Cisco Systems, Inc. (CSCO)
June 18, 2013 11:30 a.m. ET
John Choi - Investor Relations
Kelly Ahuja - Senior Vice President and General Manger, Mobility Business Group
Brian Modoff - Deutsche Bank
Vijay Bhagwat - Deutsche Bank
Kip Clifton - Deutsche Bank
Grady Burkett - Morningstar
Hello and welcome to today's Deutsche Bank Tech Talk. My name is Colin and I will be your web event specialist today. At the end of today's presentation we will have a question-and-answer session. You can submit questions at any time during the event. (Operator Instructions) It is now my pleasure to turn the webcast over to Mr. Brian Modoff with Deutsche Bank. Mr. Modoff, the floor is yours.
Brian Modoff – Deutsche Bank
Thank you, and welcome everyone to our monthly Tech Talk. This month we are pleased to have Cisco Systems with us. We have got Kelly Ahuja, who is the Senior Vice President and GM of Mobility Business at Cisco, as well as John Choi of Investor Relations. Kelly is going to talk about network function virtualization or NfV and how it applies to mobility, as well as to discuss all the assets the company has acquired over the years starting with Starent and Intucell, Ubiquisys and Meraki, how all these wireless assets come together in Cisco and how Cisco monetizes these into the future.
So I will turn it over to John who will have some brief words and he will pass it on to Kelly, and then after that we will have some Q&A.
Thanks, Brian. So before we get started, I would like to remind the audience that today's call we pertain strictly to Cisco's strategy for network function virtualization and mobility. 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 risks and uncertainties outlined in detail in our documents filed with the SEC, including our recently filed 10-Q and 10-K. Actual results may differ from statements today.
Now with that out of the way, let me turn it over to Kelly Ahuja.
Great. Thanks, John. Welcome everyone and glad to be here and hopefully I can walk you through a few things, but my key takeaways for you all on this call today are going to be the following; which is, the first, I want to educate you on what network function virtualization is and how it's different and it applies to other areas such as mobility. Second thing is, I want you to walk away with a sense of confidence that this is not something that’s going to be a threat but it's actually more of an opportunity for us, for Cisco, and an incremental opportunity at that. And it is something that we are leading the way on. So, I want to make sure that I convey those few messages through the content that I have prepared and also through the Q&A.
So without any further adieu let's get started. You know as we get started the first thing I usually like to talk about is, what problem are we solving. Because many times we always talk about technology but without having a context to what the customer's problem is, it's really difficult to kind of figure out what's exactly we are trying to do. So if you take a look at it from the customer's lens, whether it's an enterprise or an SP, they really talk to us about three things. First of all their challenges are the following; which is, one, the total cost of ownership. They want to try and figure out how to reduce the overall cost of operating the systems and services and infrastructure that they have got.
The second things is that they feel that the velocity and agility has to be much higher and those slow down because the level of complexity that they have in the systems today and the network and the infrastructure that they have. And the last part is, they want to be able to reduce the risk because as they move faster, they want to move faster, but as they move faster they want to make sure that they are reducing and managing the risk that they have in terms of doing so. So you hear lot so buzz words, there is lots of industry trends that talk about several things, virtualization is one of them. SDN, APIs, Cloud, programmability, and it's a smorgasbord of acronyms and terminology that you can come up with.
And the purpose of today's talk is really to focus on network function virtualization or NfV. So let's walk through the next stage, which is, let me help explain what NfV is and how it's different or similar to SDN. So first of all, what is NfV or network function virtualization. Well, NfV is an initiative that was launched recently in October 2012, and it was really focused to solving the problem of dedicated hardware and systems that are deployed to act upon several functions inside the network. And the goal for the initiative is to take all of those dedicated hardware appliances and virtualize them on to a general purpose compute platform.
So what are some of these functions that we talk about? Well, the functions include things such as application load balance or session border controllers, network address translation, deep packet inspection capabilities, firewalls, and even some routers and more. The initiative was launched as the SDN & OpenFlow World Congress in Darmstadt in October of 2012. And essentially what it's become is an industry specification group within (inaudible), and it's about 1.5 year to 2 year effort to basically come up with it. So what can we expect out of this? Well, they are not really working on defining specifications, but what they are working on is delivering white papers and working with other standard SDOs to ensure that the standards development organizations are all aligning around us and addressing the problem.
So, again, what is an NfV? It's taking all of hardware based appliances and function and putting them on to a virtualized or compute platform. And that’s a very important distinction because it's taking that separation of hardware and software and differentiating the software and hardware. So now let's go to the next part which is how is it similar or different to SDN?
So if you contrast NfV to SDN, SDN was really initiated in academia with initial focus on data center. And the focus was really separation of data plane and control plane. Now remember what I said about NfV, it's about separating software from hardware and SDN is separating data plane from control plane. And essentially what SDN focuses on is the concepts of controllers, network views, APIs and kind of the -- keep making those programmability of network elements possible. So programmability becomes a key part of SDN. So is NfV same as SDN or is it NfV and SDN?
Well, if you kind of look through the details, network function virtualization is not technically related to classic SDN. But it could use SDN technology. For example, virtualization is part of the problem, of course, part of the solution I should say, and the second part is other elements that SDN brings in. Such as APIs, the controllers, the orchestrators, all become important and can be combined with NfV to be able to solve the customer's business problem. Now what does that mean? Well, for certain functions where the regional physical requirements SDN can provide the connectivity fabric that’s required as well. So those are the kind of pieces that we should be thinking about when we talk about SDN or capabilities like that.
