Hello All, I have some "meta-Linux" questions regarding platform choice. I am wondering if anyone knows of any research, papers, or other work that has been done on the topic. Given two major types of IT infrastructure: - Heterogeneous, where each application/service uses the platform to which it is best suited. - Homogeneous, where one bends a particular platform to fit every edge case. where the platform can be anything from and OS to a language to an application stack. o Under what general conditions is one infrastructure type more efficient than the other? o What other types of infrastructure are there? o More generally, how does one compare infrastructures in a rigorous manner? - How does one rigorously define "efficient" when comparing concepts like infrastructures? - What other metrics, besides "efficiency" (however defined) may be useful in comparing infrastructures? Does anyone know of any work where this topic has been abstracted and analyzed mathematically? Or of any trade papers where this question has been looked at in practice? So far, I found the following paper via Google: http://uic.edu/cba/cba-depts/ids/IDSNewsFiles/JGITM_Evaristopaper.pdf that seems to be on the topic and some hits on arxiv that appear related but not specifically on the topic. This relates to Linux in that convincing an organization to switch entirely over to Linux seems to be prohibitively hard in practice, but if one can show that the heterogeneous case above can be more efficient, then one can make the case for using Linux where it fits best. Thanks for any info/opinions/personal experiences! -Adam