In terms of future-proofing design, many applications are better off keeping it simple and (at least initially) refusing the temptation to try and solve some pr

In terms of future-proofing design, many applications are better off keeping it simple and (at least initially) refusing the temptation to try and solve some problems “like the big guys do”. To me, this is the most important argument against knee-jerk reliance on memcache, sharding, (ironically) MySQL, NoSQL, or what happens to be the new fashion[1]. Understandably, buckling down and squeezing more out of the existing, tried and true, technology, is neither sexy nor exciting. However, you're no"big guy" yet, and, once you are, you'll kow that a bit of early tedium[2] can pay off. [1] Like with clothing fashion, novelty is periodic, rather than monotonic. [2] e.g. Google's early "cork board" servers could only be sexy in retrospect