So the question becomes for a CIO, what should you optimise for?... I am going to argue, that... the only thing one should optimise for, is for change... Engineering methodology has to change to optimise for these new things: Evolvability, Specialisation, to allow for Experimentation, and this is a place where I think network operators... have to learn that unless they do that, that if they avoid the risk of experimentation, they would be taking the biggest risk you can take, which is to not take risks. But you cannot have that without change isolation.
- Vinod Khosla (Former CEO @ Sun Microsystems) ONS2014 Keynote (https://www.youtube.com/watch?v=q61VkqZRjck)