but I think that the phonelines should be taken from the phone companies.

I think governments should control them and regulate phone costs to something reasonable. As it is all the phone companies as they are split up are just baby Bells, with their own small monopolies for local phone work, just as the old Bell had it's own big monopoly.

Mind, I also think that water, power, heating and basic television and radio services should also be under the domain of a government controled company. So my opinion is a little more left on this matter than most people's.

I don't quite buy his argument

[Anonymous Coward]

The argument he makes is that big providers will offer their own VoIP offerings, and will give their VoIP traffic precedence on their networks, in turn degrading service for all other traffic (and thus, competitor's VoIP traffic).

However, without realizing it, he also explains why it won't happen. He argues that currently, all traffic is routed using "best effort". His argument then sxtends that these large organizations will effectively restrict other VoIP traffic as they give priority to ther own. I don't see how this necesarilly holds, though.

Imagine a high bandwitch connection. A certain percentage of that bandwidth is the used to service the "preferred" VoIP traffic. This leaves the remainder of the bandwidth to be divided amoung the other traffic. For this to actually affect the competitor's VoIP traffic, the amount of preferred traffic must be large enough to use enough of the available bandwidth that the remainder is unable to service the remaining traffic effectively.

Thus, this practice would not have a significant effect until a large amount of the VoIP traffic is "preferred" traffic - which supposedly would be the goal of starting to do so in the first place.

The only effect that creating "preferred" traffic will have is to provide better service for that traffic. I think that the actual effect on other traffic (even competitor's VoIP), will remain small.

Re:I don't quite buy his argument

[miu]

It doesn't hurt other VoIP providers by making them worse, even if they maintain their current service levels they will look bad in comparison to the lower latency, higher quality offering of the ISP.

Since the ISP can send their VoIP traffic through dedicated virtual circuits (of whatever variety) and offload at preferential peering points (or to another subscriber on the same network) they can deliver a much better experience for their own VoIP apps. No more robot voice, random spots of dead air, or occasional electronic bursts, they can probably even do better e911 implementations - all those things will be very important for mainstream acceptance by people who expect VoIP to work exactly like their old land line.

That is all well within the bounds of legality. Add in the fact that the ISPs will play around the edges of legality in finding ways to actually degrade competing VoIP traffic and cover their asses at the same time and there is an actual problem.

This indeed disproves the myth of capitalism

[Travoltus]

The propaganda that capitalism is the most powerful medium for innovation, falls on its face here.

Capitalism with sensible government regulation is indeed the best path to rapid innovation.

Uh oh... did I just say that?!!

Re:This indeed disproves the myth of capitalism

[BobPaul]

Running Linux is like owning a Lightsaber. It's "a more elegant weapon for a more civilized time."

And it cauterizes as it cuts off your arm...

Re:This indeed disproves the myth of capitalism

[Anonymous Coward]

you're thinking of "chaos" not "anarchy"... do some reading.

Re:This indeed disproves the myth of capitalism

[Slack3r78]

Do the mods really not understand capitalism to the point that this troll actually got modded up? Capitalism in its unregulated form trends toward cartels and monopoly. It's hard for the consumer to make a choice when there isn't another one that's doing anything differently.

That's where the problem lies, and why your parent poster stated that capitalism needs some level of governmental regulation to be successful. Or would it be better if Standard Oil and AT&T hadn't been split up?

Re:This indeed disproves the myth of capitalism

[FleaPlus]

Anarcho-capitalists would argue that the combination of anarchism and capitalism is the most powerful possible medium for innovation. ;)

Re:This indeed disproves the myth of capitalism

[UranusReallyHertz]

The biggest distinction between Laissez-faire capitalism and anarchy is that the former depends on a state and police to enforce property rights, along with myriad other laws like murder, rape, civil courts, prisons, etc. Who handles all that in a truly anarchic society? The concept of private property would be severly limited to only the things you can prevent other people from taking, and it would be a really nast free-for-all since there would be nothing to dissuade others from trying, except maybe lethal force from your gun. The closest thing to anarchy i can think of is extremely tribal societies with very little law enfocrement, like in rural pakistan and afghanistan (where guns are VERY abundant, along with bombs and RPGs), and the chaos of Ethiopia.

Re:This indeed disproves the myth of capitalism

[hany]

Well, IMO capitalism is good but has one quirk which nobody solved yet:

Capitalism stops working well when "the thing" stops growing because there is no space to grow left anymore.

