Hi Bob,

While it’s true that carriers have a tendency to see QoS as a means to lower their support costs, the technology being promoted by HomeGrid Forum and currently being standardized at ITU-T as G.hn/G.9960 (which recently achieved a significant milestone) is not for carriers only. HomeGrid technology will be used for Consumer Electronics applications (such as streaming HD content from your Blu-Ray player in one room to a display somewhere else), PC applications (such as backing up files from all your computers to a NAS box at hundreds of Mbits/s over your home wiring) and Telco/Carrier applications (such as streaming live IPTV content from your broadband connection to your TV or STB in your living room). All those applications (specially if used simultaneously) would benefit from some kind of coordinated QoS to ensure a good user experience. This QoS could be managed by the user (if he/she chooses to do so) or by the carrier.

Chano Gomez
DS2

A key element of the Internet is the decoupling of the transport protocols from the application protocols. This is what has enabled the rapid evolution of each. The reason I cited IEEE-1394 was that it’s the antithesis of this approach in building in application considerations into the basic infrastructure and thus preventing unanticipated solutions as well as limiting the architecture. This is one of the problems with QoS and why it can’t work unless you lock down the entire path and even then you can’t really control all the aspects. As I’ve written in http://frankston.com/?name=IPGENI you can’t solve application level problems entirely within the network.

Today applications work by adapting to the dynamic conditions rather than having a brittle dependency on the transport.  The danger is in assuming the goal is to solve a list of enumerated requirements rather than create the opportunity for the discovering new possibilities. Too bad good engineering is often perceived as solving the problems as stated and, in the name of local efficiency, nor more than that.

As to unifying the physical layer – perhaps there is value in coming to some agreement for sending bits over the power lines within homes but it’s unclear why we’d want to apply the same solution to all physical media when we need devices to bridge them and we already do find by normalizing at the IP layer. Better to take be free to evolve each according to its particular strengths.

Bob,

I agree with you that in general applications that want to be reliable and provide a good user experience should not rely on QoS features of the underlying network. By not making any assumption about the quality of the network, they can work (more or less) well under many different conditions.

That does not mean that underlying PHY/MAC networks should not try to provide some level of QoS, and then let the applications/services decide whether they want to make explicit use of QoS features or not.

For example, when designing an IPTV delivery system (application layer), it’s a good idea to deliver HD-quality content when the network is not congested, and have mechanism to fall back to SD-quality when the network is congested. At the same time, it’s also a good idea if the underlying network (PHY/MAC layer) tries to ensure a consistent connection (in terms of bandwidth, jitter, packet loss rate) so the the high-level application can experience a “non congested network” most of the time.

Regarding your comment about “it’s unclear why we’d want to apply the same solution to all physical media”: I tried to explain this in this blog post (”Why do we need a unified standard at all?”) a few months ago:

Having a unified standard will increase competition and decrease costs:

• Silicon vendors can address two or three markets different simultaneously, with a single chip. NRE costs are reduced, unit volume is increased and per-unit cost is reduced thanks to economies of scale.
• Silicon vendors that used to supply to different, isolated markets (to the powerline market only, or to the phoneline market only) will now compete with each other for the single unified powerline+phoneline+coaxial market, accelerating innovation and driving prices further down.

Once we take all these factors into account, it’s clear why having a unified standard across different physical media is the best way to ensure that the wired home networking industry can develop its true potential.

Also, a unified G.hn/G.9960 standard simplifies the decision process: some vendors don’t deploy wired home networks today because there are too many industry specifications and they don’t know which one to choose, for fear of choosing the wrong one. By having a single standard, you automatically remove that problem.

There is traditional telecom/CE issue which is about delivering products at low cost to a large market. The key is to find the commonality that creates a large market for silicon and while creating opportunity for the unanticipated. Normalizing on simple packets without complex protocols may be such a point.
The problem as I see it is that here is no problem except for a spat over using the home power line as a particular medium. I don’t have a big stake in either solution but there does need to be some commonality.

Perhaps there is a cost advantage in having a single chip for all physical media. But that’s not a solution to home networking because the new chips will have to interoperate transparently with the existing ones. What I am concerned about is idea that providers should be responsible for how our homes run and they cannot fail to use that power to their advantage.

The lesson of the Internet is that empowering people to find their own solutions is a far more powerful dynamic.

In fact, why do we have providers at all – shouldn’t communities do their own networking rather than being limited to what benefits the revenue model of distant providers?

I’m asking this rhetorically rather than trying to debate at this point since I’ve written more than enough at my site and hope to post an overview essay in the near future.

“Why can’t we just run our own network wires to their set top boxes”

Who says you can’t run your own wires?  You CAN run your own network wires to your heart’s content, but most people don’t because they don’t really care to build a home LAN.  What they want is to be able to watch TV and if a trained technician takes care of that for them, all the better.

So it’s not a case of Verizon “taking over” your home network; it’s Verizon building you a home network where nothing existed before.  That does not prevent you from building your own home network using your own CAT-5, CAT-6, wireless, or Ethernet over power plug network and plugging it in to the Verizon FiOS service.

They only support Coax to the STBs and their VoD makes it very difficult to use a dualWAN router. I can only run the wires the way they choose to allow me to.

I am the professional and they are newbies who are imposing their naiveté on me and reducing their costs by shifting the burden to me.

