One should not make tradeoffs with the major benefits expected from
distribution. This would force to reevaluate the basic reasons why
distribution is wanted in the first place.
As an example, one could consider to sacrifice scalability. Doing
so, one should however go back reconsidering the reason why strategies
like "buy a bigger machine" have been (implicitly?) side-stepped
priorily.
Citation from Robert Martin, himself refering to Scientific
American:
However, the software crisis is more than just this. The
real crux of the software crisis is that it appears that the
probability of failure increases geometrically, perhaps
asymptotically, with complexity. This is scary. It implies that there
is an upper limit to the complexity that we can control with software.
That beyond this complexity limit, no project can succeed.
This relationship was nicely documented in one of the 1995 issues of Scientific American.