There are a myriad of issues both large and small for the telecom industry to solve in regard to virtualization and new technologies, but having a clear monetization plan in place is probably tops for vendors.
Adtran's Robert Conger, CTO of Americas, spoke about the need for a clearer commercial plan during last week's Adtran Connect conference, which was held in its Huntsville, Alabama, headquarters.
In June, Adtran was one of four vendors that signed up to work with the Open Networking Foundation (ONF) to help further the ONF's supply chain through develop and deployments of its reference designs. The reference designs that were announced in June were SDN Enabled Broadband Access (SEBA), NFV Fabric, Unified, Programmable and Automated Network (UPAN) and Open Disaggregated Network.
The ONF described the SEBA reference design as a lightweight version of R-CORD (Residential Central Office Re-architected as a Datacenter.)
"SEBA's the one we tend to put most of our focus on because it's related to our core broadband access business," Conger said. "SEBA is actually deployed in the field, but it has got more gaps to close on its roadmap."
In the first part of this FierceTelecom Q&A, which was lighted edited for clarity and length, Conger also talked about network functions virtualization (NFV) and the move to white boxes.
FierceTelecom: Going forward, what does the industry as a whole need to work in regards to virtualization?
Robert Conger: I think the biggest one, which I think operators are all aware of, is just having more of a clear commercial plan. Everybody seems to value the open source and the crowdsource ability, and the more modular, usable components and all that. But ultimately, to really get the vendors to put some more effort into it, there needs to be some kind of financial model that makes sense. I think the savings are obvious to the operators. What some of the vendors get out of it isn't as clear.
Now some of those smaller start-ups are willing to come in and do whatever, but they're not necessarily the guys that can stand behind a network. We do feel like there's been some progress on that front. That was one of the reasons we stepped up to be a (ONF) partner because that was becoming a little more clear. In the first couple of years, to be very honest, there were a lot of people saying, "Let's do all this CORD work," but then there was no clear monetization plan.
Some of that's become a little bit more clear, such as whether they (service providers) start to realize they need solution integration partners to really get these networks rolled out. I think they understand that they'll need commercial software to either augment or accelerate the deployment of some of these systems, and so there's more of an openness to working on the financial model than there was before for all of the parties involved.
FierceTelecom: Is that what Cisco is doing with reoccurring revenue from software subscriptions and licenses?
Conger: Theirs is a little bit different, but that's one way. Theirs is obviously very targeted to hosted Wi-Fi and other things, and it's on a subscription based-service model. That's what they've been doing in SD-WAN with Meraki and the Viptela acquisition. So yeah, that could be one way to do it, where as a solutions integrator you help operators deploy this and in return you back up the SLA and do all these other things.
It could be under that type of model. It could be a recurring subscription model or one-time integration fees or a combination of the two. Just the details of all of that have to get a little more ironed out. I wouldn't say it's 100% complete, but it's getting much closer to where the operators understand that if they want to get this to happen, they have to put some money into it. There will still be a huge benefit to them overall, but they've got to put some money into it.
FierceTelecom: While there are obvious benefits to NFV, operators have learned that it's also complex. What advice do you have for somebody that's maybe taking a first stab at it and hasn't been happy with the results so far?
Conger: What you see now is a lot of the projects, even in the ONF, have probably pulled back on some of the data plane VNFs in the first phase. Where NFV has struggled is data plane performance, software versus the previous embedded hardware, or embedded software on hardware. So just the cost points to do equivalent functions, and that still depends on what the function is.
Size is the main thing. Virtualize where it actually makes sense. Save the data plane virtualization for the really complex functions where you can't easily do that in hardware. But even that list is growing shorter because there are advancements in programmable switches, like the P4 programmable switches, where now a lot of the things you couldn't do in hardware you can actually even do in hardware now. So, I think you'll see maybe even more and more of a shift to flexible, programmable hardware as a replacement even for some of those applications that were heavily targeted with a VNF.
If you can go buy a white box switch with really flexible programming, you can write different prefilled programs for security DPI-type applications that may become more popular than putting it on a server. A traditional server is expensive, and power-heavy and it's limited in scale, whereas programmable silicon can give you a lot more capabilities in terms of watts per terabit, and all of that. You still retain a lot of flexibility."