Seeking Alpha
We cover over 5K calls/quarter
Profile| Send Message|
( followers)  

Cisco Systems, Inc. (NASDAQ:CSCO)

JMP Securities' Technology Research Conference

March 3, 2014 02:00 PM ET

Executives

Soni Jiandani - SVP, Marketing

Analysts

Erik Suppiger - JMP Securities

Erik Suppiger - JMP Securities

Okay. So we’re going to kick it off. Thank you all very much for joining us. For those of you don’t know me, I’m Erik Suppiger, the Networking and Security Analyst for JMP. And with us is Soni Jiandani, the Senior VP of Marketing for Cisco. And this is going to be an hour long session. This is entirely a fireside chat format. The object here is to keep you engaged with questions. So please do feel free to ask whatever questions you have. The reason we wanted to set aside an hour for this is because rarely do you have such a major transition going on in a market as big as what SDN could possibly be and for Cisco ACI. And so I really wanted to give us the time and attention to get into some of the nitty-gritty on Cisco’s positioning here.

So I’m going to start this off. Soni has some slides, but for the most part this is a Q&A session and I do invite all of you to ask as many questions as you have. So I’m going to just start it off. I want to make sure that people have a good grasp of Cisco’s perspective on what ACI means. So, there is a lot of different views of what Software Defined Networking is, I don’t think everybody has an in-depth understanding of Application Centric Infrastructure, that’s ACI, for Cisco. So could you just basically describe what does Cisco, what does ACI mean to Cisco and what does the Nexus switching platform address in terms of SDN?

Soni Jiandani

Very well, thank you. So if you take a look at the traditional networks that we have seen built over the last 2.5 decades, they have been built keeping in mind the focus on large scale networks that can -- that are stable in nature and that deal with static connections that are made between your applications and your servers and your storage devices across a more of a static IP network. And that fundamentally has been the approach for networking over the last 2.5 decades.

Now it becomes physically important to recognize with the evolution towards cloud and evolution towards mobility, the applications are very quickly evolving in this new paradigm. And as you construct customer datacenters and you look at both the emerging applications that are coming into their environments as well as a traditional applications, you need a new model. And the new model has been attempted by some companies that are purely software based approaches, which try to apply the constructs of SDN, which includes network virtualization on top of an existing IP network which is static, basically. And they try to recreate another network on top of an existing network using a software only approach, and that is what I would call the approaches that are taken by companies like Viscera/VMware is trying to construction an overlay network, all in software to ride on top of an existing network.

And you are recreating all the problems that you already have now in another network which is running purely in software because you are recreating the networking constructs all over again. We tried that about 15 years ago with ATM LAN Emulation. We try to emulate Ethernet on top of ATM and our customers told us you’re not solving any of my problems, you’re now giving me two separate networks to go run and manage, then you’re trying to bridge the gap between these two underlying technologies. And you continue to stay within the constraints in the process of trying to have a network abstraction layer, or what’s called network virtualization.

It became very clear to Cisco as we were looking at the transformation required that we have to keep the application at the center of this transformation because you wouldn’t even have datacenters if you didn’t have applications. You wouldn’t need servers if you didn’t have applications. You wouldn’t need storage if you didn’t have applications. And you wouldn’t need a network frankly if you didn’t have application. But the language that is spoken by the application teams is very different. They think about how can I rapidly provision applications or decommission applications across this datacenter, across this infrastructure, which is compute network and storage. They think about the notion of where do I locate this application, where do I place it, what types of compliance and security requirements does that application have for my business. Those are the things that application teams think about.

The network language on the other hand has been focused around how do I deliver a highly scalable infrastructure, how do I deliver highly stable network in a very consistent manner that can move packets around and can deliver traffic across point A to point B, whether it’s the market data application to run Wall Street applications or whether it is Mission Critical Oracle Database applications or Mission Critical SAP/ERP applications or the emerging HANA applications or the upcoming virtualized or Big Data applications.

But the network has been primarily focused on reliability, performance, scale and stability. What needed to happen is that the connection needs to be made where you can drive this next generation of an architecture, of a network transformation that can have an application centric view, that can now provide an abstractions, which is not just network centric, but is application centric. And it transcends beyond having to recreate the network all over again at a different layer.

