Cisco Systems, Inc. (CSCO) Next Generation Networking Innovations Conference Call September 26, 2013 10:30 AM ET
Surya Panditi - Senior Vice President, Service Provider Group
John Choi - Investor Relations
Vijay Bhagavath - Deutsche Bank
Subu Subrahmanyan - The Juda Group
George Notter - Jefferies
Good day, ladies and gentlemen and welcome to the IR Tech Talk on Cisco’s Network Convergence Systems. My name is Celia and I will be your operator for today. At this time, all participants are in listen-only mode. Later, we will conduct a question-and-answer session. (Operator Instructions) Before I turn the call over to your host for today, Mr. Surya Panditi, Senior Vice President of Cisco’s Service Provider Group, there will be a few short words from Mr. John Choi, Investor Relations. Please proceed sir.
Thank you, Celia. So before we get started, I would like to remind the audience that today’s call will pertain strictly to Cisco’s strategy around its network convergence system. No new financial information regarding Cisco’s overall performance is intended or implied and this call should not be viewed as an update to the quarter. We may make forward-looking statements, which are subject to risk and uncertainties outlined in detail in our documents filed with the SEC including our recently filed 10-K and 10-Q. Actual results may differ from statements made today.
With that out of the way, let me now turn it over to Mr. Surya Panditi.
Thank you, John and good morning everyone. Just to set the stage for the technology discussion, I thought it will be good to start with a market discussion. We all know that IP traffic has been growing and service providers have to deal with and contend with and then use their network to monetize the increased interest in cloud, video, and mobile traffic driven in a large part by 4G or LTE. In addition to this, there is a new dimension now with the machine-to-machine events and I’d like to focus a few minutes on what that means for the network, because this really sets the foundation for what we call the Internet of Everything.
So if you look at the Internet of Everything, the difference in the way that the network traffic has been evolving is from just what we have seen so far is anytime, anywhere to anyone and that’s something a concept that we have gotten comfortable with, with the next generation network, which was based on IP, but now we are also adding the anything component, which means people to people, in addition to people to machines and machines to machines, which are a change from the network approach going from a centralized to decentralized to really a distributed or a mesh network. And the machine to machine and the real-time nature of that is driven by events. And events require real-time handling, because that’s when the service provider and the network can actually take meaningful action, which means that you have to have programmatic interfaces, because these devices will be driving events that require real-time action, which will require some kind of a non-human intervention in order to make the real-time nature happen. And what we call this is the evolution from the NGN to the Evolved Programmable Network or EPN. And the three characteristics of Evolved Programmable Network really are it’s not just about bandwidth, it’s also about compute and control in order to be able to handle these events on a real-time basis.
Now, let’s take a step back and say is that sometime far in the future or is that happening now. The fact is that already we see smart offices, we see industry, we see homes, all of these already getting this kind of event-driven capabilities, and we are seeing these go from kind of the trillions to tens of trillions of events and this requires a new scale of event handling. Likewise, we also see new emerging applications that again are possibilities for the service provider to monetize around smart agriculture, where you are using intelligence that comes from the devices, but also gets embedded into the network. And of course we are seeing opportunities in things like smart automobiles as well as in the healthcare industry. And these are emerging markets that offer unique opportunities for our customers and service providers to monetize and the events are going from billions to trillions. So the net result here, the takeaway for me is that there is a significant increase in scale of events in addition to bandwidth, which requires really the Evolved Programmable Network that I mentioned.
So what is the Evolved Programmable Network? It’s a combination of having the network, the usual things we think about feeds and speeds and capacity, those don’t go away those are very important. We are also bringing the element of compute and storage connectivity to the network as well as the control that applications are demanding from the network, which needs to be embedded on a real-time basis. And therefore the network needs flexibility and needs to be obviously high performance, but also needs the elasticity around how it handles functions that are in the network. And what we need then is an open programmable model that the network can present to applications.
