Have always thought the industry published a decent NFV white paper in October of 2012 and then failed to follow its own plan. The white paper talked about leveraging the capabilities of the hyper-scale IT cloud platform operators but then the industry decided to reinvent the cloud platform.
The industry lost track of the distinct separation that needs to be maintained between Tenant VNF / Service Orchestration Layer and Platform Resource Management / Orchestration Layer. As a result, Network Function vendors focused on solutions that encompassed both their version of a cloud platform and their version of Software. This was exacerbated by the absence of standards. The resulting VNF software was not truly cloud aware and not portable across even more mature commodity cloud platforms. This becomes a huge problem in a digital era when virtually all content delivery is a multi-cloud, multi-service provider endeavor.
All Network Functions are not the same. They do not impose the same low-latency and networking requirements. The industry could have chosen to start with those network functions most easily re-architected to be cloud aware so as to run optimally as SaaS on a multi-tenant "commodity cloud". The more difficult, low-latency, Control Plane VNFs could have been addressed in a second or third wave. Not too late to make this course correction.
As for Cloud Platforms, the closed loop, AI enhanced, data analytics driven, software defined, resource graph/state management capabilities of a globally distributed cloud platform like Microsoft Azure significantly exceed the vision of ETSI MANO or Open MANO. The TM Forum has attempted to fill gaps with the ZOOM initiative and OSS Future Mode of Operations work but even that has difficulty keeping up with the accelerating advancement of hyper-scale cloud platforms and the latest software architecture and DevOps trends.
The telecom NFV ecosystem focus on commodity x86 hardware and combining it with Open Source software got everyone off track. It was fully engineered and commercially stable Cloud Platforms that were becoming the actual commodity. The hyper-scale cloud developers and operators were far more advanced with VIM and NFVI than the telecom industry realized. They had already abstracted hardware lifecycle management from software lifecycle management to create apparent unlimited scalability. This may have been difficult to recognize in 2012 but it is becoming very clear in 2017.