According to Mike Nelson, the relevance of the Internet in all facets of modern life is decreasing as cloud computing takes the stage. Nelson, a Georgetown University professor, gave a speech at the World Future Society in Boston on Friday, where he predicted cloud computing would eventually be more important than the Internet as services evolve.
"In the mid-90s there was a clear conscience about what the Internet was going to be," Nelson said. "We don't have as good a conscience as we did in the '90s, so we may not get there...the cloud is more important than the Web."
One of the problems when discussing cloud computing is that there's still no exact definition for what cloud computing actually is. If Amazon's Elastic Compute Cloud is cloud computing, what about online backup programs, Facebook, or Flickr? Are XBox Live and the PSN console versions of cloud computing? How we define the term has an impact on how we perceive the future importance of the cloud.
Professor Nelson is confident of the cloud's long-term importance, but we don't share his vision. One of the goals of cloud computing is to make information device-agnostic; your mobile phone would be capable of displaying a real-time 3D rendering that's actually being computed by a massive rack of servers somewhere else. There are significant technical and financial barriers that would have to be overcome in order for cloud computing to actually push the Internet out of the way.
It's no small irony that while various handset providers and telcos profess full support for a content-rich, media-heavy mobile experience, they're quietly killing unlimited data plans and returning to the days of AOL, when exceeding your hourly plan resulted in the company charging you heavily for each additional hour. Traditional ISPs have experimented with bandwidth limitations, but the ceilings are typically much higher than a wireless plan. This is part of what allows cloud-based services to prosper—confined to a standard 3G contract, users might well avoid a data-heavy cloud compute resource so they don't exceed their monthly data plans.
The other major issue is the high discrepancy between upload and download speeds. In the Louisville, KY area, AT&T's best offer is 6Mb down / 512kb up. Translate this into better terminology and that's 750KB/s download and 64KB/s upload. In practice, 500KB/s is the highest download speed we've seen, with the upload pegged at 50K.
If the cloud is going to grow, both consumers and businesses need access to faster upload links. Even at a steady 50KB/s, it would take a person 231 days to transfer a single terabyte of data, assuming the connection never broke. Businesses have access to faster uploads, but such services are more expensive than standard data plans. Finally, there's some question as to whether or not cloud systems can even provide access to certain fundamental Internet capabilities. Social network clouds are easy. Servers-on-demand compute clouds are easy. Can anyone see a cloud application taking over Google we think not. Information search and retrieval is just one example of a critical Internet function that no social network site can duplicate.
In the end, the debate over cloud computing is the kissing cousin of the fight over thin-client vs fat-client systems. In this case, we definitely expect cloud computing's popularity to grow, but we don't see it becoming more relevant than the Internet anytime soon.