Bandwidth bandwidth bandwidth

[mik3cap]

These are the three most critical things for the future of personal computing. Why? Because the next thing coming down the pipe is massive, massive processing power. Which is great! We can keep bending Moore's Law over all day long, and make humongous irons available in people's living rooms and offices. We can do all kinds of great calculations... but what are we going to do with all this calculating power?

People will tell you that the possibilities are limitless - with huge data sets, you'll be able to figure out amazing things and make predictions and perform calculations previously only available to giant enterprise computing teams. But here's the root problem: you can crunch data all day long, but where are you going to get the data from? Well, unfortunately, you're going to spend all day downloading that data from the "wonderful" intarweb0rs. Bandwidth is going to become a huge problem; you're going to have this awesome computing monster at your feet, but have nothing to feed it!

People are going to want to keep massive databases locally so they can do cool computing tasks - so storage is a key thing, but thankfully terabytes aren't that expensive any more. I'm sure we'll see 1TB drives as standard shipping options by next year, if not the end of this year. And the advent of solid state storage will also make accessing data faster, so that's awesome too. The only piece missing from this equation is faster networks! Sadly, the computer network infrastructure of the U.S.A. is pretty pathetic when compared internationally. If our country has any hope of competing, we need to get on the stick and dismantle the network monopolies, institute real competition, and get innovating!

Comments

[harrison-ripps.livejournal.com]

FiOS isn't available in your neighborhood, yet? The top tier of FiOS right now is 50 Mbit/s. Admittedly, that's still almost 2 days to download a terabyte, but then again, that's 2 days to download a terabyte. As opposed to using a telephone modem and spending 1.5 days to download 100 MB.

And frankly, if I were gearing up to process a terabyte of information, I'd want to figure out a way to queue up the work so that it could start while the download was going, because I bet whatever I'm doing will take more than two days to finish. Hell, even running a checksum on a TB-sized chunk of data could take a while.

But I think the demand for high bandwidth amongst the population, say, of your apartment may be higher than that of the population at large. HD video webcasting will push the margins faster than Joe Sportsfan's burning desire to capture and recompute the statistics for every game ever played in every sport ever invented.

There is a real risk that by changing to optical transport, Verizon has gained some immunity from the FCC in their bid for world domination. However, for the masses, cable modems and DSL won't die until FiOS is 1) actually available to them and 2) cheaper than what they've got. I think that it is going to take Verizon a while to get the same sort of nationwide saturation that the cable infrastructure has...

[mikecap.livejournal.com]

FiOS is only available to a small population currently. I don't have it, my parents don't have it, rural populations will be lucky to ever get anything at all. These speeds still don't really compare on a "per capita" basis with what's available to people in places like Europe and Southeast Asia.

And it is undoubtedly true that people on the leading edge will have higher demand - but look at the trend in games. You see stuff like Spore coming out, where simulations will use and generate massive amounts of data, and graphics processing generally having ever higher demands; developers will take advantage of the local processing power, and bandwidth will just continually suck, so the imbalance between the leaps being made in processing and the crawls being made in bandwidth will just become more and more apparent.

Plus who can say what other applications will be made when processing power moves to higher and higher limits? Maybe video applications and real time rendering/animating will be popular with the YouTube set (and again I point at Spore and machinima for examples of movie and character making, those are really easy applications that any Joe Nontechie can use)? It's hard to say what people will do.

[narnarthinks.livejournal.com]

Was just thinking about this -- maybe exclusivity with AT&T was the price Apple had to pay to get any network to agree to be the carrier at all. Maybe the networks see the writing on the wall and are reluctant to support internet amounts of data transfer. All this time users have been clamoring for network choice, but maybe bandwidth considerations were driving the politics of Apple's decision to contract with AT&T.

[mikecap.livejournal.com]

It's always about money. The carriers don't want to invest in infrastructure, because that will reduce their profits. There's no incentive to build better networks, whether it's a wireless or cable provider. Our system only encourages monopolies, not competition - if it were truly a free market, companies would be competing by building bigger and better networks and innovating ways to make them cheaper; instead, we have fat and slow behemoths that only care about milking their customers and providing the minimum possible level of service at minimal effort.