Jon Udell: Scale-free networks and mirror worlds
The inimitable Jon Udell writes about networks - about Javaspaces and Tuplespaces, loose coupling and grid computing - and I can’t help thinking that these uber trends which would seem to be profound leaps and bounds in the punctuated evolution of computing are often to do with very banal issues, often involving the relative strengthening and weakening of various points in the information processing chain that only exist because of cost.
Is there a trend towards clustered cheap servers because the weak link between disks and RAM (that justified expensive Sun hardware with a fancy BUS) is no longer needed if you keep things in cheap memory? Is it in turn, the availability of these swarms of cheap boxes that is really driving the possibilities of grid computing? What would happen if the on board memory on a chip were as cheap as a hard disk?