There are stages (but beware, I'm not economist, that's just my observation or my opinion, whatever):

1. While "the market" is in development and there is a lot of "land not taken", there are lots of businesses wich are growing and "taking the land". And there is competion and all the "fruits of competiton" which are good for customers.

2. Once "all the land" is occupied, bigger businesses start to either eat or kill smaler ones. At this stage there is still competition but it's dissapiering as the number of businesses is dropping.

3. Finaly "all the land" is occupied by one or very few businesses and that's when "the shit hits the fans". And that's what have to be solved somehow.

One obvious and "easy looking" solution is to make "the land" bigger. But that (at the end) effectively means to make more people for which we need to expand into space. With that approach we can solve, mitigate or avoid "stage 3" till we reach another limit (like we fill all the glalaxy).

Another ideas?

That makes it worse...

[katharsis83]

Actually, adding on another layer of encryption makes the problem worse. From the article summary:

"...the beauty of this approach is that they're not explicitly doing anything to the 3rd party service applications. They're just identifying and tagging their own services, which is within their rights."

The service providers are prioritizing THEIR VoIP traffic; so unless you can encrypt and then mask your VoIP service provider's packets to look like the ISP's, all encryption will do is increase the latency for voice - remember encryption/decryption requires time. The ISP doesn't explicitily delay Vonage's packets, for example, it simply upgrades the QoS priority of their own packets; this conveniently screws over 3rd-party providers like Vonage while not getting the ISP's in legal hot water.

Encryption can protect your 3rd party call from evesdropping, but can only increase latency under this new sneaky scheme.

Begs for clear labeling

[cait56]

This isn't an issue that requires direct oversight.

It requires clear labeling of products so people know what they are buying.

One set of ISPs offers "Internet Service", by which they mean access to the web, and then a collection of other services that they will offer.

And there is nothing wrong with them offering that service. It is what many, perhaps most, customers want.

The problem is that it is not the "Internet Service" that others want, including most slashdot readers presumably. Which is basically unrestricted access to the Internet with at most a total bandwidth constraint (and protect-the-net restrictions like no forged packets).

If an ISP is clearly labeled as providing "Internet Access" then they could not violate their service guarantees to you to favor their own traffic. If you want to use Vonage, host a server, select your own email provider, or any of a number of things that "power users" find desirable you would look for an "Access Provider".

If you only have a vague idea of what the difference between VoIP and email is, then you probably want a "Service Provider" who will provide you with services and take responsibility for integrating them.

The key problem right now is the ISPs are bluffing at providing open access to the Internet. There is probably a strong case that stealing from the common pool of "best effort" capacity without explicit disclosure.

But the solution is not to restrict what business Service Providers go into, it's to make sure they clearly label what business they are in.

Gets Worse

[MBCook]

In his newest article, he talks about the Burst.com settlement, but in the last 3 or 4 paragraphs he gets back to the topic of this story... including this little titdbit:

"And there are other dirty tricks available to broadband ISPs. Telecom New Zealand, for example, is reportedly planning to alter TCP packet interleaving to discourage VoIP. By bunching all voice packets in the first half of each second, half a second of dead air would be added to every conversation, changing latency in a way that would drive grandmothers everywhere back to their old phone companies."

Re:Gets Worse

[The Vulture]

Well, hell, anybody using TCP for voice communications gets what they deserve. I seriously hope that Cringely meant UDP.

TCP is a poor choice for VoIP, because of the reliability factor (believe it or not). With something as free-flowing as a phone conversation, you would rather lose a packet here or there than wait for retransmission delays caused by TCP.

-- Joe

Re:Tag your own packets?

[The Vulture]

Well, you've got the idea correct, but it has nothing to do with how DSL can run aside POTS. DSL can run along POTS because one uses the low frequencies, and the other uses the high frequencies. If you have DSL on your phone line, and don't have the filter, you'll likely hear a hissing, that's the DSL signal.

However, should you get a combination cable modem/MTA (the VoIP box) from you cable operator (i.e. Comcast), it works like this:
* The DOCSIS 1.1 specification calls for a nifty little feature called "service flows". Service flows have their own QoS, and can be triggered by a variety of criteria, including TCP/UDP port numbers, Ethernet frame type, etc.
* From there, the cable operators will provision two (or more) service flows for the cable modem. One would be for the voice, which would receive the highest priority possible (but with a lower bandwidth), and the other one(s) would be best-effort, with a higher bandwidth allowed.