I wouldn’t necessary conclude that they are trying to take over the home network and are limiting your internal cabling options simply for the purpose of control.  I too am a professional networker that would rather have options, mostly would rather they dropped off fiber to the home with a wireless link that served the whole house and all telco / TV utilities.  But the reality is that the majority of their customer base is not networking pros that can deal with that, and a telco / TV provider must choose the path of least resistance, simplicity and consistency in order to keep costs down.  They choose the obvious, to bring FiOS fiber to the door then use the usual internal infrastructure, coax, to complete the internal connections.  If Ethernet or fiber in the house were more prevelent, they would add that as an option, but it is not.  If their installers had to deal with too big a variety of options once they got to the home, costs would go up, and that would get dispersed and passed through to all customers, even those not interested in having many (or any) home cabline options other than the existing coax.  The majority want the provider to do it end-to-end and just turn it on and it works.  It’s only a smaller percentage like us that want flexibility and options.  It will get there eventually but will take a long time.  Builders are still not really making Ethernet or fiber the standard internal cabling in a house; you have to ask, push / force them (and pay) for it.  Of course, I still can’t get FiOS despite living in a populated area where many adjacent neighborhoods have it, so I wish they would just work out that (political) issue first!

“If Ethernet or fiber in the house were more prevalent [sic], they would add that as an option, but it is not.”

And that’s the problem—the dead hand of the past and its death grip on the future.
Network is it Anyway?

They require Coax to the STB today because they had to support analog broadcasts and TVs without any set top boxes.  Once they stop supporting analog broadcasts, then it may not be necessary to use Coax.

Still, why the obsession with replacing that coax network?  The point is that you can still run your own network in parallel with that network.  As far as the complexities of dual-WAN setup, there are ways to do it and no residential broadband provider supports it today nor should they have to because the market in general doesn’t ask for it.

A lot of us love to hate telcos and cable companies, and typically for justifiable service-related and/or price reasons.  But sometimes this unnecessarily morphs into speculation that would make Hollywood blush regarding their sinister motives.  It is typically oriented around anti-competitive practices that they are consciously executing, but is often spoke of without any real tangible supporting evidence and usually in the face of other very credible evidence of a simpler explanation.  We went through this (and still are going through this) with all the hype over the Comcast/FCC/BitTorrent saga and all the speculation around their anti-competitive practices.  I have no love for the legacy utility companies and wish there were more, better, faster, cheaper, more flexible and higher-tech options.  But I also don’t think what is being said sometimes is fair even if it is interesting speculation.  The reality is that we are talking about very large legacy telco and cable companies that have trouble with even simple things like getting the bill right or giving you an installation window more narrow than 9am-5pm!  So I’m not sure they are really capable of executing some of the anti-competitive strategies that so many people assume they are.

In this case, it is quite simple.  A telco made an extremely bold move to invest billions to deploy fiber to the home, something that has been talked about for over a decade but has not yet really happened.  This should be applauded.  It is a big investment and big risk for them.

(Side note:  If anything, more should be said on how great a thing this is and how other carriers are stifling service capabilities by continuing to invest in technology advancements over existing copper or coax infrastructures.  When DSL was still theoretical, everyone was talking about how it was going to be just a stop-gap measure to get us beyond what analog dial-up modems could deliver over copper wires and the telco network until fiber got to our front doors, which so many in the mid-90’s were predicting would happen by the end of that decade.  We’re near the end of the next decade and it still hasn’t happened.)

But at some point the carriers have to contain the business model to something reasonable.  Verizon have chosen not to extend the fiber past the demark at the front of the house and have chosen instead to make use of existing coax to deliver the last few feet.  (This is not unlike the obvious selection of telcos using copper to deliver DSL-based Internet broadband service and cable TV companies choosing to use coax cable to diversify into broadband as well.)  For the vast majority of their customer base, this is more than a sufficient solution.  There is very little demand right now from the average customer for more than that, and it is very difficult to build a viable retail business model that allows for a large number of variations and customizable options tailored to a few unique end users.  Not everyone is Starbucks!  If they also decided to cater to the possibility of various other home infrastructures - Ethernet, fiber, wireless, power line - the requirements, capabilities and training (and thus the costs) of their installation team will go up, and eventually that would get passed through in the pricing to us, all to satisfy the minority of their customer base.  Hopefully it will get there and we’ll have a better-integrated “home utility network”, but it’s not practical right now.
 
Regarding the cabling within a house, it is not necessarily the responsibility or fault of the telco or cable carriers that coax or telco-grade copper is still the norm.  It is still not automatic that builders wire new houses with Cat5e/6 or fiber, even though it is certainly easier to do before the dry wall goes up.  That is something you have to push them for and typically pay extra for.  This is something that needs to change too, although wireless networking may be making it a moot point.

Hopefully at some point new options will emerge for home networking that incorporates all utilities, but until there is better demand from the average consumer and not just the super user to create the necessary volume demand, we’ll have to wait.

I don’t hate the carriers as much as I feel their pain and we suffer their problematic business model. They are in an untenable position in having to fund their infrastructure by selling services. It’s a structural defect and that’s why DSL has had one bump when they created it twenty years ago for ITV and then mysteriously it has shown no progress since.

If they didn’t do both the wires and the content then there would be no “catering”. And, as I demonstrated with home networking, there would be a vibrant marketplace for solutions from DIY to pay for a service plan.

Imagine catching up with twenty years of Moore’s law from the days when CPUs ran in the low Mhz range to today’s hyper threaded quad CPUs running 3Ghz. And from when home wires were modem speed to today’s gigabits over copper.

So let’s befriend the carriers and help relieve them of the burden of putting billions at risk in betting against the empowered users. Let’s recycle them into connectivity companies and service companies.

As I’ve noted we can go around on this again and again so best to go to my site then repeating it all here.