When I say open, what I mean is that applications in a manner that is important for that particular application can take advantage of the network’s intelligence and capabilities. And when I say elastic, what I mean is that, the types of functions that are on the network need to be available as and when needed and taken down as and when not needed. So the elasticity is an important function of all the capabilities in the network and of course it needs to be resilient and secure. So again, we are calling this to Evolved Programmable Network and what we see is that service provider plays a central role in monetizing these applications using the Evolved Programmable Network. Irrespective of the access technologies, they sit in the middle and they are in a unique position, because again they have network assets and now they can apply intelligence on top of it.
So I think it’s important to see how the service providers, kind of business models have evolved? Certainly, we are used to the bandwidth that SPs have been offering and that really the focus was on cost reduction and speed. Then we saw quad play capabilities from service providers, where they were able to start offering essentially a mix of different kinds of services. But when you look at these, these are still primarily connectivity driven. With the evolution of the things I talked about in taking advantage of all of the cloud, video, mobile applications they are driving today now they can be – they can package these four applications in new ways to derive new revenues, which gives them an opportunity to get new addressable market, but that also puts a new demand in the network for this event scale that I talked about driven by really these programmatic interfaces and programmable event-driven opportunities.
So how big is this market? Well, if you look at the market around the healthcare, we see that growing in a decade to $30 billion a year. And then on the manufacturing side, we see the opportunities going for – this is opportunities for our customers. So this is the addressable market by the service providers of $43 billion in manufacturing and then at the home $120 billion. So that sets the stage for why the importance of the Evolved Programmable Network and the fact that it is in addition to feeds and speeds also requires other dimensions of event scale.
So with that in mind, we just introduced the Network Convergence Systems and I wanted to take a few minutes to talk about the NCS. We see that, again feeds and speeds are important, so the bandwidth, the NCS is capable of going to over a petabit of bandwidth in a multi-chassis configuration. And even today what we have announced we are shipping 1 terabit per slot, which is an industry leading capacity per slot. And as I mentioned, we need this event scale. So the NCS is capable of supporting trillions of events. We have built into it advanced programmability and intelligence. And what this does is it gives us support for the virtualization, I am going to talk about in a little bit as well. And then the kind of control that I have mentioned a few minutes ago, which is incredibly important. And by adding all of these, we see that you can lower the total cost of ownership by 45% in the cases that we have studied.
So what gives us the capability of doing this kind of scale? There are two fundamental technologies I want to talk about. The first one is silicon. And what we have done in the nPower silicon that we just introduced a couple of weeks ago is we built a massive scalable performance. And even what we are shipping right now is the industry’s only single chip 400 gigabits per second. And we can scale that into multi-terabit scale. And we are leveraging that in both the NCS that we just announced as well as the CRS-X, which we announced recently I think a few months ago. So really industry leading and the reason we are able to do that is because we have unique capabilities on the silicon development side and we have access to the industry’s leading foundries, so we are able to come up with a high performance product.
Now, I mentioned event scale. When you are dealing with a significant number of events to really come back and think about trillions of events and you think about all the high performance in terms of going to gigabits and terabit speeds and multi-terabit speeds scaling up to petabit speeds, you need to be able to do this all in silicon. That’s an important capability that you cannot rely only on the software and I will come to where the software matters, but silicon matters as well. And we have in fact over 300 cores in the nPower silicon that we are shipping. And so we can reach, go from millions of operations to trillions of operations because of our capabilities, which means really 8 times the throughput at a quarter of the power per bit that we have been able to accomplish.
Now, the second foundational technology is virtualization. And what we have done with the NCS is build on top of a hypervisor and we are allowing various types of virtualization and I want to walk through the various, the three kind of basic elements of it. The first one is the ability for our customers to use a virtual version of IOS XR to be able to run and see what the applications look like in without really running it on the hardware itself, but just to make sure that they were able to get rapid service deployment by testing it out in their lab. The second piece is being able to do a virtual control plane, which is allowing them to have the control distributed between the hardware and running it on the datacenter or in the cloud again leveraging the hypervisor.
