Latency and bandwidth?
None of these guys, including oddly enough network people, spend much time addressing latency and bandwidth. This is a shame because the reason we have all of these shenanigans is that latency and bandwidth are hopelessly inadequate all around.
By the 21st century I would have hoped that a good portion of the resources we all used would be de-localized and distributed to the point where nobody even cared anymore.
You should never attribute to malice what you can attribute to incompetence. In this case, I am all for incompetence because the issues are pervasive throughout computing. The same issues they have in supercomputer fabrics affect us as well. Only the supercomputer guys seem to care and not even all them it seems at times.
One of the things driving this is, I think, a conflation of levels whereby things that should be exclusively implementation details end up bleeding over into the architecture standards.
Were a reasonable standard in place for communications most of the software written would neither know nor care where things were or what their characteristics were. Drivers should be taking file designations and credentials and figuring out how to contact the next step in the chain.
One thing that may be perverting this is that a secure location service and secure distributed data is not what some people (*cough* NSA) want to see us use.
As long as latency is low enough, bandwidth high enough and security sound, nobody would either know or care what back-end infrastructure looked like except the small percentage of people actively building the stuff.