Why Cloud Computing Won’t Lead to Dumb Terminals

It’s easy to fast forward to a time where cloud computing will once again lead to an era of dumb terminals. However, I don’t believe that to be true.

Back when dumb terminals were the norm, the world of computers was quite different than it is now. Most importantly the cost of computer systems was staggeringly higher than it is now and the pace of advancement was much slower.

Moore’s Law is at a point where our individual workstations have processing power equivalent to high end servers of only a few years ago. It only makes sense that horsepower at the workstation provides for a promising future for distributed computing and at some point the ubiquity of grid computing. Leveraging millions, and even billions, of computers as opposed to a much smaller set of monolithic servers is a more likely model.

A server-based model provides for central points of of failure; distributed and grid computing does not.

With cloud computing, the proximity of our data is less important (and likely to be less obvious) but it does not necessarily point to a future where my laptop plugs into the network as a throw back to mainframe days.

I believe it actually points to a time that compares more closely to peer to peer computing, similar to Groove’s replication of data across a network of computers. This eliminates central points of failure, promotes greater power (horsepower and  volume) by leveraging large numbers, and is more conducive to offline computing and synchronization.

We are all guessing at this early stage of cloud computing, on which every server and workstation is merely a node, so I’m interested in your perspective.