And what that does is you can start to see where you can bring in virtual assets as needed and physical assets as needed or take them down as you don’t need them, and building on top of that, now they can start running network functions. And the really important part of the network functions is you can actually start to think about how you service chain in order to have multiple services available to a particular user, a consumer, a business user can now say okay, on demand, I am going to provide for example to a business, a distributed amount of service capability, mitigation capability. And by the way because it’s an unpredictable event, I didn’t have that capability on the network infrastructure, but all of a sudden let’s say there is an increase in the DDoS attacks, you can spin up virtual machines on the fly, real-time, programmatically driven and say I am going to bring up these resources, use the scrubbing function in the cloud, so I can still handle it, bring up virtual machines and then once that attack is mitigated to bring them down again in a elastic, real-time, close loop manner. So it’s a very powerful technology that we have built as an underpinning for the NCS system or family.
Now, I mentioned silicon, I mentioned virtualization, so what is that? Let’s now see what that means for the service provider. So today, they are dealing with Ultra HD, for example, or 4K video, cloud, I mentioned the machine to machine and mobility what that requires is it requires the network, which is the Evolved Programmable Network to have certain capabilities. So let me talk about what I mean when I say EPN? I see five things as being important for the Evolved Programmable Network. First is it needs to be open programmable, which means it needs APIs, it needs the ability to have applications talk to it. The second is it needs intelligence, intelligence convergence. And convergence is convergence of IP plus optical. And when I say convergence of IP plus optical, I don’t necessarily mean it has to be in the same physical box, but it has to act as one system. And in fact we saw this being an important capability and we have built our portfolio over time to address it. So when we did the CoreOptics acquisition, which is a few years ago, it was with this in mind, which is the IP plus optical. When we did the Cariden acquisition, it was also with this in mind, because now we are able to build on top of the physical, also the intelligently managed bandwidth using their three and their one capacity planning. By convergence, I also mean convergence of the physical and virtual, which we see as a hybrid approach is what our customers will do, because there will be some functions they want running on the physical infrastructure and some functions that they will want virtualized and some that will do both as I mentioned in the previous example of DDoS mitigation.
And then convergence of fixed and mobile, that’s also an important one because we see as I mentioned a few minutes ago access agnostic but common services that our SP customers are offering. We see the need for automation. Again, automation in real-time are important capabilities of the Evolved Programmable Network, because we cannot afford to have human time scale when dealing the data in motion, it has to be done on a real-time basis. And for that reason, you need automation and you need a closed loop capability. So those are the things that those five capabilities, I mentioned are really important for the Evolved Programmable Network and that’s what allows our customers – service providers to leverage the network they currently have.
So at the very – on this slide, you could see kind of three layers, the very foundational layer is the Evolved Programmable Network, which is all of the technologies they have currently deployed core, edge, access, aggregation and now a fabric that connects all these together and provides this event scale that I mentioned was important in their new services and new revenue monetization opportunities offering open programming interfaces to an orchestration layer, which then supports applications that sit on top of it. And then the orchestration layer, we have been building our portfolio and then the application layer obviously we are going to have some functions that are network functions and some functions that maybe other functions that the service providers themselves may develop or third-parties may develop, which also requires this open programming interface.
The important thing here I wanted to make two very important points. The first is that the Evolved Programmable Network needs to have network intelligence, visible upstream and needs the programmability to also be downstream southbound, so that you can actually have the network behave in a manner that supports the needs of the application. The second thing is that the orchestration layer is an important part of what we are bringing to market, because that’s what allows the service provider to have the kind of control and flexibility they need on the network and to support applications as they can monetize them rather than just have some open uncontrolled access to the network. And that’s an important part of bringing this package together that helps the service providers monetize their investment.
So I mentioned that the Evolved Programmable Network is really, it’s an evolution, not the replacement. So what do I mean by that? If you look at the existing network they have and our products for example that fit into those, into that existing network capability, it is things like the CRS core router which we have been investing in. And in fact as I said a few minutes ago, we rolled out the CRS, we announced the CRS-X and we are using some of the same silicon in the CRS-X that we have introduced with the NCS family. Likewise, we have the ASR9K high capacity edge platform that also acts as a metro core and both of these use the IOS XR operating system. What we have done now with the NCS is provided that fabric I mentioned a few minutes ago. So that brings together the key events scale in addition to the capacity scale that is required for the network to evolve.