Now let's talk specifically about the next part, which is, what is what is Cisco's point of view and what is our approach on network function virtualization. Now first thing you should know is virtualization is great, right. We are leading the way on virtualization. I will show you some of the things that we have already virtualized and we have a very strong road map on where we want to go. But by itself it's not alone to address the customer's problem and we have to do more to address the customer's problem. So the first part is, as we talk about NfV or virtualization, is to decide what to virtualize and what to not virtualize. And you got to have almost a decision criteria on ensuring what is that can benefit from virtualization.
So what's a simple kind of a thing that I can point to in terms of how you should think about what should get virtualized and what shouldn’t. Well, first of all anything you can use general purposely to you for control, for signaling, and any high touch but low bandwidth application where cost, flexibility and fast development are required. So many of the control plane capabilities, signaling capabilities and low touch bandwidth applications, absolutely virtualization is a good thing for them. Now contrast that to other areas where you have got to use customer ASICs, NPUs, a lot specialized silicon for implementing high bandwidth, high performance and processing, where the algorithms and well known in the industry. So those are both important. And as we think through that, those distinctions, then you have got to figure out where virtualization makes sense and what to virtualize and what not to virtualize.
Now as I have said before, we have already been leading the way in terms of virtualizing many capabilities. And there is a sample list here for all of you in terms of what are the things that we have already virtualized. Everything from Nexus 1000V to cloud services router, to security appliances, to virtual wireless land controllers, mobility services engine, identity services engine, prime portfolio, as well as our Quantum portfolio, are essentially all virtualized elements today that you can purchase and deploy.
So a lot of things we have virtualized already. And this is a common theme in terms of how thing evolve. So first of all, as networks and systems have evolved, you always have gone from purpose built to things that use common hardware to translating them to virtualized environments with hypervisors to almost becoming clouds are elastic clouds where you need to bring these functions together and be able to expand and contract them as needed. And that’s a big part of the challenge which I don’t think virtualization by itself addresses and that needs to be addressed.
So let's talk a bit more about what that is and why that’s needed and so on. So first all, as I said before, virtualization alone is not enough, right. The goal for the customers, the business problem is, how do I make things faster? How do I reduce my cost, how do I reduce my complexity so I can move fast, and how do I reduce my risk. So if you think about the way they are solving those problems are they way they are asking to solve those problems. It's one they are saying, well, I want networks more programmable, right. I want to be able to program an abstract out of the network so I can do things much faster.
Second things is, I want to be able to automate as many functions as possible and I want to make the network self healing and adapt to different changing conditions. The third is, I want to make the network more elastic. Why? Because as many of these functions become deployable in virtualized environment, I need to be able to grow or expand or contract the capacity, because that’s how I am going to get the efficiencies out of it to be able to do that very effectively and very dynamically. And lastly, I need to make the network very resilient. I need to make sure that I have enough capacity in there but also enough security in there to be able to make sure that I am addressing the needs that I have for the business.
So what does that translate to? Well, if you think about all the things that we have been talking about from Cisco, things like open APIs. Programmability is a key part of being able to open up, expose the network and then program the network very effectively. Automation, all of the dynamic orchestration capabilities that are required not just overall in the network but different parts of the network and how do you do that. Virtualization providing the elasticity and also the cost efficiencies and of course resiliency through security but also high availability architectures. And those are functional foundation building blocks for what the customers, where it's an SP or an enterprise, need across a complete portfolio to build up the network architecture and also to ensure that they are actually solving their business problems.
Now the other part is, why such problems need to be addressed? I think I talked about this already but just to repeat that, because I think this is a very important construct. First is elasticity. Elasticity is required to be able to expand and contract the number of instances of an individual function or perhaps a service, right. Because now when you virtualize something, you have got to put into a compute environment, you need to turn out many instances of it. Whether it's a firewall instance or a session border controller instance or something else, to be able to scale or contract the capacity as and when you need it.
Second thing is, is you got to automate things. You have got to automate the infrastructure, you've got to orchestrate the infrastructure, both inside the data center where these virtual functions are going to reside but also not just in the data center, you've got to automate and orchestrate the entire network as well. The wide area network, the radio access network etcetera. And, of course, we have got to make things programmable. Why? Because to be able to move faster, to make things simpler, we have got to be able to turn on programmability but do it based on a policy and intelligence that can be extracted from the network and applied back into the environment.
So a good news is that both NfV and SDN are already part of the Cisco next generation architecture which we refer to as Cisco ONE, or the open network environment framework. And if you take a look at that, we have a set of physical infrastructure assets, we have a virtual infrastructure upon which a lot of these network services and network functions could reside. And then above that, the applications. And all of these are things that are currently, as we have talked about, heavily being worked inside the company, with the customers, as well as in standards environment through many, many initiatives. So as you hear about OpenDaylight or onePK or all of these other things, they are all part of this overall framework and they tie together to solve the customer's business problems. And they are open, right. The key thing here is that they are open.
Now with that in mind, let's turn it to maybe some specific examples. So where does NfV apply? Now there are many use cases for NfV. We can apply them to video or Videoscape. We can apply them to business service or VPN. We can even apply them to mobility. Given I run the mobility business, Brian asked me to make sure that I can apply this NfV kind of examples to how we are actually solving it for mobility. So let's maybe drill down a bit more on mobility and share with you what we are doing there.
Now if you take a look at the mobility architecture, for the lack of a better word, this mobility stack. This high level mobility stack has different components or different layers in it. There is a radio access network and there, there is of course small cells and macro radio. There is a wide area network which is of course the backhaul but also the backbone area. There is a core network where typically you have the packet core but a lot of the other core functions which deal with the per session or per user or per subscriber level basis. And above that are many of things that the operators may have inside their data centers to be able to apply optimization capabilities or even monetization capability.