Cable operators can also use this to throttle any arbitrary connection (i.e. P2P), and in fact, have done so in the past, I would imagine.

A "side effect" of this would be that Vonage boxes would considered as best-effort, simply because they don't get classified into the voice flow by the software of the modem. This is because they won't meet the characteristics of the voice flow.

-- Joe

Two reasons: open VOIP will survive

[aqui]

Technology and Free Market Competion

1) Free Market forces:

As you all know the ISP business is a very competitive business. If I am a paying customer and I am paying for high speed internet access, I will get this from my provider. This suggests that my packets will get these preferential tags for my internet (http, port 80 access).

2) Technology
Now if I use a VOIP software program that happens to:

(a) encrypt traffic (err like Skype for example)

(b) happens to run its traffic over an http proxy like mechanism through port 80 (which automatically separates the VOIP traffic from browser traffic), how can the ISP distinguish my VOIP packets from my internet packets?

The answer is as far as I know they cant (I'm not a VOIP expert, so please correct me if I'm wrong). I'm guesing they cannot distinguish a long high bandwidth legitimate transaction (which I am paying for) from a VOIP conversation.

It sounds like to me that innovation has changed the business model in the telecomunications industry, and players that missed the boat are now trying to compete by blocking these innovations...
However since they're not innovators they don't understand that theses bumps in the road will be simply be innovated around.

We heard this same argument in a different flavor about people being able stopping P2P filesharing before.

But hey what do I know. ;)

Also

[Sycraft-fu]

If the amount of prefered VoIP traffic was enough to screw over non-preferd traffic as low bandwidth as VoIP (80kbps in the heftiest implementations I've seen), it would also screw over all other non-prefered traffic including normal web traffic, FTP, etc. Well I don't know about the rest of you, but I get pissy if my transfer rate drops below 300KiB/sec, if it was less than 10Kib/sec, I'd be looking for a new ISP the next day.

I'm not saying I particularly agree with the practise, but I hardly see it as being able to kill VoIP. If I have a fast broadband connection, I'll have more than enough bandwidth for VoIP. If that gets cut back, well then no reason to pay for it right? I'll jump ship for someone else.

i can see the headlins now....

[teksno]

NEW CAUSE OF DEATH: LAG!!!

operator:"hello, 911 emergency."

random person being killed my a madman: "HE...... ......LP... .........me. my address i...(10 minutes later)lane. HEL...P ME.....pleas.....eeeeeeeeeee"

operator: "miss can you please repeat that."

R.P.B.K.B.M.M.: I SA.......

hours later police arive on the scene to find a psycho wearing our poor victams skin as a trendy new blazer. The coroner arrives shorlty there after and rules that the cause of death was none other then...LAG!!!

wow i guess gamers had it right all along. lag really does kill

But will it be bad in practice?

[louarnkoz]

Much is said about "quality of service", but in practice the Internet as been working quite well without any of that. In practice, Skype does work, sounds better than the average telephone, and does not use any particular priority labelling.

The "best effort" service is far from being a "bad effort". The users want to download files fast, so the ISP has to oblige and provide bandwidth. They want to play video games, so the ISP has to oblige and provide good latency. Guess what, voice over IP requires less bandwidth that downloading a file, and is more tolerant to latency than playing a video game.

In practice, we have been observing over the years a "raising tide of quality". The speed of the average connection over the Internet is more or less proportional to the speed of the user connection, because it is what the users expects. 20 years ago, 9600 bps was considered great. 10 years ago, 64 kbps. Today, users expect to use the 256 kbps of their broadband connection. Tomorrow, users will probably get connected through 100baseT Ethernet, or 50 Mbps WIFI. Yet, voice barely needs more than 20 kbps.

There is no doubt that some ISP somewhere is concocting some evil plot, but the chances are that the evil plot will fall on its face. Probably not much to worry about.

same mistake all over again

[idlake]

You don't need special QOS guaratees or priorities for VoIP: regular TCP/IP service is more than enough for VoIP; if they degrade regular TCP/IP service to the point that VoIP doesn't work anymore, games and all sorts of other applications won't work anymore either. The thought that voice needs special networks or service classes is why telephone companies missed the boat on VoIP in the first place--they just didn't get it.

The only way to kill VoIP is through explicit, service-specific filtering, and that's technically hard to do in general, and quite anticompetitive.

