Wednesday, February 07, 2007

Google throws a Hail Mary pass to the Cable companies!

As reported in Reuters, Vincent Dureau, Google's head of TV technology, spoke at the Cable Europe Congress. Some net-neutrality die-hards will undoubtedly feel betrayed by this Machiavellian business move, but it's smart business politics for Google.

According to Google's Dureau:
"The Web infrastructure, and even Google's (infrastructure) doesn't scale. It's not going to offer the quality of service that consumers expect,"

You read that right -- A Google Exec telling the world their infrastructure can't "scale" to the bandwidth needs of "TV over the Internet" (!)The Reuters story goes on:

Google instead offered to work together with cable operators to combine its technology for searching for video and TV footage and its tailored advertising with the cable networks' high-quality delivery of shows...One cable chief executive, Duco Sickinghe from Belgian operator Telenet, said it was "the best news of the day" to hear that Google could not scale for video.

Google is bluffing, folks. They have more fiber rights than some telcos. They don't want to become a telco and there's a big difference between net neutrality in principle and in application. Network operators should be able to tweak their networks for optimum delivery and QoS (Quality of Service) (without screwing the customer for it preferably)

It's puzzling that Google is speaking for themselves and Joost...According to the CTO of Joost, its delivery system is designed to accommodate relatively lower bandwidth & bit streams. In fact he's pretty bold about the advantages of Joost's delivery mechanism vs. YouTube and Google's:

He's very clear about the economic folly of using big servers to mainline video to millions of desktops. "You can try to cut costs by allowing only crappy little images. You can limit the run times or make money selling hardware. But eventually the bandwidth bills will eat you alive. YouTube, iTunes, and the rest of them haven't got a chance. We just hope they take their time realizing it."

It's ironic that Google throwing themselves under the bus, actively seeking the partnership of the Cable Companies as an internet "white knight." Could it be more of a desperate attempt to build an alliance with them? (preemptively against Joost) It has the added benefit giving them better leverage against the traditional network operators too.(e.g. AT&T, Verizon, BT, etc)

Business is Business...but, Don't be E***, Google



Joe said...

You clearly don't know much about video networks. Whether Google has more fiber than a telco is irrelevant. Cable operators, satellite operators and telcos (to the extent they implement IPTV) are multicast networks - they transmit a show or channel once to as many people as want to watch. This is extremely efficient in its use of bandwidth. Internet video is unicast, and thus extremely inefficient in its use of bandwidth. I know Vincent and he is a straight up guy - he doesn't bullshit. He is speaking the truth, whether you like it or not.

Anonymous said...

Google is not bluffing. They DON'T have what is needed.

Greg Horne said...

Dear Joe,

IPV6 is multicast. Think about it.

Utunga said...

I absolutely agree with Joe, your article doesn't take into account the massive difference between IPTV style multi-casting (each person selects their own stream) as opposed to the broadcasting model (where each person chooses from one of the pre-set stream/s).

Neither the telcos, nor the cable operators, nor the cable operators have anywhere enough bandwidth in place right now to do IPTV.

Lets do the math. For the sake of argument, lets just assume that VOD requires about 4Mbps, and that everything is 100% on demand and uncached.

Lets say that an ISP has, say 20,000 customers in its local neighbourhood, thats 240Ggps of switching to the ISP. And the 'backbone' provider who may be supplying, say 10 big ISPs of this size in any given metro area is looking at a 2.4Tbps pipe.

Thats absolutely huge, and way beyond what *anybody* has in place right now, even for the relatively small uptake (as a percentage) of 200,000 customers in a metroa area. Furthermore, if you want to do this in HD then we're talking 10 to 15 Mbps in the last mile so multiply everything by 5.

Sure, Google could *choose* to massively invest in fiber and scale up to IPTV - but then again, not exactly 'core' business for Google. More's the pity, though, because I hate to imagine how long its going to take the idiotocracy in places like AT&T to figure out how to scale bandwidth that much (or figure out how to implement super-smart caching/m-bone strategies).

Anonymous said...

Why would you center all paragraphs on the page? Do you realize how hard that makes it to read things?

parry said...

H.264 and MPEG4 can push the bandwidth requirements down pretty dramatically. for NTSC video at moderate speeds (not sports) the data rate can be as low as 1Mbps, HD can be as low as 8MBPS. Even DirecTV using MPEG2 transmits at 3 or 7MBPS. While lower than some of the estimates from Utunga, these are still pretty high.

Even google wont want to pay for unicast traffic to tens/hundreds of thousands of users or more. That's the youtube approach only bigger.

Now if your willing to flip from channel to channel like cable and satellite today, you can have a finite number of multicast streams and even a few hundred channels is manageable without too much trouble. Today that works great on a network with one centralized administration. Comcast, or SBC in a given city could multicast their video in a region without too much trouble. In fact this is precisely what the phone companies want the FCC to permit. You may have a hundred thousand users, but there's only a hundred streams they can subscribe to and that's all you backbone has to transmit.

Much testing has been done for multicast on a larger scale but the IP address allocations for IPv4 are pretty limiting considering every channel anywhere on the internet would need one. I don't see this happening without v6.

So as a summary I'm of the opinion that it would not be economical to expand the youtube method to full scale TV. The phone companies and cable companies are naturals to compete with each other on cable or IP without any trouble. Sometime in the future we may see internet TV as the killer-app for IPv6. We learned a lot from the mbone. My crystal ball just shows a warped image of an old monitor though so maybe that's a vivid imagination speaking.

H.264 whitepaper:

Anonymous said...

FYI, this centered-text layout sucks. It makes things harder to read.

David said...

dude, Google is right... They currently can not do what the cable providers are doing. Go here and listen to Cringely's last few podcasts about the subject. You will quickly realize that the internet is no where near ready for everyone to be viewing video or using P2P. Good find on the Google news though. Lets keep their feet to the fire on the net neutrality stuff.

Joe said...

Greg Horne-

Sure, you COULD do multicast over the Internet, but why? The pay TV operators (cable and satellite) already do that quite well. Competing with them by doing mulicast over the Internet isn't going to give you a cost advantage because the programmers will make you pay MORE than they do (which is exactly what they did to DirecTV and Echostar). There's no advantage.

Anonymous said...

So the key is better compression;
Is it true that the upcoming compression algorithm 'DHOSa' squeezes video up to 8x more than H264 - If so would this make it possible ?

Anonymous said...

True true and true.

Maisie said...

People should read this.