Now if I take this architecture and turn it 90 degrees and simplify it, a typical user session whether it's a smartphone or a tablet, would require you to cut across all of these environments inside the network. Now keep in mind that I can change this from a service provide mobility environment to even a enterprise environment where the radio access network is essentially the essentially the Wi-Fi infrastructure that the enterprise may have. The wide area network could be their branch office environment and the core network could be where they actually maintain the information about the sessions and users going into the data center and services. And then that’s where you go to the internet.
Now where does NfV apply in here? Well, let's first talk about how the current network is stitched together. Now I am going to take the privilege of shrinking this down a bit in terms of picture to be able to show some examples, but also tie in to this for you, as Brian said, where some of the assets that we have been acquiring and also building, fit in inside this architecture. So first of all, let's talk about the architecture itself. Now virtualization may apply in any one of these environments. As I talked about before, virtualization just isn’t limited to one area. But let's talk about first in the radio access network for example. You may have physical assets but you may also have some virtual assets.
Now typically in the RAN you need cells, you need access points, you need Wi-Fi or licensed or converged access points. And that’s where our acquisition of Ubiquisys allows us to have licensed small cell capability which we can combine with our own internal assets of licensed or Wi-Fi based small cells. Now if you look above the top of each of the layer, I have actually highlighted this triangle. That triangle is absolutely critical in each one of these layers. Why? Because that’s the area where you can actually extract the information from the network. You can actually make that analysis of what's going on within that domain. The radio domain, for example, or the wide area domain or the user domains. And then based on a policy apply something back into the network. Program the network back.
So using a construct of begin apply that back, so that’s closed loop. So there ThinkSmart was a company we acquired to allow us to do that closed loop in the Wi-Fi world, Intucell allows us to do that in the macro radio and also small cell heterogeneous environments using SON or self-optimizing network capability. In the wide area network, we acquired a company called Cariden which allows us to do the same closed loop at a path level, optical or IP or MPLS level in that layer. And then in the core layer, this is where the acquisition of Starent a few years ago picked this off for us our focus in mobility, but now the acquisition of BroadHop allows us to get that high performance scalability in policy to be able to apply that back into the network.
Now if I take that whole infrastructure and close in the loop on that infrastructure between the RAN, WAN and core, let's go back to the user session and talk about how user sessions are set up. So a particular session may have to traverse a radio area network, a wide area network, and go into the core. Inside the core, the last element that you hit is the GGSN or a PGW, which are essentially gateways for 3G for 4G environments. Behind that are different APNs or different networks, virtual networks that are established. And each one of these networks is established for a particular session type or a particular user. So for example, a smartphone user going to the internet may have to go through a web proxy, a deep packet inspection, a firewall, or an address translation.
A video user may go through a video optimization engine and perhaps a firewall for security. An IMS user may go through a session border controller and enterprise user may just connect directly into the environment. Now most of these assets are physical appliances and in many of those cases, sometimes they are Cisco in many cases they are not. Because you may see load balancers in there because one appliance isn’t enough, you got to scale multiple appliances and that’s where load balancers come in. Now what's the problem?
Well, the physical appliances that operators deploy are difficult and complex, because you got to design the -- maintain the capacities of the systems. You got to think about redundancy, load balancing and also there is lot of, sometimes incompatibility in how they stitch together or connect together. And it creates a lot of complexity and operational headaches for the operators. And even for the enterprise example, the same thing would apply but just in a different use case. Now the other thing is, I am an operator and I need to reconfigure, right. I have got some firewalls dedicated for my internet users and I want to apply them to the video side, I have actually got to go do physical work. And that work is difficult, it's very cumbersome and not very straight forward and carrying forward in that.
Now the last but the most important part is, all of these service chains, and each one of these you can talk about as a service chain, they are all hardwired to the APN which is that virtual connection. And the reconfiguration is very complex and that requires a lot of manual operation. So remember back of the business problem which is what we are solving for, is speed and reducing the risk, right. This is a very high risk and very complex environment and makes things a lot slower and difficult for the operators to move fast on.
So let's talk about how NfV or network function virtualization solves the problem. Now if you take a look at that data center environment that was sitting back there, now that’s a bunch of compute resources, of course Cisco UCSs that we have deployed in the infrastructure. And each one of those physical appliances that we had before is now a virtual service. That virtual service or virtual function has replaced that physical appliance. So now, and of course looking at the core network, instead of having multiple APNs we have simplified that by consolidating all the APNs and being able to do this in a dynamic way on a per session or a per user basis as the operator needs it. Now this is an example of how virtualization can solve an operator's problem in the mobility space. Now as I said before, there are many other examples that we have talked about or can talk about, and let me mentions some of them here.
The first one is that of a femto, for example. If you think about small cells and a lot of the small cells that are deployed today, many of those functions that are deployed in the core network of the femto or actually virtualized and stitched together just the same way. Another one could be our recent acquisition of Meraki. Meraki was an acquisition that we did primarily for getting into a new business for small and medium kind of area, mid-sized market. But if you take a look at what they do, they have access points but they also have a lot of the functionality in the cloud. And the key value they bring together is this virtualized capability that’s stitched together to be able to tie these service chains much like this one is. So another example of what we have done in this case. And the same thing can be applied to, as I said before, video or enterprise or VPN capabilities.
