The Paradox of Cloud Computing and Moore’s Law

A recent article by RW Lucky in IEEE Spectrum, highlighted how some of us still remember the age where we were not allowed to play with those huge mainframes. Big and expensive computers kept behind glass walls and time-shared computing as a dominant paradigm, computation per second was rented for your dumb terminal to consume it.

The advent of inexpensive hardware (accoring with Moore’s Law) changed the picture. Users can have millions of CPU cycles and Terabytes in their homes.  In this setting, why is it useful to give a step backwards start renting computing power again? Cheap hardware means you can have millions of units of inexpensive devices you may want to “share” with others (big data centers are already doing so).

So, very interestingly, what took us all away from leasing computing power is bringing us back again to the same usage model. However, in the old times the end user rented the computing time, while Clouds are not leasing their huge power to dumb terminals, but to corporation’s services.

Tags: none