Author’s Note: There’s a lot of talk and plenty of noise related to NFV in the telecom industry. It’s a time of market transformation with many executives and marketers making claims for how their companies are doing NFV better than others, or how they are ahead of the competition in one or more ways. I think we are missing an important voice: The CTOs. I am talking with my peers from service providers and suppliers to get a sense of what is real. I am sharing these conversations in this series called The Real CTOs of NFV. The title is fun but the intent is serious. Following is my second Real CTO conversation with Doug Suriano of Oracle.
For other entries in the Real CTOs of NFV series, please click here.
Prayson: Doug, tell me a little about Oracle’s role in NFV transformation.
Doug: I think the best place to start is at an industry level. When you look at the goals of NFV it’s really about transforming the network side of the business into a more IT-like scenario. We are striving to adopt the things the IT guys have mastered over the years. Things like virtualization. Things like using a development operations (DevOps) environment that allows for more flexibility and faster time-to-market. We are taking advantage of the IT economics of commercial off-the-shelf (COTS) rather than the hardware-based, inflexible, longer-time-to-market scenarios that we see on the network side.
When you look at convergence of IT and network through this lens, you see that the IT side of the equation has the potential to lead it, especially if the network DNA is there.
Oracle is building on a strong IT heritage, and we acquired network DNA with Acme Packet and Tekelec. There is not only the talk of convergence of IT and network, but also the true melding of one of the world’s best IT providers, focused on software, with some of the best network equipment providers. We are blending those two pieces of DNA together just like the industry is doing right now.
The second part is that we are leading within the standards efforts. To make all of this stuff work, you shouldn’t have proprietary implementations. There needs to be set standards that we all work toward so that we can interoperate and play our roles to make this ecosystem work.
Participating in things like ETSI is a no-brainer from the network side. On the IT side we are using OpenStack in the IT environment and moving that into the network and making it part of our operating system.
Atypical standards on the network side would be things like the tele-management forum (TM Forum). Here, we are putting the focus on the business and OSS side of this NFV transformation. One of the things lagging with NFV is truly tying network transformation to business transformation.
We are starting to get our arms around the operational support systems, especially as you look at orchestration. But then how do you turn it into a model you can actually design from a business perspective — from the monetization side of the business? And, then couple that with all this great technology work in the network to make it a complete package? I think the industry is very immature in that area and it’s one of Oracle’s sweet spots.
Prayson: That’s a great segue to the next question about obstacles to NFV transformation. You mentioned one that I think is big, and that is tying the newly virtualized network elements to the business systems so that you can bill for it. If you can’t bill for it there’s no point in doing it. Is that the biggest obstacle that you are helping CSPs to overcome? What are the other obstacles Oracle is helping to address?
Doug: I would summarize it in two words: business transformation. It’s a network transformation and there has to be a commensurate business transformation, too. It’s not only the management and operation of this new paradigm; it’s also the monetization.
So, how do you monetize it? It involves how you take an order, deliver that order into the network, deliver the device around it, deliver the service, configure the service, monetize the service, and maintain the customer relationship around that service. All of those pieces have to be transformed along with the network transformation for it to work.
Agility doesn’t just come from the network side. It has to come from the business side. That’s the area we are going to be able to contribute to in a big way, and the area that will likely be the most difficult once we clear those early technical hurdles.
Prayson: On the business side there’s another interesting trend and that is operators believe that by moving to these software-based systems they will be able to pay less for these functions or perhaps get them for free. I think that a lot of the traditional equipment manufacturers rightfully see that as a threat. Now, Oracle is coming at it from the IT, data and software side of things so it looks like this could be an opportunity. Has licensing been a struggle or an obstacle for this transformation?
Doug: Prayson, you would think it is. But, what we have been focusing on is to see if we can’t transform our pricing into a subscription-based model. What operators have done in the past is acquire a bunch of hardware, and in some cases the software came along with it. But as the move to COTS started to take off four, five, six years ago, there was a disconnect between the hardware and software. They would buy a COTS server and then load software on top of it. They would pay up-front and depreciate it over time. They would have Op-Ex in the form of software support that they would carry along for three or four years.