So now let's get to the summary and I would love to get into some questions and answers. But let me just summarize quickly by saying, one, NfV or network function virtualization is really about virtualization of functions and services in the infrastructure. NfV an SDN are not the same and they can co-exist inside the environment. The third part which is the most important which is virtualization alone is not enough. You need programmability, you need automation, you need elasticity and you need resiliency. And virtualization addresses some of the problems but not all of the problems that the customers have.
The next part is, as you saw before, we are leading the way in the industry by virtualizing many of the functions today and many of them are available today for our customers to use. And both NfV and SDN are part of Cisco's next generation internet architecture or Cisco ONE. So with that I want to thank you Brian and thanks to the team for listening, and I would be happy to take questions.
Brian Modoff – Deutsche Bank
Thank you, Kelly. And now we will jump into Q&A and please have your questions ready so that we can get to those as well. So, Kelly, sticking with the NfV focus, who are the customers that were deploying NfV? What is the value proposition? Is it a TCO or is it a monetization play? You can maybe tie in some service chaining on the monetization side?
So first of all, NfV is a concept or construct that you can apply to SPs or enterprises, either one, right. And it could in the SP world, it could be mobile operator, it could be converged operators, video folks, cloud folks, or anyone. I think it applies probably to anyone and everyone. In terms of applications, it's really going to be dependent upon the specific customer and their pain points. And the pain points that I used today for mobility is the key pain point that we hear about from the operators in my business, right, mobile operators. So much of the way the pain points could be different, depending upon which customer you speak to.
In terms of monetization versus optimization, you know as much as there is great focus on monetization, the first piece that we have to address is the ability to move fast and be able to have programmability and be able to have elasticity. And optimization is part of that, right. So I think unless you do that, unless you get the infrastructure to be able to turn on new services or new intelligence based monetization capabilities, you are not going to be able to do that. So I think they are tied together. And in most cases what we hear from customers is a different thing. In some cases, customers what to lead into this virtualization model with new services. In other cases they actually want to do go down and optimize the current services and then talk about monetization. So we are seeing both of those in terms of our engagement with customers.
Brian Modoff – Deutsche Bank
And you know NfV enables, and you just aggregated a deployment model such as cloud-based architectures for mobile packet core. Will Cisco deploy a virtualized ASR5K? What are your customers' points of views for deploying a virtualized ASR5K? And then I want to talk about competition after that?
Yeah, so first of all, we have been leading the market in the deployment of packet cores, right. As a result, we get to engage with many operators around the world. And the one thing that they tell us is, look, you packet core products are great and we need to be able to use them for many of the services that we have to deploy today. Because most of the services that are deployed today are really mobile broadband, right. Mobile internet infrastructure or perhaps other capabilities. And for that service, the call models that are required, really require function of bandwidth, compute and memory. All combined together to be able to solve the problem.
Now there are other use cases in mobility, and as we talk about the internet of everything. As we talk about machine to machine. There are new use cases that are coming up which are much more low bandwidth, if you go back to that characteristic that I had. Low bandwidth, high control plane, all those kinds of things. Those are the kind of the examples where operators are talking about this. Now clearly, what that becomes for us is an incremental opportunity as I have said before, right. And certainly we are working with many operators and many folks around that.
Brian Modoff – Deutsche Bank
It sounds like that actually is a key point, because one of the things that people will start to get confused on is they think, well, NfV, okay, we virtualize everything. I don’t need the high performance boxes anymore. I just need to virtualize on standard compute and away I go. And you are saying that there are specific elements that you apply NfV to, there is still other elements that you need the high speed - you need the CSR-X and you need these kinds of high speed, high capacity box.
That’s it, for quite some time. And then in the same way you can talk about core routers, for example, right. You are not going to apply NfV to a core router. However, if it's a low bandwidth or a small router like a CSR or cloud services router, you can virtualize that function and put that inside and put that inside a [data centric environment].
Brian Modoff – Deutsche Bank
And that’s a key understanding. That’s right. Okay. Competitively, what is Cisco's sustainable differentiation in telco, NfV and SDN compared to say Juniper with Contrail and Alcatel-Lucent Nuage solutions?
So if you take a look at the examples that I laid out for mobility. If you take a look at the different elements of that full architecture. Like, how should we get into some of these constructs about virtualization, SDN and NfV and about programmability, it's not just about having a point product in one of those areas, it's about having the ability to have horizontal solutions and stitch them together. And that’s where I feel we have the widest breadth and depth as well as not just product and technology but also our services capability that’s world class. And we can do this for operators and do it in a much more faster way and a simpler way than perhaps a point product that's just kind of focused on only one domain or one particular instance of it.
Brian Modoff – Deutsche Bank
This is the architectural systems approach to it.
Absolutely. This is it. This is absolutely and architectural approach and an architecture way. We have got to minimize the complexity from a technology standpoint but also deployability and serviceability standpoint.
Brian Modoff – Deutsche Bank
So let's talk a little bit about mobility. You have acquired Starent a few years ago. It's funny we had Intucell on one of the tech talks and about two weeks after it you bought them. So you have acquired Intucell here recently. You have acquired Ubiquisys, a femto company. And you have acquired Meraki. And people tend to think, oh, Meraki, how does that apply into this. So can you kind of stitch these pieces together? What is that makes Cisco and what is the kind of thing that Meraki brings to the equation, particularly with regard to femto deployment that the market may not be appreciating.