And we are now bringing together support for the virtualization, with the virtualization I mentioned a few minutes ago, the need and support for high capacity with the silicon that I talked about a few minutes ago and then the support for control plane that allows you to start taking advantage of these existing assets that are already in the network by adding a dimension of scale and event scale, multi-dimensional scale, so that SPs can bring together their fixed and mobile networks on a common fabric that they can have the monetization opportunities for the Internet of Everything and you can see SPs are already starting to offer that to bring together the physical and virtual. So really bringing that linkage to the datacenter and the cloud and to be able to have a global visibility I mentioned the orchestration capabilities that can be built on top of the EPN so that they can start to offer real-time programmability to the applications. And really what this does is gives them a new revenue and monetization opportunities that are incredibly important as well as the opportunity to optimize the network for capacity and for managing that investment in the network.
So with that, I’d just like to say, I think that what we have pulled together with the NCS is frankly a network fabric that is the foundation for the Evolved Programmable Network. So the EPN really sits as a part of our Cisco ONE Architecture, we have talked about the SP architecture, which is shipping now. And the Cisco ONE is the set of APIs that we have announced in the last few years and is supported on top of IOS XR and the NCS families, really family that we have announced is three systems, the NCS 2000, NCS 4000, and the NCS 6000. Both the NCS 2000 and NCS 6000 are shipping today. And in the announcement we have made a couple of days ago, we had three customers who also announced that they are customers of the NCS 6000. And the NCS 4000 will be shipping in the first half of next calendar year.
So with that, I want to thank you and John, I will hand it back to you.
Great, thank you Surya. Now before we open up the phone lines for Q&A. Since our launch on Tuesday, we have had a number of meetings. And I wanted to be able to address some of the frequently asked questions, but I think maybe helpful for the audience. And after we get through some of those, we will open up the phone lines for Q&A. So first question Surya that we get is we spend a lot of time on the Evolved Programmable Network or EPN, do you think this is a new category within the service provider network?
I see the Evolved Programmable Network as kind of the next NGN if you will, the Next-Generation Network. As an industry, we were very successful in using IP as a convergence layer for lots of different application. So we used to have TDM. We had a whole bunch of dedicated networks, all of them collapsing on top of IP. And I see the Evolved Programmable Network as evolution from the NGN, where now we are saying it needs the functions to be virtualized and being able to run physical and virtual for the network to be programmable, so applications can take advantage of the network and the network can be tied to applications. It’s an incredibly powerful capability that service providers have and they are now in a unique position to both monetize as well as optimize, because they have that Evolved Programmable Network.
Great. And the second question we get on this topic is, how does NCS fit with the current portfolio of CRS and ASR? Does it substitute or complement those products?
So the traditional way of thinking about the network was hierarchical. So you had large core routers and you had somewhat less large edge routers and then access and aggregation routers that were smaller. And frankly the biggest, one of the biggest issues for SPs has been to take complexity out of the network, make it simpler, make it more easily manageable. So we start to think about flattening of the network. And so I think that the days of talking about core edge access kind of in the sense of size of the box. I think it’s not the way to think about it anymore. I see, for example, the ASR 9922 is a high capacity edge system. Why is it an edge system, because an edge and maybe a metro core system, because it has things like multi-service edge functionality, support for broadband those kind of things? It’s not just capacity driven, it sees other functions. Likewise I would see in the core like CRS driven by capacity, while we offer multiple flavors of CRS, both for high-function multi-service capabilities, but in some cases that’s not something that we have and SP wants, so we offer simpler versions of it, too.
