With all due respect, I think you need to be a little more open-minded and dispassionate about this. If there's "no way in hell" you'll change, then so be it, but you may be overlooking opportunities by being unwilling to fully consider alternatives. Try to avoid letting this be a religious matter. At this point nobody knows for certain how this will all progress.
If you're a one-person shop, or a mobile user, you'll probably continue pretty much just as you are for some time to come, with a Java-enabled browser on a PC; word processing will be done locally (on your PC) with local software (either legacy-style -- e.g. Office xx -- or perhaps a new Java word-processor). However, see comment below regarding mobile computing.
But suppose you're a business with several employees. You might opt for a small server and a series of NC's (e.g. JavaStations). In terms of the amount of iron on the floor/desk this is not much different from what many small businesses do now (e.g. a group of PC's and a file server connected together with a Lan). The main technical difference is that persistent storage (hard disk) is formally centralized; local temporary storage for most tasks will be in RAM instead of on disk (I believe I've seen discussion of the possibility of adding cache disks to future NC's for truly large tasks). RAM is cheap enough now that you can load up an application once from the server and thereafter processing proceeds pretty much as in the PC/Server model (except now we're almost certainly talking Java). Many shops already require their employees to store work-product on the server so that it can be shared and so that backup can be centralized; thus, except for the initial program load (IPL) there isn't much difference from current practice in terms of load on the network; future JavaStations will probably include flash memory so the IPL will only need to be done when software updates are made (flash memory periodically resynchronized with servers may also solve the mobile computing problem). Note that the Java model may help cut down on "bloatware," and also minimize the bandwidth required for software downloads, by allowing more selective loading of features; I believe most people actually use only a small fraction of MS Word's (12 megabyte?) code.
Think "intranet" rather than "Internet."
Obviously checkpointing and similar techniques (e.g. periodic saves) can be used to mitigate problems with network failures, just as they are now. I suspect the severity of problems caused by network failures will be similar to current experience. Overall system reliability should improve because of the elimination of mechanical components on the desktop, and because of better controls over software (e.g. avoidance of accidental kills caused by installation of identically named but incompatible DLL's in user-provided SW).
The economics I've seen suggest the Server/JavaStation combination will actually be cheaper than the current Server/PC model, both in initial cost and, more importantly, maintenance and support. I believe Sun now has some pretty inexpensive servers. Software charge models may or may not change, but that's not unique to NC's. Security should be improved over the current situation.
These are just some thoughts offered in good faith to (hopefully) help you think about the possibilities presented, and are Just My Humble Opinion. Not intended as investment advice. |
