The rollout of 5G will be an extended affair -- according to Heavy Reading's Gabriel Brown -- which is not surprising because there's a world of LTE deployed and a switchover takes a good deal of time.
This is an even more complex undertaking than earlier transitions, such as that from 3G to LTE, because of the integration of the cloud deep into the structure. In the second and final part of a conversation with Telco Transformation, Brown, principal analyst, Heavy Reading , says that using the hybrid approach will set the stage for commercial launches in 2019. In the first installment, Brown spoke about how LTE and 5G would coexist. (See HR's Brown: Longtime Coexistence for LTE & 5G Core Networks .)
Telco Transformation: Where are we in the development arc of 5G?
Gabriel Brown: I think there is a view in the industry of how to get this to commercial operation now. It's reasonably clear. The first step, really, is to have the early drop of the standards in December this year. This is very much a phase one of the standard that works in non-standalone mode. Really, that's going to define the 5G new radio interface, 5GNR, operating in that mode.
If we get the early drop, that gives chip makers and modem makers enough confidence so they can go and create. They will take the prototypes they have today and go create samples in silicon. That would take six months or so. They can test those devices. In another six months, perhaps at the end of 2018, we'll maybe have the first sort of pre-commercial type devices available. I don't think you'll really see any proper commercial launches in 2018, but it does set you up for the first commercial launches in 2019 in quite a few different geographies.
TT: How does the move from LTE to 5G differ from the earlier transition from 3G to LTE?
GB: The new core network, in many ways, uses kind of tried and tested principles. I think what's different in the 5G case is how the industry can look to progress in cloud in other industries and apply those techniques to the 5G core. So, in this sense, 5G core is being developed as a cloud-native architecture, so to speak. It's designed to run on a cloud infrastructure, which hasn't been the case previously.
TT: What is the impact of it being built with the cloud in mind?
GB: In theory, you should start getting the benefits of cloud, which are greater efficiency due to shared resources and a greater flexibility in how quickly it takes to set up a network or set up a different service.
But in practice it's quite difficult because the telecom networking industries have different approaches. For example, consider how redundancy is handled. Failover in telecom networks are in mission critical infrastructure. They're vital to many industries. They are expected to have a very high reliability and, so, there are a lot of engineering processes that have developed up around that. Typically, in a core network, you may have devices and appliances deployed in a one-plus-one redundant configuration, or in a pool of devices, so you can failover from one to another. That works very well, but it can be quite expensive. It works well, but you need more equipment than you necessarily are using.
You can be extremely reliable in the cloud model. Facebook's hardly ever down. Google is hardly ever down and so forth, but there are a whole set of different design principles to how you get to that state. Making that transition I think is going to be very challenging.
TT: Where does network slicing fit into all this?
GB: From an economics standpoint, that's one of the real big transformational opportunities in 5G, if they get it right. Itís making 5G inherent to many different industrial processes, but supporting them on a common infrastructure. If they get it right so it's a really big deal. That's a good example of how you can kind of do steps towards that on a 4G core, but that's an example of where the full 5G core will give you much better capabilities.
TT: Many of the sensor networks will be cloud-based, so it sounds like all of this architecting and planning has to be done together, comprehensively and holistically. This really defines a key difference between the NSA and the SA approaches to this. Is that a fair assessment?
GB: Yes, certainly in the first part, you're right in that you're going to have many services on a shared, infrastructure platform. So how do you ensure a particular service gets the resources and performance it needs, which one is more important than another and so forth. These are just going to be scaling in and scaling out of these different slices.
That is a big, new challenge. Now, you could also apply that in 4G -- it still exists in 4G. You can say well, we want to do this in the cloud as well. As you sort of ramp it up into 5G, the size of the challenge multiplies.
TT: What strikes you as a major building block of getting this right?
GB: 5G core should be a multi-access core network. People want the services they use on fixed networks. They also may use them on mobile. An operator wants a common infrastructure. They want to sort of follow the user services, so they're really looking for multi-access. I think the 5G core network should be multi-access so when you authenticate, it's a common authentication across different access types. I think it's worth emphasizing that that's one of the goals and objectives the industry has for the 5G core.
— Carl Weinschenk, Contributing Writer, Telco Transformation