What I am seeing now is there is going to be this hardware layer and it’s going to be commodity for sure. The software layer is going to be very dynamic. These 18-, 24- to 36-month cycles of deployment and operations are going to shrink. To pay for that, it’s going to be a subscription model where you pay so much a month rather than having to capitalize it all up front.
Not only are the equipment providers threatened by not being able to monetize hardware like they did before, but also the customers are going to have the expectation that they pay as they go. The big capital purchases and big support fees that have been contributing to the mainstay of the equipment providers’ revenue streams are going to change. You are going to see a subscription model that includes both the service and the support for that service all in one fee.
At Oracle, we are going through that right now as we move our enterprise applications to the cloud. You really see a shift in your business. That is going to be a challenge not only for the service providers but also the customers.
I do think the value will still be in the software. They might say they are not willing to pay for the software, but I think they’ll be willing to monetize what you provide, but it will be about monetizing as a service over time.
Prayson: I would agree. The models you’ve laid out are answering some of the desires of the operators to move away from these fixed chunks to pay-as-you-go, shared risk/shared reward, revenue-based, whatever you want to call it. So, what they are paying is tied somewhat to what they’re earning, which makes their business cases much simpler and much better.
There are a lot of benefits here for NFV and virtualization but it’s just now starting to ramp up. When do you see it becoming mainstream and widely deployed in various parts of the network?
Doug: Certainly in your business, the CPE, NFV is a logical choice. We’re starting to see some of that especially on the SDN side with Carrier Ethernet services.
There is definitely some early adoption going on that side. In the core of the network I think it’s going to be a little bit slower. There are lots of POCs and lots of projects going on.
As an example, I believe AT&T has stated they want to have 75 percent of all their core network elements virtualized by 2020. I think a safe window is three to five years for when we will start to see main adoption coming along.
I think this trough of disillusionment that we’re in right now – I think we’ll agree we’re in it – is going to be prolonged. I think at three years we start coming out of the trough. Five years is when we start really seeing the full-scale type of adoption. And, we are going to be at it for 15 years, Prayson. Just like any transformation.
And the resulting industry dynamic will be interesting. I think you are going to see the blending of the enterprise, the service providers and the OTTs. You are going to see a lot of MVNOs and VNOs standing up. You are going to be serving things like devices that aren’t even attached to humans. Billions of devices are going to be connected to these networks.
We are in a big transformation cycle and it’s three to five years to get to mainstream adoption around virtualization and NFV, but the sky’s the limit after that.
Prayson: There are some things that are expected with this transformation – there are a lot of technology changes and you mentioned the standards. What have you seen that you really didn’t expect or that was a surprise from a technology standpoint, or business standpoint, or organizational standpoint?
Doug: I think you touched on one piece of it. There’s a massive cultural change going on inside the operators and service providers. It reminds me of my dad when new technology comes out and he says, “I’m too old to embrace it but I’m not old enough to ignore it.”
There is also a group that says, “just throw it on a VM and we’ll ship it.” However, the reality is the industry has to refactor most of the software that’s out there. The stuff that’s been running for 20 years is very monolithic and that’s not how you deploy. You really do have to decompose into micro-modules of software because that’s the way you want to deploy it.
You want a mode where you can orchestrate these software modules, control them and scale them exactly how they need to scale, rather than scaling to the least common denominator, to get tons of flexibility. The way you develop, test and deploy is really becoming an IT dev ops model. You can do a feature a month and deploy that feature. You can do it very quickly, easily and painlessly. Those cycles to benefit were taking a year to 18 months, maybe two years, before, and now you get that benefit on a monthly basis.
So, I would say the surprise is the entrenched culture and the way we develop, test and deploy. I think that’s taken longer than anyone imagined, and it’s even internal change that you have to convince your teams that they have to do that because they’re hanging onto “babies” that they have been growing for 10 to 15 years.
Prayson: Will operators push back on feature-a-month type of change?
