[n_are_q]

Eric Schmidt has mentioned several times that google is very fond (focused on he would say) of the "hockey stick" projects that are created by developers and not by the managers on up, and as users discover them, experience exponential growth due to the value they provide. I assume his strategy is to attempt to create an environment in which such projects can first be created, and then be successfully scaled to the masses if they prove popular. So the "use google infrastructure" mantra will be difficult to let go if you are in that mindset.

They need to either remove at least a major part of the overhead and constraints of these highly scalable systems, or create a well planned and organized transition path from something with low deployment and development overhead to something "google scale". This could be at least at first a completely non-technical thing, such as requiring teams to at least have a plan of how they will migrate their redis, solr or whatever data stores to whatever google is using at scale. But reading this makes me cringe, that sounds like a very frustrating environment to work at and sounds very ungoogle-like to me. It appears their public image diverges much farther from reality than I would have thought.