Yup. Be a business, get a enterprise grade router doing BGP and just buy IP-transit. It's $2-8/Mb. Buy all the bandwidth you want. This is dedicated bandwidth. I've see $895/month Global Crossing Tier 1 bandwidth for a full GigE (1Gbps) port in a co-location/data-center recently.

FiOS is not enterprise level, it is small business level though the shared PON infrastructure has a shit-load of b/w.

If you're in Manhattan and willing to pay up to $200-$800/month (enterprise class with SLA) you can get dedicated IP-transit of 1Gbps+. This is NOT consumer/small business internet. This is 100% dedicated bandwidth. If it's a High-rise most-likely VZ/Cogentco/XO and a few other Tier-1 and upstream providers probably already have GigE/OC-X fiber termination router in the basement and it doesn't cost them much to hook it up. There are a few carrier hotels with Internet exchanges there (Equinix, 111 8th street, a few others, I frogot.) If this is for a business. Check with your high-rise mgmt. and see what carriers have fiber in the basement. Every single high-rise these days has a fiber-termination gear of GigE/10GigE/OC-X in Manhattan due to business needs of certain firms in certain buildings.

You want 1:1 dedicated bandwidth? PAY FOR IT. It's like $900/month at the cheapest for GigE level bandwidth of IP-transit in a data-center. That doesn't include transport to your location either. Consumer bandwidth is all oversold of 1:10 to 1:50 ratios.

The majority of traffic in South Korea and Japan stay WITHIN the country making it extremely cheap. The US is a large spread out country with a-lot of users accessing international content, etc. It's like 1Gbps in SK but 10Mbps international.

How much upstream bandwidth do you have? IP-transit costs?

Yup Singapore and Japan are technically "1st world" countries in Asia in that their per-capita income of workers is similar to the US/Canada and Europe. Companies have offices in Singapore for both the talent that's similar to what they can get in North America and Europe but also rule of law that they are familiar with.

The above was posted ANON for obvious reasons. A-lot of the research about the GFW is public information anyways.

Just did an analysis of the protocol used and apparently google is using the open source XMPP protocol with some SSL/TLS for encryption on the handshake and plain old g.711 RTP as the voice transport. I could decode my voice traffic from a test call as it was unencrypted but some of the handshake/call setup was. The login for the relay (I'm behind NAT) was unencrypted. I did 2 seperate test calls and the login to the relay (relay.l.google.com, replace l with your local google server cluster/cache), is dynamically generated on the fly with a magic cookie for every call so no unless one manages to crack google's algorithm and how it ties to your google account. Google's public implementation info: http://code.google.com/apis/talk/open_communications.html

I totally agree. Spend some money and do it right. This way you have a stable and reliable network. As mentioned you can EASILY run high-bandwidth applications (Up to 175Mbps or greater) over a Coax network in your house. Just make sure the right protocols/products don't interfere with the frequencies running on the Cable Modem if you're using a cable provider for internet/tv/phone, etc. This is very important. They simply do some modulation (like QAM256 for cable modem's these days over a channel space or similar for in-home frequencies, etc...). For the Coax network you need to make sure make sure they don't have any interference for any of the cable companies products. Some protocols like MoCA are designed to work around this and hopefully the newer G.hn. Others like HPNA3.1 (being phased out and being replaced with G.hn) have interference with DOCSIS for example so you need to put them on separate wiring circuits/networks. I did some research a while back and there's essentially 3 solutions: MoCA: http://en.wikipedia.org/wiki/Multimedia_over_Coax_Alliance (175Mbps) HomePNA: http://en.wikipedia.org/wiki/HomePNA (320Mbps) G.hn: http://en.wikipedia.org/wiki/G.hn (Coming soon...) If you choose a protocol that does have interference don't worry, just put it behind a router/switch and separate them via wiring. For example a Coax data network with a splitter just for your data at one end and the converter at the other (like Ethernet). This makes it simple by keeping the networks separate. If you really want to get around this with the same frequencies as the cable company on the wire you can look into a frequency engineering method device called a Diplexer which is what AT&T uses for their U-verse installations. For all those naysayers that say that Coax doesn't have the potential, their all wrong. Coax has the potential, to have even MORE bandwidth than copper. It just requires signal processing. Like DOCSIS on cable or DSL on phone lines. Ethernet is the "pure" solution without any processing (technically very cheap processing...)