Yeah, so I think it's -- you have asked multiple questions. First of all, I still recall you inviting me to that tech talk from Intucell but luckily by then I already knew what I needed to know. The second thing is, right, most of the moves that we are making in the marketplace seem as kind of a one off move, but there really is a broader thinking behind these. When you take a look at, particularly my business, as we looked at Intucell and Cariden, and BroadHop and Ubiquisys, it wasn’t individual piece parts, it's how we cloud stitch those together and make the one plus one plus one equaled six, not just three. The second thing is, when you talk about Meraki, the one key thing around Meraki is that most people think that it was around the access points etcetera. Well, yeah, those are interesting, but it's really about how they have actually approached the service deployment. How they have addressed the simplicity for the enterprises, with the operators about the business model. And how they have actually looked through and done all of these functions in a virtualized way inside the cloud and more importantly, stitched them together to be a solution so that you can have Wi-Fi service and internet service and VPN services and all those things that you would inside a very physical environment into a more virtualized environment.
Brian Modoff – Deutsche Bank
Operator, I would like to open it up for some questions.
(Operator Instructions) And your first question comes from Grady Burkett.
Grady Burkett - Morningstar
Thank you for hosting the call, it's been terrific. Just a question on the hypervisor asset. How much software development or integration happens between Cisco delivering the services layers with the hypervisor supplier? And does it ever make sense for Cisco to kind of develop its own hypervisor or build on top of an open source hypervisor? Thank you.
Good question. And first thing is, when it comes to hypervisor, our customers have many choices. And they really want to make the choices themselves, because as you look at this model, right. There is a different level of stack emerging and each layer of the stack, they want to make their own choices on what's compute environment, what's hypervisor environment and then above that what applications they want to choose. So our approach and our goal is to be open and to be able to work across any hypervisor environments that are out there today or might come up. Whether it makes sense for us to look at one or not, that’s really going to be led by our customer's thought process, but right now they have enough choices today and they are making those choices based on the current environment.
And your next question comes from the line of [Peter Callahan].
Thank you for the call. You guys gave a lot of great color around the NfV strategy for Cisco here and I was just curious, does this potential roll-out of your solutions potentially result in lower ASPs given the heavy software features? And does Cisco have any plans to offset this, if that's the case?
Well, Peter, so the first things is is, like I said before, we see this is an incremental opportunity for us because it allows us to go in and sell into many functions or appliances areas, where most of the customers are actually getting them from someone else. That’s one. The second thing is, actually the margin story on this is quite good because they are being sold as virtual functions. And of course the model has to be different because it's not a physical asset you are selling, it's a virtual asset you are selling. So you got to monetize in a different way. And those are the conversations we are having customers around what are the models that make sense for them and also for us in terms of how to evolve in that model. But you have got to see this as more of an incremental opportunity and allowing us to get into those areas that we weren’t in before.
Brian Modoff – Deutsche Bank
Another question is, you have recently rolled out the ASR 5500. Can you talk about what are some of the key use cases you are seeing for that product and how has carrier reception been of that new high capacity box?
Sure, Brian. So the ASR 5500 we launched last year. Today we have seen global deployments in pretty much every geography today, every continent. And most deployed cases have been around either 4G or 3G packed cores. And what's interesting is is that, our conversations with customers are really around the call flow models that I talked about before, which is what they are anticipating in LTE and 3G combined. And it's very unique because as I said before, you know there are certain things you can virtualize other things you can't. And this is one area where the combination of bandwidth, memory, IO, and some of the layer two assists that you need to do to do those functions is absolutely required by operators in that environment.
Now it's really interesting because as you have seen, we have been growing our market share, have been very successful with customers based on the platforms and we are continuing to do that. And the conversations that we have with customers once we get installed in the packet core, are really about some of these other complex challenges that they are having. And in many cases, we are actually engaged with them on some of these other complex challenges and it allows us to have a conversation about the packet core and for the need to evolve the packet core into a different approach. Whether it's a hardware-based approach for mobile, broadband services or (inaudible) type deployment, or it's a machine to machine type core network which could be in a different way, deployed in a different way as well.
Brian Modoff – Deutsche Bank
Thanks. Any other question from the audience?
Brian, I got this one from the web.
Brian Modoff – Deutsche Bank
So this one's from UBS. Can Cisco deliver on its mobility aspirations without owning a macro cellular access pieces like 2G, 3G, 4G macro cells, especially when competing against Ericsson, Alcatel, and Huawei.
Well, I would tell you that today we are seen as the number one player in mobile IP infrastructure. So I would tell you that we are doing that. And the macro radio assets, many of the operators today have the macro radio assets deployed and in many cases it's one of the traditional RAN vendors, whether it's an Ericsson, NSN, ALU or a Huawei or whoever else. So for us, what's important is to be able to help them with that heterogeneous network environment. So for me it was more about leveraging that installed base and helping them optimize that network with a very important function called SON, right. So SON for 3g and SON for 4G or LTE, that’s where we have inserted. And that’s where we have become extremely relevant in the radio space without actually being in the radio space.
The second thing is that, if I take a look at most of the investments that is going to go on in the radio networks and a lot of the operators are talking about this shift from macro, so small cells. In that scenario we are playing very effectively whether it's with SP Wi-Fi today and also with license small cells that are coming up. So certainly in that area we have already made our play and talked about it and even made some acquisitions in that space.
Brian Modoff – Deutsche Bank
Will you consider, if you think about this as a lot of IT oriented architecture, so this is network intelligence. Can you talk a little about your, perhaps your advantage in this area and understanding this relative to say an Ericsson or Huawei. Talk about some of the use cases for Intucell and tie the cloud architecture of Meraki into that.
Okay. Well let me -- I don’t know if I will tie the cloud architecture and Meraki but let me tie this together for you. So, you are right that as we have looked through the mobile broadband, and I talked about the internet and mobile have converged. It's really happened with 3G and 4G. So IP, or internet protocol has become more relevant inside mobile networks today than it has even been before. So when we talk to customers, they look to us as being the IT experts. And it's really kind of that that allows us to have that conversation. And the radio becomes primarily an access technology, whether it's a Wi-Fi or a 3G or 4G or whichever G that might come up.
