the "grid" 10,000x faster than cable internet connections?

[Thorgrim]

Techtree.com India > News > Internet > The Grid to Render the Web Obsolete?

does anyone have any idea when the upper-middle class will get access to this....years or decades?

[Simius]

That sounds cool. I remember being excited about 56K connections.

[Dumpy Dooby]

Originally Posted by Simius

That sounds cool. I remember being excited about 56K connections.

Jesus. I was still shitting my diaper when 56k came out.




I was also 14yrs old.

[WickedLou9]

I remember dialing into BBS services at 2400 baud.

[Ardentfrost]

Originally Posted by WickedLou9

I remember dialing into BBS services at 2400 baud.

Word.

BBS's were such shit. Downloading little jokes, leaving messages, and playing MUDs. And they were ALWAYS busy. I remember leaving the program I used (Communique) on auto redial for like an hour or more before getting in sometimes.

[Ardentfrost]

Originally Posted by Thorgrim

Techtree.com India > News > Internet > The Grid to Render the Web Obsolete?

does anyone have any idea when the upper-middle class will get access to this....years or decades?

This already exists in I2 (Internet2). The problem with either is the up-front costs of instituting it. I2 has good American backbone path... Atlanta, New York, DFW, Kansas City, LA, San Fran, and a couple places in the upper midwest that I can't remember. Georgia Tech has access to it and, of course, it's blazing fast when accessing something that is on the I2 backbone.

I thought it was funny that the article describes hub-and-spoke topology that has only been around as long as Darpanet (it is the topology that the Internet uses).

Here at Georgia Tech to support our immense traffic we have some dozen parallel 10GbE drops that run across town (well, UNDER town) and pipe in directly to the internet backbone (as in, we don't go through a carrier... we actually ARE a carrier for a lot of the State of Georgia's school system). 10GbE is the fastest commercially available at this time. IEEE only agreed that 100GbE is the next target bandwidth in November 2006.

Anyhow, 10GbE goes faster than any non-High Performance Cluster can handle (HPC is almost exclusively used in research installations, and all the ones I personally know about are all connected at 10GbE). That means that it can transfer faster than your hard drive can access information. GigE is actually widely available already (although cost prohibitive to the individual), which translates to 125 Megabytes per second. Most of my albums are about that size, so I could feasibly transfer an album of music in 1-2 seconds. 10GbE is obviously 10x faster.

The problem in America is that consumer ISPs resist upgrading their infrastructure to handle these sorts of bandwidths. They have a phone and CATV infrastructure in place, and make LOTS of money off it. Verizon is the first to offer fiber optic to your house (FIOS), but currently throttle the line to 30Mbps (3.75 Megabytes per second). It's the fastest personal internet you can get without paying thousands of dollars in both up-front and monthly charges.

The resistance to a faster consumer infrastructure has allowed a lot of other countries to zoom ahead of us in average bandwidth per internet user. Granted, smaller countries will have an easier time upgrading their infrastructure, however large cities (especially those close to the internet backbone) have no excuse. The fastest internet I can get from my provider is 10Mbps in Atlanta, GA.

Sorry to rant

[Phantom]

Originally Posted by Ardentfrost

Word.

BBS's were such shit. Downloading little jokes, leaving messages, and playing MUDs. And they were ALWAYS busy. I remember leaving the program I used (Communique) on auto redial for like an hour or more before getting in sometimes.

It crazy how far we came in just a few years back then - going from local BBS to that funny thing "teh internet" and then "teh web."

[Thorgrim]

arden u still didn't give your guess on when customers can get this for say, $50-$100 a month?

[Ardentfrost]

Originally Posted by Thorgrim

arden u still didn't give your guess on when customers can get this for say, $50-$100 a month?

Shit needs to be deregulated or at least regulations need to be relaxed. That could possibly force the current ISPs to compete for the infrastructure, but right now they're just not going to do it.

There are also quite a few logistical problems, though. A startup would need to make a major initial investment, but I think long-term would really pan out. Offering plans between 100Mb to 1Gb would probably be attractive to a wide variety of people and businesses. Modems would no longer be needed by end users, but media converters would be and they run a few hundred bucks. Renting those is, of course, a possibility. Looking at a pricing sheet I got here, I could supply multimode fiber to up to 264 locations within a 2 mile radius at an average cost of $318.06 per user (total cost of $83,969 for that many) not including burying fiber, bandwidth, power to keep the shit running, a place to keep the equipment, and other variable costs.

Let me know if you know someone with a lot of money willing to invest in the project... I'm willing to fight the FCC during the startup process We might be ok as long as we don't offer phone or cable TV as well because then we wouldn't be subject to the FCC regulations regarding those things.

