I want to start by returning to this notion of Network Function Virtualization and Network Based Services. I need to start by defining these terms and helping you understand why this is important to you as a consumer. Network Function Virtualization is a really important trend in Communications. Today, many IT functions are virtualized. This means that they don't have a dedicated computer that a service runs on. There is software (VMware being the largest supplier) that allows each computer to act like multiple computers and the software shares those resources. There are three important aspects here. First, this means that you don't have to have a computer dedicated to a function such as a web server. Second, it means that you can create new function without adding more computers. Finally, it means that you can turn off functions that are no longer needed.
In practice, this means that a company like Amazon does not have to dedicate specific computers to specific tasks. This allows the company to change the configuration of the services that they offer dynamically. From an efficiency standpoint, this has been a great boon to the IT world. I can't imagine an IT group today that doesn't use virtualization for something.
The Internet Service Providers are headed this way as well. There are types of equipment that they use that are not virtualized and use custom hardware. There is a push to reduce or eliminate all this custom hardware and implement Communications Services as virtualized services. I don't think that the top end of the performance curve will be done this way, but much of the more standard class equipment will be. There should be an impact on the cost to build and operate networks because of this.
But our friends the major ISPs really want to get more revenue out of their network functions. So, they are claiming the possible need to embed services in the network in order to allow applications to operate correctly. I see examples of this at some level in Content Delivery Networks (CDN), but they are actually just cheaper ways for providers to deliver bandwidth. They don't impact the application except for cost. They invent this notion that we should allow them to build services into the network that can be instantiated with NFV in an on-demand basis. I believe this is a terrible idea for both application developer, consumer, and ISP.
For the application developer, they then end up being dependent on the implementation within specific providers that allow this to happen for the specific set of functions that they need. This means that application developers are building a premium version based on this service. I can't imagine what must be in the network, because the function would be replaced at the application layer without the network service. The application is going to be successful without this or not at all. Which is why this is a bad idea for the consumer. If there needs to be a specific function available in the network, then if they travel they might lose the use of the application that they depend on.
Finally for the service provider, they would have to invest a lot of money to make this happen. There are challenges with setting up, tearing down, billing and maintenance for these systems. Unless there is a huge interest in it, then they will lose their money. On top of that, there will be Service Level Agreements (SLA) involved. It will be difficult to make a service that has an SLA that brings value to the application makers and the ISP at the same time.
So, this is just a bad idea. Its bad for everyone involved. Let's just locate the intelligence where it belongs, in the applications. Jim Sackman Focal Point Business Coaching Business Coaching, Executive Training, Sales Training, Marketing
Change Your Business - Change Your Life!