Now what happens in say, particularly with your example in the LTE or in the Intucell case, most of the networks today are congested in the 3G environment. Because that’s where the network sees a lot of load on it and a lot of things going on. And the challenge that the operators have is how do I apply this case again to the constructs that we have talked before, which is to be able to optimize the network I collect all this data from the network. And I take a look at the data, the RF data, then I have got to manually go and process that. I have go to figure out where the neighbor lives I need to optimize. Then I have got to go manually reconfigure different elements of my network and I have got to go program the network but do it manually, or do it through scripts.
Now what Intucell does, it does exactly the things that we talked about. Which is extracting intelligence out of the radio network and then it takes this information, analyzes it and it applies it back and has a policy based [thing] and it programs the network by itself. And it actually does it on a near-real time basis. And the near-real time in the radio world could be in the order of minutes. And going from the orders of minutes from weeks is dramatically better for the operators. Why? Because the network can actually self tune itself to be able to adjust the capacity, have self reliability. And if one of the changes doesn’t take effect in the right way, you actually measure it and then you can reprogram the network back dynamically. And we have had operators that have been doing this for a while and in a very successful way. And the improved user experience by providing lower dropped call rates and also improved throughput inside the network.
Brian Modoff – Deutsche Bank
Can you tie -- I mean a lot of investors were familiar with Ubiquisys. We knew the company. It's been around for over three years and we have seen them in Mobile World Congress. But a lot of investors aren’t familiar with Ubiquisys and some of the capabilities they brought to you. I have seen them built base stations inside of a cigarette pack. They have exposure to companies like Vodafone. Can you kind of -- what does that angle bring to you to Cisco.
Sure. Now I will tie this into Intucell as well because this is again an important part. Because for us to be able to insert small cells into a existing macro radio network we first needed to be able to optimize the macro radio network, and that’s what Intucell brought. What Ubiquisys brings to us is the following assets. One, they bring us a lot of knowledge and technology in the RF domain that we need along with the Intucell folks. The second thing they bring us is a set of assets. The assets are, one, the physical assets which is the licensed small cells that they have been building. And they actually build them in two different models. They either build the hardware themselves for the high value and the premium customers in particular areas. And for other areas where it's commodity business, they actually just provide the software and service provider can have the option of buying that hardware from a contract manufacturer for example, right. But they provide a hardware based reference design that those folks can build.
The other thing they provide is, again, that software asset that sits on the AP itself or the access point. The second thing they bring is the ability to be able to turn up a small cell very easily and very dynamically. So, for example, when you buy a small cell from an operator, you take that, bring it into your home or into your business and you turn it up. All of the provisioning, all of the programming, all of the management that’s done is done from the cloud and it's done on a per provider by provider basis. So many of the constructs we talked about in the virtualization, NfV, all of those things are virtualized sitting in the cloud and you can actually provision those things very easily and very fast without much touch points.
Now you have got the small cells and you have got the macro cells and you can optimize with the Intucell capability and the self organizing capability that the small cells have, optimize the performance of the entire network. The other thing that this allows us to do is Intucell gives a map view of the RF domain and gives us the view of where the hotspots are for the operators. And where those hotspots are, we can now go in with the Ubiquisys technology and plug those hotspots for the operators and give them the capacity that they need to be able to fulfill the demand.
Brian Modoff – Deutsche Bank
So [network consolidation]. Vijay?
Vijay Bhagwat - Deutsche Bank
Vijay Bhagwat, Deutsche Bank. Kelly, it's a very helpful call. A question for you is, what are your top carrier customers telling you? What I mean by this is, if you look at NfV, it's primarily a cost savings value proposition. It takes cost out of the equation. If you look at SDN, it's a programmability and automation value proposition. If you look at big data it's a monetization value proposition. So if I was one of your top service provider customers, what would I be telling you? Would I go about SDN, NfV, big data in a horizontal manner, look to virtualize the infrastructure assets, look to automate some of the assets, look to introduce monetization? Or would I go about it in a very piece meal manner. Look at a very specific use case or bringing a new mobile data service to market. Look at a very specific use case of automating the radio network. What are your carrier customers telling you?
Actually that’s a really interesting point, because it’s exactly that that we are seeing which is -- when you think about the three problems we are trying to solve which is the cost of ownership, the speed, right, the velocity and the agility and reducing the risk, many operators are going about it in a different way. They all agree with the end vision, right. But what they are actually taking is different steps on how to get there. So on example is an operator in the wireline context that wants to be able to provision services a lot faster and a lot quicker. And this is just broadband services provisioning, right. That’s one operator in Europe that’s doing that.
The second operator is saying, hey, you know what, I have got a lot of complexity in my video service chain that I have got to build out and I keep having a lot of challenges because I have got to scale that environment, I have got to add more appliances in and that’s just getting too complex for me. So that’s their problem that they are trying to solve. Another example is, is an operator that’s trying to just do some smart billing around -- billing plans around family usage and all these other things or the toll free data or the kind of sender pays user model. That requires a different level of a service chain that can be deployed. So many operators, depending upon where you are, coming at it differently. But we are seeing all of them. So the interesting thing is we are seeing all of them. And people call them use cases, I call them specific problems or pain points that the operators have. And depending upon which operator you talk to, they may have a different pain point. And the value in this is that we have got all the piece parts that we can stitch together very quickly and very effectively and show them how we can solve their problem.