Re:And kill the net as a whole?

[The Vulture]

The best way to think of it is really like a FIFO queue, or standing in line at the Post Office.

All packets that the ISP favors (their own VoIP packets) go first in line. All other packets have to fight for a spot in line. (Non-VoIP packets are treated the same as every other packet*).

Now, assuming that there's enough spots in the line for all of the packets, nothing is dropped. The ISPs VoIP packets go out first, giving them a slight advantage, but everything goes out. If there aren't enough spots, then some of the packets get dropped.

*In practice, this isn't quite true. There are also packet priorities built into the IP specification, and it is likely that VoIP packets are using these as well. Therefore, the line would really look like this:
1. ISP approved packets
2. Non-ISP approved packets with high priorities
3. Every other packet.

Once these packets leave the ISPs network, it's "catch as catch can" again, however, it is likely that the ISP voIP packets have IP priorities as high as, if not higher than the non-ISP VoIP packets, causing them to still have a slight edge.

-- Joe

WRONG

[WindBourne]

we are where we are at, because of gov. regulation. Gov. allowed monopolies to be held and consolidated At first it was ATT. Then, they allowed a small number of cable companies who are quickly becoming just one company.

The way out of this, is to either forbid monopolies(as in, allow competition in), or minimize the monopoly. Personally, I think that by minimizing the monopoly (fiber/cable to the home from the CO; NOTHING ELSE), society will be furthered as the interesting piece is in the service.

Nobody wants to supply residential access

[r00t]

There just isn't any money in selling pure access to:

• clueless types who should have bought service
• a few odd nerds
• spammers

Even the nerds won't buy it, because normal service is way less expensive.

Regulation is required because competition has been blocked, both legally by the government and economically by the prohibitive capital costs. You can't just get a business loan and start stringing fiber all over town. Probably you'd go to jail. If this were possible, the sky would be blacked out by overhead cable.

capitalism isn't dead, but ...

[bezuwork's friend]

If an ISP starts quashing VoIP traffic (or not handling it properly), consumers will, if it matters to them, move to someone who does things right.

That's part of the point here - if ISPs do it quietly enough, most consumers might not realize it. And for many that do, there's always that service contract - $99 if you stop the service before a year is up, for Verizon, IIRC. $99, I doubt that many will incur this cost in order to switch to a different ISP just for VOIP reasons only.

An interesting throught

[bruns]

An interesting thought - people are starting to get their 911 service through VoIP.

What if, god forbid, because of providers tinkering with QoS, someone needs to make an emergency 911 call and can't or results in a call thats utterly unable to be understood?

Wouldn't that make the ISP in question doing the tinkering liable for interfering with a life or death situation?

*Cringely's* not being fair - or accurate

[billstewart]

Yes, there are some ISPs that want to kill every application that generates upstream traffic, so that any consumer in their right mind would buy services from somebody else. And there are third-world telecom monopolies or ex-monopolies that use their position to strangle new competition. (In the US, the worst offenders are mostly cable TV companies, but even in the bad old days of Excite@Home banning anything server-like, they understood that the main reason people bought their service was to file-share pirated music. And in the Pacific Rim, unfortunately Australia's telecom industry has a third-world attitude toward data users.)

But fundamentally, the things Cringely's complaining about aren't accurate, because he doesn't understand the technology or the resulting economics. Yes, telcos are dealing with the threat of VOIP, and it's making their heads explode, and VOIP is much *much* harder to integrate with an old-fashioned telephone infrastructure than to run as a pure-VOIP business. (The technology's difficult, making it scale is difficult, different parts are centralized or decentralized, all the assumptions about who hands money to whom are different, the regulatory infrastructure doesn't match well at all, etc.) And the telcos are making sure that their data networks will support any VOIP services they develop with as close as they can get to traditional telco voice quality, and they're not sure how to deal with the fact that cellphones have convinced the public to accept lower-quality calls and newer codecs with much higher frequencies can support speakerphones much better.

Some big ISPs happen to be owned by telcos, or by telco-wannabees like the cable TV companies. Most of them are working on adding CoS capabilities to their backbones, but that's the least critical part of the network because most of them own their own fiber plants, and it's cheaper for them to burn more wavelengths on their fibers than to add fancy engineering capabilities to their routers or to hire fancy engineers to run them. It's the friendly mom&pop ISPs (that Cringely's not worried about) who are most likely to have backbone congestion issues that need CoS support to prioritize VOIP over best-effort data applications, because they're running at a different scale and don't generally own their own fiber networks.