Anyhow, what's far easier is for the current consumer-level ISPs to want to provide fiber to the homes, which isn't likely to happen unless the FCC allows them more leeway with their infrastructure. But then people get mad when AT&T charges another carrier to use their infrastructure...

[Donkey®]

The Local - Sigbritt, 75, has world's fastest broadband


A guy I work with did that for his mother. It was a great demonstration of what CAN be accomplished when you actually try.

[Ardentfrost]

Originally Posted by Donkey®

The Local - Sigbritt, 75, has world's fastest broadband


A guy I work with did that for his mother. It was a great demonstration of what CAN be accomplished when you actually try.

If you have obscene amounts of money and are physically close to a 10GbE-capable area

[Donkey®]

Originally Posted by Ardentfrost

If you have obscene amounts of money and are physically close to a 10GbE-capable area

Before you have a "should we" discussion you have to answer the "can we" part. Obviously having your very own personal CRS in your computer room isn't going to happen, but when you show something can be done, it starts the innovation and adaptation process. At least that's the way I see it.

[WickedLou9]

This doesn't really seem like new technology though. It's just faster versions of existing technology. Both the the major carriers today, Comcast and Verizon have extensive fiber-optic networks at thier cores. My cable connection is throttled down to whatever like 8mb/sec or something, but the ifrastructure is capable of much much more than that. It's just that it;s a shared medium and congestion limits the actual throughput. This "new" technology would suffer from the same problem I would think, even if it is capable of overall greater speeds.

[Ardentfrost]

the ability to take 10GbE (load balanced to any multiple of that) to your house has been around as long as 10GbE has been. This COULD have been done in 2003. Someone doing it now doesn't change the prices or regulations involved. Logistics isn't a matter of doing it once, but doing it millions of times.

[Donkey®]

Originally Posted by WickedLou9

This doesn't really seem like new technology though. It's just faster versions of existing technology. Both the the major carriers today, Comcast and Verizon have extensive fiber-optic networks at thier cores. My cable connection is throttled down to whatever like 8mb/sec or something, but the ifrastructure is capable of much much more than that. It's just that it;s a shared medium and congestion limits the actual throughput. This "new" technology would suffer from the same problem I would think, even if it is capable of overall greater speeds.

The problem is that their core can't handle that much traffic. Even if they let everyone loose they only have so much transport bandwidth. (I don't even know what their network backbones look like...my guess is OC48 at most.) Even my network is limited on transport bandwidth. We just recently turned up our 1st 768 in the past couple of months. The rest of the network is 192. The biggest obstacle to us beginning to upgrade was hardware. Cisco has had their GSRs out for something like 10 years at it's STILL a pos with so many issues and that's not counting the fucking IOS. Once we deployed CRS, got a halfway stable code, we ramped up our conversion. People say "just use another company" and it simply isn't that easy. I am positive Juniper or Nortel or whomever aren't much better and are probably worse. I guess the point is that we can turn up as much bandwidth as everyone needs only if there is a reliable and stable piece of equipment tying everything together. We just don't have it.

[WickedLou9]

Originally Posted by Donkey®

The problem is that their core can't handle that much traffic. Even if they let everyone loose they only have so much transport bandwidth. (I don't even know what their network backbones look like...my guess is OC48 at most.) Even my network is limited on transport bandwidth. We just recently turned up our 1st 768 in the past couple of months. The rest of the network is 192. The biggest obstacle to us beginning to upgrade was hardware. Cisco has had their GSRs out for something like 10 years at it's STILL a pos with so many issues and that's not counting the fucking IOS. Once we deployed CRS, got a halfway stable code, we ramped up our conversion. People say "just use another company" and it simply isn't that easy. I am positive Juniper or Nortel or whomever aren't much better and are probably worse. I guess the point is that we can turn up as much bandwidth as everyone needs only if there is a reliable and stable piece of equipment tying everything together. We just don't have it.

Interesting timing on this

Comcast to Conduct First Live 100G Network Trial with Nortel Optical Solution

Comcast Corporation (NASDAQ: CMCSK)(NASDAQ: CMCSA), the nation's leading provider of entertainment, information and communications, is testing a 100 Gigabit/sec optical solution from Nortel(1) (TSX: NT)(NYSE: NT). Comcast(2) is conducting this test at the 71st meeting of the Internet Engineering Task Force (IETF) conference which it is hosting in Philadelphia.
Comcast's network trial will be first time real traffic is run over a 100G wavelength on its existing network that is also carrying live 10G and 40G links. The trial will run in parallel over the same optical infrastructure that supports a combination of high definition or HD video, Internet and voice traffic on Comcast's 40G national backbone network. Nortel's 100G solution, based on the Nortel 100G Adaptive Optical Engine, is designed to operate with comparable performance to today's networks, enabling operators to dramatically increase capacity.