Brian Modoff – Deutsche Bank
Kelly, there are two questions that are coming in from the web. One of them is, what role does a router play in NfV? Do you sell less routers or actually the router becomes a source of deploying NfV? That’s the first question. And the second question, we hear this term service chaining quite a bit. Can you explain what service chaining is and how Cisco's approach to service chaining goes?
Okay. Great. So if you go back to the network pictures that I drew, the picture about the radio access network and the wide area network, and the core network and the data center. If you think about routers, right, we use routers today in the wide area network, right. And if you think about all the core and the edge routers we build up, those are still required and play a very important role inside the network infrastructure. And that’s part of the reason why I call this is an incremental opportunity, because anything that we do is not going to change a CRS or an ASR 9000 to be virtualized, right.
Now if you go inside the data center, you will require some additional routing capability. Why, because you have got a lot of elements inside the data center, not just the data center interconnect router but also the other elements that are inside it, meaning the routing function and the routing capability inside that. And that’s why we developed the cloud services router, to be able to allow us to do that. So there is routing required in a virtualized manner inside many of these constructs and that’s why we did the CSR and also the NK1V because you require a virtual switch as well. Because if you think about all of these services and let's talk about the second part of the questions, which is all of these functions or so called services, just because they are virtualized doesn’t mean they automatically connect together. In the old days we would physically connect these appliances through Ethernet switches or routers and then kind of configure them. In the virtual environment we need to do the same thing. And that’s where the Nexus 1000V is really about connecting the services or functions in a layer 2 way. And then a layer 3 way as well with vPath or cloud services routers to be able to connect those boundaries.
Now the other part was, what is service training. Now if you take a look at all of these functions, right, that I had around the internet user or the video user, all of those instances that have to be connected on a path basis or a session basis, are what's called the service chain. And the service chain could be a NAT function or a firewall function or a video optimization function or a combination of all of those stitched together in a way that’s done today in physical environment but doing it in a virtual environment. That’s what the construct of service chaining is. And the other part around the service chaining is that, as you create more service chains, the difference between the today's world and the tomorrow's world is, those service chains have to be very dynamic, you have to be able to configure them on the fly on a per user or a per session basis, which is not the case today. Hopefully that helps to answer that question.
Brian Modoff – Deutsche Bank
I think one of the things I had to think about with regards to NfV in service chaining is really the velocity -- increasing the velocity of rollouts and reducing the cost of rollouts, and allowing carriers to become more like the OTT vendors in terms of being able to delivering new services, a lot more experimentation without a big penalty in terms of cost associated with that as has been the traditional way of rolling out new services. A couple of other questions around femtocell and 802.11ac. There has been a lot of hype around small carriers, small cell deployments. When do you think we start to see these accelerate and then how much do carriers and enterprises care about 802.11 ac and will the transition to ac happen faster than the transition to N and what will be the key driver for that?
So ac has started already and we have actually introduced products that allow us to make all of our Wi-Fi access points have been shipping for a while to be ac capable. So that transition has started and it's well down its path. So I think that it is going to take time. Anytime these things happen, I mean we are talking about massive amounts of Wi-Fi access points that have been deployed today, whether it's with enterprises or SPs and now they've got to go and upgrade those. So I think the path is along its way and it is going to solve a challenge and a problem for the operators.
The N again is a same thing. As N comes up, because our path today on the ac is through a module and it's going to get integrated together inside the access points, but there is a lot of other dependencies in these environment as well. So for example what happens on the client side or the device side is also a key part of the equation. And that’s where some of the device guys have been moving along faster than the others but I do think that we seize the momentum picking up now, the wireless broadband alliance meeting that happened in the UK over the last couple of weeks actually showed some good traction. And not just traditional Android folks but also other device folks kind of planning up behind ac. So this is all great stuff and we are really excited about that.
Brian Modoff – Deutsche Bank
And then on small cells?
Brian Modoff – Deutsche Bank
In terms of the acceleration of small cell deployment.
So the small cell, it's kind of interesting area because most of the operators have got a hype around small cell. But really where the operators are interested in small cells is to go address that problem. And in many of the cases, the problems is they need coverage and capacity in the most cost effective way in the areas where they have challenges with those. And in many of those areas those challenges are not necessarily outdoor, they are indoor. Because most of the usage and most of the data usage that’s happening inside the network is actually happening in an indoor environment, whether it's an office or a home or a venue. So that’s where I think the small cell transition is going to start and we are starting to see a lot of activity since our announcements at Mobile World Congress and the acquisition of Ubiquisys, a lot more activity coming our way around that.
Okay. I have got a question coming in from the web. What advantages does Cisco when you talk about architecture with your data center solution, things like UCS, Cisco ONE. You mentioned Nexus 1000V that separates you from potentially a Juniper or Alcatel when delivering NfV. Can you leverage Nexus 1000V, can you leverage UCS as you think about a solution around....?
So, we know the key advantage and this is one of the key things that I talked about, this is an architectural approach right. An architectural cell and a solution that you have to deliver. And you need different aspects of that solution. Right. So when you think about NfV and all the things that you need to talk through and stitch together, some of those are radio access network in the mobility domain or the wide area network, but in the data center by itself, what do you need. Well, first you need the infrastructure. UCS and the physical assets of the Nexus 7000 switches or the Nexus 3000 switches. Then you got to compare and say and say does competitor a, Juniper have those. I mean clearly they have switches but what about UCS. So the second thing is, once I have got those assets is their approach kind of more open or more close.
