It's been a while, but then again I haven't really had the time or inclination to sit and contemplate much over the past month or so, and the holidays are a good time to slow down and chat with people. So, that being said, I'd like to talk about some conversations that have emerged about technology over the past few days of visiting with family and friends I haven't seen in a while. This is as much my sounding board for my own ideas and long term storage, so bear with.
This whole post is spawned by an article I saw which is basically a retrospective of the technological changes we saw in electronics in 2007, and predictions for the future.
Now, some of what I got for christmas was a 1GB flash drive smaller than the tip of your pinky finger (the storage is possibly smaller than the USB plug) a 16 GB flash drive, and a GRPS card for the laptop (which is what I'm using right now to post from a house that will probably never get a wireless or broadband internet connection.)
Now, I was talking with a network engineer about Moore's Law, which states, roughly, the power of computers per unit cost - or more colloquially, "bang per buck" - doubles every 24 months. Yet we're about to hit a physical problem with this, as microprocessors have been engineered so tiny that there is no more capability for them to become more efficient. This means either adding additional processors in tandem (i.e. dual- or quad-cores) or completely throwing integrated circuit technology out the window and adopting one of the other possible solutions being developed for number crunching. Either solution will come with problems for the home (or even worse, the mobile) user, and so the question develops, why do you need the vast processing power of these advanced systems for local processing use. (Supercooled... which means 4 degree kelvin... processors will probably reach 1 TeraHerz by 2009... half a million times faster than your home PC) Getting back to my earlier train of thought, a computer needs Input/output, storage, Memory, processing, communication (bus) and power. Power is getting better, memory is smaller than ever, storage is absurd (see previous comment about 16GB stick) and I/O is limited by the human body's mechanics. But I think we will soon see a revolution in communication between these elements, as wireless systems grow faster and more robust... and processing. Of course a local processor will always be neccessary, but the more and more I see online super-processors (or process clusters) that time will be farmed out on. As bandwidth becomes cheaper and easier, profit will be gained by super-fast processing time.
If you think this absurd, consider the fact that in 2007, you could rent out server space (processors) to put up "virtual servers" to allow emergency redundant virtual machines to back up your physical network hardware. Yes, silly, but when you can replace half a million dollars in hardware (which will age, have power requirements, admin, etc.) with Ten thousand dollars a month of rented processing and storage... the choice is relatively simple.
Thus the prediction that, as bandwidth increases, the shape of the typical end-user computer (especially mobile computers) will change vastly. A simple I/O interface with minimal required onboard processing, memory and storage that is linked through a robust network to a fully online, (infinitely expandable depending on price) virtual machine.
Problems with this? Security, obviously. Recurring cost (as opposed to one-time cost for hardware.) Problems with the reliability and trustworthiness of your "virtual computer" housing. Need a very, very good wireless network.
Requires more thought.
Wednesday, December 26 2007, 04:31 AM