The places that CoS matters most are the skinny parts of the network - the ingress from the customer's premises to the ISP's POP, and the egress from the ISP's POP to the customer's premises. The ingress direction is really a customer hardware and management problem, making sure that VOIP packets get on the wire before data packets, but service providers (including Vonage) typically handle that by forcing the customer's data through the same box that converts traditional-phone signals to VOIP, and software-based providers like Skype handle that inside the user's PC. This doesn't require the ISP to do anything, though it's sometimes cheaper to build those capabilities into the DSL/cable modem.

The egress direction can benefit from CoS marking, or from other fair-queuing systems that share bandwidth between remote sites or protocol types, or even from dumber systems that prioritize UDP over TCP. In a symmetric environment, like most business T1 connections, this is the most critical part of the system, because data applications can drown out voice unless there's some QoS approach. But most consumer connections are asymmetric, with much faster downstream connections than upstream, so there's less of a problem. Also, in most home applications, if downstream bandwidth is the bottleneck, it's usually because of some application like downloading music that can be turned off or slowed down during phone calls, which isn't a practical approach in most multi-person offices.

Cringely's arguments are especially bogus because the impact of backbone QoS / CoS features on network performance is much smaller than the impact of slow upstream connections in ADSL and Cable Modems. A 12

As someone who's worked on archetecting this ...

[Ungrounded Lightning]

As someone who has worked on archetecting this, let me clue you folks in.

Giving quality of service guarantees means you treat some packets better than others. There IS no alternaitve.

You do this because some packets are more VALUABLE than others. Voice packets, for example, are FAR more valuable the file transfer packets - but only if they receive preferential handling. Delays and drops just slightly slow down a file transfer, but play HELL with a phone call.

Voice packets are also a drop in the ocean. A two-way phone call, with no compression whatsoever, is less than a megabyte of payload per HOUR. So giving its packets preference over, say, file transfers, won't even be noticed. Even giving it priority over best-effort VOICE traffic won't be noticed - except maybe in the very narrow pipe from the edge to the customer - because it won't interfere when there are no fat transfers going on, and when there ARE fat transfers the best-effort voice connection will still be broken.

If some packets are to be treated better than others because they're more valuable, it's fair to charge more for them. (Why should people pay as much for a packet that gets second-class treatment?) This also lets them subsidize the plumbing for the second-class packets.

ISPs only get a little for supplying fat dumb connectivity. They're looking for ways to sell "value-added services" to enhance their revenue. Providing a phone-network quality connection at far less than phone-company costs and prices is a good deal both for them and their customers - they can split the savings with their customers and both come out ahead.

If they're providing an extra-cost VoIP service, they are involved, not just in the payload traffic, but in the connection signaling. This makes it easy to identify the payload flows that need special handling. To do the same for other people's traffic they'd have to spy on the traffic to identify it - and then give it preference equivalent to their own extra-cost packets, for free? Why should they do extra work for free to help their competition? (Especially when it involves spying on the traffic and its routing, which some people might not want?)

What CAN be done, at a profit all around, is one of the following:

- The VoIP providers and ISPs can engage in agreements to handle each other's voice traffic at higher quality of service, and split the extra fee.

- Protocols can be arranged for a client application - VoIP or otherwise - to negotiate higher quality of service (at a higher fee) for its flows, and the ISPs can again engage in suitable contracts to handle the traffic prefferentially and split the extra fee. (This generalizes the service, uncoupling it from strictly VoIP applications.)

You wouldn't have to have a single tier of extra-price service, either. There are different levels, at different price points, that would be useful. (Even within VoIP: POTS emulation at a level that can handle appliances like FAX machines and 56k modems {without reencoding bridges} requires very tight guarantees - essentially every packet must go through with a tight limit on delay variability. Something suitable for compressed voice can accept more drops and jitter.)

And anybody - peer-to-peer or budget service - who doesn't want to pay extra to get their packets special treatment can still take best-effort delivery, and get service about like they get now. VoIP traffic is a very small drop in a very large bucket. Except at the very edge (like a narrow-band drop from the edge router to the customer site), giving company VoIP packets preference over non-company VoIP packets won't appreciably affect the latter: They'll still get through if there's no fat application competing with them, and still get creamed when you're downloading a file or browsing the web.