So for example, what hypervisors might they support? What orchestrations did they use across those resource environments, right. And we have got (inaudible) resource environments orchestration that we acquired, we have got kind of elastic controllers that we have been building to solve that problem. The third thing is is, now when you take those applications, right, specific applications, these are firewalls, these are Nexus 1KVs. As I talked about before 1KV is about stitching across in a virtual environment providing that layer 2 switching capability. You have got to look into, does somebody have that or not? The third thing is, how do you do that at the layer three level which is what we passed out. And the last part is, how you stitch all these things together, the orchestration part. And I think folks, other folks in the industry, you know other competitors may focus on specific use cases that maybe applicable only in one environment. But what we are building as an architecture is something that the operator can take and build and use it in a mobile environment and a video environment, or in a security environment. Or in other environments as well, a VPN environment.
So that’s the commonality piece that I think that some of our competitors will struggle with and the lack of architectural depth they will struggle with.
Yeah, Kelly, a question for you. You talked about NfV as mobility. Is Cisco currently involved with any NfV SDN oriented project with the telcos and when can we see some of the earlier deployments of NfV and SDN oriented projects at the telcos.
So clearly, the first things is we are engaged on the standards standards activities in ETSI. So we have working groups and our teams are well engaged there. In terms of working with operators, we have several engagements of operators around the world on what could be qualified as NfV. In fact we have been engaged with those operators much before NfV as a construct came about. In fact some of that construct came about because of the work that we were doing with some of these operators because many of those are part of the initial NfV kind of group that was set out. So we are engaged in those environments and some of those capabilities are starting to show up in LANs or in trials. And the key things around NfV is not so much the technology capability, for the operator it's a very different environment. Right. How do you do capacity planning, how do you do operationalization. And that’s why they are taking a problem containing in a specific environment and getting their team used to it because what happens today in an operations environment is, operators have very contained group that focused on specific parts of the network. And what virtualization does, it actually breaks those boundaries, SDN breaks those boundaries. So they have got to come up with the new models of how do they build some of these environments where a lot of these capabilities moving inside a compute environment and how do you do capacity modeling and operations for those.
So because of that, that journey started for the operator a long time ago and they are continuing to make progress on taking those models, changing their processes, changing their procedure and working while with them on that journey.
And then a quick follow up is on your thoughts on Cisco's big data strategy. And the reason I ask is, the mobile network is a rich treasure trove of information and data, what are your thoughts on laddering upon NfV programmability and then moving to service monetization. Your thoughts on Cisco's big data strategy.
Yeah. So big data is just as clear as cloud and SDN have been for a while. Everyone talks about it but no one can really put their finger what the data is. To me, there was really two types of data. One is data that we know about an individual service or things that we have. That’s really data address. And that’s not something that’s of interest to the operator because that’s something they can mine and look at and analyze. What's of interest to us and particularly in mobility motion, right. Data in motion. Because I feel that data in motion is what the operators want to be able to change the network environment for the optimization capability while also be able to leverage it for monetization. So one small example but a very very critical one that we have actually been working on aggressively with many operators is that of location based service. Particularly where it's shown you first is actually in Wi-Fi, an indoor location based service. That’s become a really really interesting way for the operators because that’s where they see a huge opportunity to tie context and location. And we can get accuracy of user down or a device down to a very very thin, very fine point where retailers can advertise or promotions or other things can be done to be able to drive growth.
So that’s starting to happen and there is more instance of that starting to show up. The other part is, is big data is not something that can be looked at across the entire network. You got to segment the network into different chunks. And that’s why I think take a look at that three walled triangles that I drew, each one of those has analytics and data coming from that particular domain in the network to be analyzed and acted upon. So data is good. If you don’t use the data for something, which is to go back and program the network or change the network or apply a service, it's useless. So you got to do it in near real time and close the loop on that.
Kip Clifton - Deutsche Bank
Hi, Kip Clifton from Deutsche Bank. Would Cisco ever consider partnering with another vendor, for example like an NSN to offer a complete macro and micro, femto, backhaul and APC offering?
So interestingly enough, we actually work with several macro vendors already today. So, for example, for a couple of the RAN vendors, where in particular situations or accounts, we provide them the backhaul solutions that they need for cell site routers and aggregation. In other cases we actually do a reference cell for their microwave solutions. And you've recently seen at the Small Cell Forum, what we were talking about in terms of our partnership strategy for wireless backhaul leverages several partners including small players as well as big players as well. And as we look at some of the new assets we have acquired, many of those have actually sold through traditional RAN vendors as well, whether it's small cells or SON capabilities. So we have been actively working and even in our traditional portfolio of routing and switching many of these traditional RAN vendors are launch partners.
Brian Modoff – Deutsche Bank
So one thing I remember back from Mobile World Congress, not this last year but the year before, was at that point in time you were in 23 of the top 30 carriers of all packet cores. Any updates on that? Gained share....?
Well, given that we have been gaining share, we have grown that number fairly well. I can't recall what John talked about it -- I don’t think he has talked about it. So I can tell you that the last number that was published was over 40 design wins in LTE, and that number has gone up from the original number that we talked about last year for at least the mobile environment. But the result is what you should be looking at, which is we are number one in mobile IT infrastructure, we are number one in mobile packet core, we are number one in mobile backhaul, and we are also number one on SP Wi-Fi.
Brian Modoff – Deutsche Bank
Okay. With that I think we will wrap it up. I would like to thank Kelly Ahuja and John Choi from Cisco for joining us today. We will have a transcript of this call available later on today. If any of you have any follow-up questions for any of us, you have our numbers, please reach out to us. And I would like to thank all of you for attending this call and have a good rest of your day. Thank you.