In this respect, it would seem that Google's "horizontal" infrastructure stack makes it more adept at handling web-scale problems across multiple domains (versus Microsoft's "vertical" problem-by-problem approach).
In this respect, it would seem that Google's "horizontal" infrastructure stack makes it more adept at handling web-scale problems across multiple domains (versus Microsoft's "vertical" problem-by-problem approach).
Some queries that would be relatively easy (and not super-scalable) have to be reformulated so computation is done on write and not on read. An example is writing an app that uses ranking and decay: http://code.google.com/appengine/articles/overheard.html
There's a different set of tricks to learn, so in the context of cloud computing, if you aren't willing to confine your approaches to those amenable to truly distributed cloud computing, you're not addressing the basic problems. Of course, there will be apps that can work effectively without web-scale, but are you talking about the "cloud" at that point or just hosted apps?