Small Cell Forum Lays Down the Law
On virtualization: We are seeing virtualization happen now in the core network, where you have large banks of equipment of common hardware platforms and the flexibility to port software around. The natural expansion as you move to a small cell-based network is, you've got quite a lot of processing power sitting at the edge. With the drive of standardization of equipment, that starts to open up a tremendous opportunities to take very adaptive and flexible techniques developed in the field of NFV and extend it to the edge of the network.
What you can do by doing that is leverage all that additional processing power at the edge of the network, and you can move functions to where they are best placed in the network. That could save you latency, transport, could deliver tremendous potential with virtualization and really extend it throughout your network. I'm really excited with the study phase we have going on covering all the different areas of the network, and the desire and appetite I see in the Forum in that area right now. I think it's a very exciting area with tremendous opportunity there. I'm looking forward to working hard with the Forum to make it a reality.
On virtualization's roadmap: I think you could start to see the first exploitation of virtualization in small cells relatively quickly, within a few years. To push it out to every small cell asset will take more time. There's clearly more work on how you control that and need to work on whether there's additional standardization required on that. As an example, if you take SpiderCloud Wireless , for instance, and what they have with their services node, they have in essence a platform there potentially capable of running applications at the edge of the network already. In essence, you could start to exploit NFV in certain areas relatively quickly.
On where virtualization starts: You'll see this transition across all the different product types on having that standardization structure that lets you push certain capabilities to the right part of the network. It could be anywhere -- I don’t think it'll end up starting in urban or rural. When you have a product developed to have the right framework in it, you'll look to exploit those functions right away. You'll see a progressive creep across the entire network. As you add cells capable in dense urban areas, you'll exploit that. The features and functions you move in dense urban or rural could be different. You might end up with the same hardware in dense urban and rural, but the software and functions you run over the top being different; the logical split of where you run those capabilities, it's extremely interesting. There's lot of scope for study and making sure we have all the enablers, and we address the challenges that are there to make sure as an industry we can move forward to that goal as quickly as possible.
On what functions will get virtualized first: It could be as simple as if you take some of the signaling you use for location. As a mobile device moves, it reports measurements to and from networks. Having to pull all those measurements back takes a lot of bandwidth. So if you wanted a location function, you would push that location function to the edge of the network to process all those functions being reported, so you get the best granularity of the data, but without the burden of having to port those features deeper into the network. It depends on what you want to achieve and the capabilities of what you push, and where you push it to.
— Sarah Reedy, Senior Editor, Light Reading