Ummm what? I use AndChat from the Market and it works fine. I can also install any Android app I want even those without official approval by enabling debugging mode and disabling signature verification in the options. Downloading some apps (*.apk files) from 3rd party sources (souceforge, etc.) and it works fine. Verizon's Android platform is TOTALLY OPEN if you know what you're doing I've noticed (aka RTFM/Google). Maybe you're having coverage or connection issues with the server? I even tether through my Droid using a NAT proxy app and haven't found ANY blocking.

Totally agree with everything you said. I suspect the REAL problem at the moment internally at Google on YouTube is one of the following: 1. Bandwidth costs are going to a point where it's unsustainable even for a giant like Google (just compare text versus video). If Google is still buying IP Transit then it'll get real expensive even with cheap bandwidth costs now-a-days. Though this is greatly lessened with me suspecting Google getting alot of settle-free peering with eyeball end-user ISP networks/ASes. 2. In relation to #1, Google still has to maintain and upgrade an ever-growing internal distributed network linking it's datacenter's as video takes lots of HD space to cache over their CDN. 3. The real problem is the business model I suspect Eric Schmidt is trying to work out. Personally for me as a consumer Advertising CAN WORK for professionally generated content. This is one of the reasons I personally love Hulu. It's a great value proposition, allowing me to watch all my favorites with advertising funding it. For user-generated content it's a completely different case. I suspect what Google is really working on (and being pressured by investors to generate revenue) is applying the Googlesque algorithmic magic to to match user-interested adveritsing in multiple forms (visual/flash, txt, gifs, etc.) with the video one watches. My personal estimate is that advertising CAN offset the bandwidth/network/Hard Drive costs but will it sustain the consumer to keep going to the site? This is the REAL question. They need to show a delicate balance of sustaining advertising without driving the user away. User-generated content is not like premium, content where the end-user wants to see it. Personally I'm not going to pay for youtube unless it's worthwhile content. Google *ahem* should if they ever go with this model develop videoRank or something similar to pageRank and pull up the most popular content and monetize those. For me personally, the problem with user-generated content I see on YouTube is 25-50% of it is SHITE! I mean low resolution shite! If I'm going to pay for something it better be 1. A good price, subscription based 2. Be on the NETWORK, this is something Google can excel at, using an OPEN standard so pretty much any device using Flash Player or something similar can access it. They need to develop an eco-system for this. I suspect Adobe is pushing hard for this with their recent announcement of trying to get Flash Player onto STBs in the home living home. 3. Ability to download it. I would accept a bit of DRM but it needs to be RESONABLE and EASY TO USE on the end-user who isn't technical.

I'll let others comment on the hardware and business plans but the problem with rural inherently is that cost of delivering bandwidth from a major city/internet POP to your rural location ala backhaul. Bandwidth is EXPENSIVE when purchased in a rural location unless you have the big $$$ to operate your own fiber backhaul or a wavelength of existing fiber from the POP to your location. It really depends on how rural the location is. You really need to consider the costs and oversell ratio if you're going to do this. Try asking here btw: http://www.dslreports.com/forum/wisp In rural areas the ILEC setup is usually the minimum unless the CO has been setup for DSL/higher bandwidth services meaning it's usually only TDM based. The ILEC has to share the resources with the federally required voice service. And if they haven't deployed broadband out of the CO already then it's not profitable for them. This eventually most likely with the ILEC telling you a very high bandwidth costs and even higher if there's fiber/equipment buildout costs needed. If you're in a rural but not-so-rural location and can do wireless backhaul for the bandwidth (microwave) then you can lessen your costs that way. This is all based on what I read online. Good luck btw.