And what I see is that, both of these need to continue to increase in speeds and capacity, so that doesn’t go away and in fact we are doing that on both ASR9K as well as CRS. We have already talked about how we are increasing capacity and bandwidth and I see that as an important ability. And NCS, as I see is a fabric that interconnects these, does not replace them and in fact allows for this additional scale. And now if you are an SP, think about if you are rolling out let’s say a home automation system or some kind of a capability that allows you to address a new market or new opportunity, that’s new revenue. But if you now start thinking about what’s that revenue coming from, if you are sitting at home, I am sorry if you have a home automation system and you are sitting somewhere in your office or you are traveling or something else and there is an event in your home that event could trigger video speed to your mobile phone, your smartphone plus some kind of event triggered to the alarm company or police or fire, at the same time it may require some other adjustments to your schedules, all those events that are triggered is a new scale. So when you are the SP, that’s new revenue to you, but you also need to build that new scale in the network, which is what the NCS now provides.
Great. Now the third question here and we have two more before we open up the phone lines. This question is about SDN, as we look at the role of software in SDN what role does software play in your world of routing, can software solve all problems and how important is the hardware and ASICs?
So I see these as not either/or, what I see is that software and the kind of control that you can enable is incredibly important. At the same time, you are not going to do packet forwarding at these kind of capacities going to terabits in software. You need custom-built silicon that’s programmable in order to be able to handle all these functions or actually do something meaningful on the packets as they go into the network. And so I see this as a combination of having this. And I wanted to take one more minute to say SDN and network function virtualization are really – they are two separate things even though they are – they can be used, joined together, but the fact is if I talk, when I think about the discussions I have with my customers, it’s the network function virtualization is incredibly important to them, because they see that as an elastic function capability. What they want to do is to say how do I take that and then apply it on top of the network and that’s where they need the programmatic interfaces, which is what I see as the real power is. APIs, open APIs in conjunction with an SP is I think a really important capability.
Great. So you talked about network function virtualization, there is a lot of talk about it. When do you know which functions to virtualize and which ones can’t be virtualized?
Yes. So there are some functions that today run on appliances. And so in order to make them more elastic, in order to offer the right kind of service chaining and in order for them to be even more easily manageable, those are the ones that we see first being virtualized. And so it’s a scale all the way from there to. If you look at functions that are related to packet forwarding capabilities, those will be done in the hardware. So I really see that and by the way, Cisco already has a number of virtualized functions. So for example, the Nexus 1KV switch we virtualized that, the cloud services rather which was really a derivative of the ASR1K product. RAS, the wide area application services, we virtualized that. The ASA, the security module appliance, we virtualized that. Security gateway, we virtualized that. The wireless LAN controller, mobility services engine, the network analysis module, we virtualized that as well. The ISE, the identity services engine. And so – and we have got a bunch of network management capabilities that are also being virtualized. So we see that, all of these are important to be virtualized. And then as you know John in the last 6 – maybe last year or so we have done several acquisitions in the service provider space. All of them really around the concept of how do you provide the framework of virtualization. So with the Quantum Suite, now we have not only operational efficiencies that Cariden can offer, which is now part of the Cisco Quantum Suite, but also this real-time policy management that brought (inaudible) to us, which is an incredible scalable capability we brought into the Quantum Suite. And so these kind of things I think are really what you should think about is excellent virtualization opportunities not as I said forwarding functions for packet forwarding.
Great. Surya, this is the last question and we will open up the phone lines here is we talked about the early interest on customer adoption, what has been the feedback on NCS so far with the customers you have spoken through globally?
Obviously, we had some customers in the launch who talked about why they see the NCS as an important evolution of their network capability. And frankly, I think the feedback hasn’t been very positive for a couple of reasons. One is this is truly an evolution that allows them to add rather than replace and I think that’s been an important, service providers don’t like to throw away stuff. Once they have got something in the network, it’s important for them to continue to leverage that investment. And so I think they are really pleased with the fact we have come up with something that allows them to continue. If you are a CRS customer, there is a really good roadmap you can add capacity, you can bring in new services and we have also leveraged some of the same silicon and operating systems. So they like that. And this NCS gives them a new dimension. So I think it’s been very positive from that perspective.
Perfect. So Celia, that’s the questions so far that we got. Why don’t you open up to the phone lines, we will take questions from the audience.
Earnings Call Part 2: