All right, so let's take a little bit more of a look at these NFV workloads. First of all, we've talked about the standard high volume servers or the COTS platforms consisting of compute resources, storage resources, and then network hardware resources. And in that element, we're going to dig down into each of these a little bit further. But first, let's take a look at this diagram that we've got here. This comes stylized right out of the [? IT ?] 2012 document that defines network function virtualization. And again, let's not get hung up on whether these are virtual machines or whether these are containers. You'll notice that we don't call it a hypervisor in here, intentionally from that standpoint. But what we really want to see is that those network functions, those interesting applications, can access the resources of those virtual compute platforms that exist on those resources that are there. Interfaces into the back office systems, the BBSes and the OSes. And over on the right side, still, that VIM layer that we talked about very briefly earlier. And then VNF managers, sometimes those would need to interface into an element management system or directly into the VNF themselves, and then finally up into that orchestration layer. So we've got that abstraction now that takes place where we can separate the standard compute resources and that NFVI layer from those interesting applications that are going to be then adapted to those compute resources. And in particular, inside the comms network, some of those functionalities, coming in from the edge, virtualizing the RAN is certainly a possibility at the enterprise level. We would be talking about virtualizing the customer premises equipment. And as we get deeper into the core of the network, some of those standard functions that exist inside there. As we mentioned also earlier, one of the key aspects to be reintroduced after that decomposition from the standard high volume, and the purpose built, and then those interesting workloads is that system integration level. And we see that there really is three levels of system integration that take place. And there is a question of who provides that functionality now. So we used to buy the purpose built solution, where the system integration was provided by that vendor that gave us that complete integrated solution. But now we have either the vendor of the hardware platform that's coming in, or the vendor of that VNF, that virtual network function, or potentially a third party. That third party could be the end consumer themselves, a comm service provider if they have the capacity for doing that, or otherwise a third party system integration house can certainly provide that. But this is the glue then that provides that system level functionality so we've got an operational system that can be managed and monitored from a performance standpoint. So the first thing we've got to do is we've got to make those VNFs work with the infrastructure and the orchestration. Lots of work going on currently in that area. Some efforts like [? OPNFV, ?] another functionality of ONAP are going on in that area. And there are some other independent efforts that are going on to provide those VNFs. So I get a virtualized network function, an application like that virtualized SBC that we spoke of before, but a software. And I want to deploy it on my standard high volume servers. Well, I've got to be able to instantiate it on there. I've got to ensure that if this VNF has a certain amount of resources it requires, that something, my management layer is going to know that and allocate system resources to that. And then creating the instantiation, deploy that hardware, and spin those functions up. And then do the stitching, that SDN stitching, of those functions together. And that's that first level of making that VNF work. Then we've got to align those resources with the legacy network function. So again, these software applications are going to generate events. They're going to generate information. They're going to generate logs. And that information needs to migrate its way off of that platform so that it's useful, back into those OSS and BSS systems. And that interface needs to be consistent for that to happen seamlessly. And then finally, we've got to look at the operations themselves, re-engineering that operation process so that we can exploit those capabilities of rapid deployment and the transformation of that infrastructure. [INTEL MUSIC PLAYING]