Mobile Broadband is not perfect and wireless can always be finicky. 1. The problem with Mobile Broadband is the connection between you and the tower/backhaul. This is ALWAYS variable in a wireless environment due to shared spectrum, carriers per sector, # of users on the tower, backhaul bandwidth (mostly NxT1s to the switch site) etc. 2. In relation to #1, Mobile Broadband is a MORE SHARED MEDIUM than wired broadband. Yes all consumer broadband is shared and aggregated at some point but mobile broadband has less available at the air interface. (This would be the currently deployed EV-DO Rev. A 3.1Mbps/1.8Mbps, or ATT's HSPA 3.6 (soon becoming 7.2 with this summers iPhone release being rumored)). Those are THEORETICAL SPEEDS for only 1 person sharing the same sector/backhaul/RF enviroment. In reality with all the traffic you're going to get less than 1/2 that. This will not change until the capacity of the air interface dramatically increases with 4G technologies being deployed (LTE, WiMax). 3. The other limitations already noted, ping times (200-400ms), and bandwidth limitations (5GB mostly). 4. There is a REASON carriers theoretically want to keep mobile broadband from being a replacement for wired cable/dsl/fios broadband due to the fact that the air spectrum is SHARED. Think about it, most cell sites/carriers have only 1 1.25MHz/5MHz sector deployed sharing 3.1Mbit down and less up. Your competing with all the mobile phones data traffic too. Finally the backhaul. Most cell sites only have NxT1s as the backhaul to the switch site. This is changing though with LTE. 5. Personally I would never replace wired broadband /w mobile unless it's an addition. I would wait until LTE as the data rates can dramatically increase matching what the current DOCSIS/DSL standards can do and even more with enough backhaul (once LTE launches the backhaul should be comparable to current cable/dsl/fios).

ISPs need to pay for backbone bandwidth of the amount their customers consume. This is something that most consumers do not understand. In fact your consumer facing ISP or division of an ISP is just a company/division that engineers, maintains and links your home to and maintains a link to a backbone network like Level3, AT&T, Verizon Business, NTT, etc. But what I donâ(TM)t agree with Brett is that itâ(TM)s VERY Cheap especially in an urban area and a mega corporation like Time Warner can definitely afford the 10Gb level interconnects to get costs as low as 2-10cents/GB on backbone providers like Level3 in bulk. It makes it even cheaper for companies like AT&T & Verizon since they own their own backbone networks (they just hand it off to another division of the company and can get great rates due to their Tier 1 power, settlement free peering). Sure thereâ(TM)s a competition aspect per the cable companies but in the end itâ(TM)s all about cost and profit. Right now the in the US itâ(TM)s one of the most competitive markets in the world for IP transit due to the fact that we have multiple backbone providers HQâ(TM)ed in this country and competition. Now what I agree is that it needs to extend to the consumer and access networks. If companies are going to do caps, they better price is in realm of what their actual costs + profit are. Time Warner charging $1/GB is CLEARLY price gouging, in fact I see margins of that estimated at 50-80% including the access network costs. Yes the usage distribution of end users is uneven clearly and light users are subsidizing the heavy users but Time Warner is still making profits so I donâ(TM)t see why they need to start this now. Also another problem with access networks/consumer level bandwidth is that itâ(TM)s shared bandwidth and thereâ(TM)s only a limited amount of capacity on a DOCSIS 1.1 cable network before thereâ(TM)s congestion. If utilization gets high enough there needs to be node splitting (expensive, time consuming), increased bandwidth per node (DOCSIS 3.0) or shaping so everyone gets an equal amount. Itâ(TM)s only fair that way. Personally I really like Comcastâ(TM)s idea of 250GB/month since it balances costs with the current dynamic of IP transit costs. PEOPLE YOUR CONSUMER INTERNET CONNECTION IS NOT A DATACENTER CONNECTION. If you really want to know what dedicated bandwidth costs go price a T-1, DS-3, OC-X, GigE or some other kind of dedicated connection. Of course itâ(TM)s for businesses and service level gurantees but what your really paying for is unlimited dedicated bandwidth. Now personally I wouldn't mind paying a few cents for a GB but $1/GB? Your kidding me.