Doug: There are two pieces. There’s the practical piece — the guys who actually have to operate these networks and get blamed for outages. They have one view. For instance, with our Policy and Charging Rules Function policy server, you should be able to roll out new policies in minutes or hours. Some of our customers take far longer than this to implement a new policy, mostly because they’re leery of doing anything that would harm the network work.
The other camp is driven by the C-shop, the execs, the visionaries and the strategy guys who have to push the organization to realize the benefits of all of this.
You cannot get to a flexible environment unless you adopt these principles. Our development teams are doing it successfully. But it’s the hand-off to the customer — where the customer has to take it, test it, deploy it — is going to be the hard part.
How do they architect and organize these networks to put things into production very quickly? They might not roll out on a wide scale at first, but if they start on a small scale, they can make sure it doesn’t break anything, and then roll it across the network.
You could upgrade a whole network in a manner of minutes. It’s a whole new way of operating. And, really, what we want to get to is zero-touch so that these networks are operating themselves — but that is a ways out.
Prayson: You’ve talked about the need for both suppliers and operators to become agile and use DevOps. Is that what you are talking about with the telco cloud? Will the operators be able to make that transition and what is their threat from new entrants like Google?
Doug: Telco cloud is a parallel path to the whole NFV transformation. With NFV you’re doing it on prem. The operators are transforming their networks and the way they do business within their networks.
There is another deployment model, which is really in the cloud whether it’s a public or private cloud. You actually put this service up there and operate it whether for the network providers, or it might be for a virtual network operator.
There are some services that are starting to be put in the cloud — for example voicemail and web conferencing. These applications are more IT-based. They don’t have the stringent requirements for a five nines type of requirement for real-time communication.
To get to a real-time communications environment in the cloud you have to build-in characteristics that are expected to create fault-proof, very high-quality communications. We require high availability — five nines at least. You can’t take down these cloud communications networks and expect to have the same customer experience you would with the on-prem networks.
You also need to be able to handle stateful transactions. Imagine I’m on a phone call and my network goes down in the cloud. What happens to my call?
In the network right now your call stays up because the state of that call is maintained in the network and if an element fails it will just failover to another part of the network so your call state gets maintained. Another thing we deal with is latency. In the internet, there’s a lot of latency. Much of the traffic on the internet is really “best effort.” You can tolerate that if you are surfing the web or doing a banking transaction or even a FaceTime call with a friend.
However, if you’re expecting real-time communication that connects quickly and stays up all the time for business-critical or communications applications, redundancy has to be built into the cloud if you are going to be offering those real-time services with high reliability.
We are taking all of the stuff we know how to do in the service provider network and putting those principles into the cloud so we can share applications with anyone — whether it’s the OTT, the enterprise or the service provider. This ecosystem is coming together, and I think it will come together in the cloud not in a private network.
Prayson: Would you say that companies like Amazon have telco clouds that meet those requirements?
Doug: I think they aspire to it but I don’t think anybody has yet put together a cloud environment that has the characteristics of a service provider telco network.
Why? Because their latencies are still pretty high. Stateful transactions can’t be maintained in the cloud. They really don’t have a five nines environment.
So in addition to the Oracle public cloud we have taken a more vertical approach. We call it Oracle Cloud for Industries Some industries are going to have specific requirements — like the banking industry with security or the healthcare industry with HIPAA requirements. We are purpose-building those requirements in a cloud-driven way that makes sense for each industry.
Prayson: I think that’s a great approach, Doug. That really focuses in on the point of all this and that is solving a customer problem. We’re not doing NFV and virtualization for itself. We are doing it to provide solutions for end users. I think the Oracle Cloud for Industries is a great example of honing the requirements and implementations for a particular type of business or organization.
Doug: The last piece here is that we believe there is a need for a common communications infrastructure that goes with it in the cloud.
I think the expectation is that everything is always on. This isn’t something that just the service providers are grappling with. If you are a cloud provider, you must figure out how to make your environment five nines. This is where our focus is.
Prayson: Thank you very much for your time today. This has been very helpful.
Doug: It was a pleasure talking with you.
For other entries in the Real CTOs of NFV series, please click here.