And so our approach really has been to not necessarily drive the same software defined models on existing networks but is to look at where we can focus on applications, where can we focus on bringing application centric abstraction on top of networks that we can make highly efficient, highly programmatic and to enable application agility across this infrastructure day one by providing ACI which is network focused but day two, that is expandable beyond networks and can encompass compute and storage, because the datacenter is a collection of these three elements in the infrastructure. It’s not just the network alone. So ACI transcends beyond the network and is expandable to go beyond networking and to go and encompass compute and storage as well from an architectural perspective.

So in summary, what is our day one vision with ACI? Our day one vision with ACI is to enable the rapid deployment of applications, both virtualized applications and mission-critical bare metal applications across networks and deliver it with full scale, full security and complete visibility. And when I talk about visibility I just don’t talk about visibility associated with the network and its troubleshooting characteristics. I also talk about allowing for an application person to determine where to place an application, where to decommission applications from, because if I don’t have that visibility, I'm frankly running blind across that overlay network. So full visibility of physical and virtual.

And there are really three key elements to ACI. The first key element is a controller and that controller is going to be the central point of management and policy repository information. It will provide the ability through which you can assure that your applications are deployed with the right security, with the right SLAs across this highly programmatic network. And this network has to continue to deliver the best of great price, performance, port density, programmability options and it has to be highly programmatic and meet the needs of the diverse applications on the policies that get pushed into this network in a distributed manner. And this application centric network abstraction has to map to applications and cannot just map to the network alone because if it does not map into the way customers are deploying that applications, we will fall short in terms of truly delivering value to the datacenter consumers, maybe applications are sitting at the heart of it.

Question-And-Answer Session

Erik Suppiger - JMP Securities

Is that to suggest that in the case of an alternative SDN solution that they are not mapping to the network, they are mapping to the applications but they are not mapping to the network effectively?

Soni Jiandani

And if you look at the SDN models that are software centric and are looking at a compute centric model or that are looking at virtualized centric view, they fall short in many areas and we will come to that a bit later in the conversation and doing a real comparison of one versus the other.

Erik Suppiger - JMP Securities

Okay. So, talk a little bit about the Nexus 9000, how does that switch interface with ACI? What is the unique aspect of having a combined switching and ACI architecture?

Soni Jiandani

So, if you look at the picture on my, I am facing -- the existing three tier design picture on this chart, our customers have built hierarchical networks. They have built networks within two tiers and three tiers. What became very important for us is that to recognize that as they were building the traditional networks within their datacenters, we had to accommodate the Nexus 9000 to fit into their existing designs. And as it fitted into the existing designs, it would provide them with the programmatic APIs and allow for their operations team, their network operations team to drive high degrees of programmability using the existing interfaces; whether those existing interfaces are tapping into the command line interface, whether those existing interfaces want to exploit JSON as an interface.

So, we enhanced the Nexus operating system to make it highly programmatic, allowing not just the cloud customers with us but also the traditional enterprise customers and extended it also encompass open stack support, allowing us to run a more modernized operating system, built upon the matured Nexus OS that we have been innovating in for the last six years and added high degrees of open API and programmability. Then there is a category of customers that want to go beyond that, that want to build overlay networks and they want to use the Nexus 9000 to build overlay networks with full routing and switching capabilities. They don’t want to reply on external servers to be gateways with merchant only boxes, right. And that’s the picture in the middle which allows them the ability to support integrated network virtualization with full overlay network protocol support like VXLAN and NVGRE switching and routing.

And they can -- by leveraging open daylight, open source based initiatives, they can use any controller that supports open flow. It gives the customer the choice, the freedom to choose what controller they want to use within an overlay network, including their own homegrown if that’s where they are going with this.

And the third alternative is really a turnkey solution. The exact same product, with a different software image will have the ability to provide a turnkey application centric infrastructure experience for customers that want the ability to future proof their operational model and their infrastructure and they want to be in a position to drive the integration of SDN technologies with full automation across their networking teams, their security teams and their application teams. Because ultimately if 80% of the dollars in IT are locked up in operational expenses, your biggest bang for the buck in embracing new technologies like SDN will be where? It’s through automation and cutting down that operational expense, and in turn have the flexibility to move it more agility and the ability to have your apps deployed the way you need them and when you need them with the right security and with the right SLAs and drive a level of automation that has not been enabled in the past across your IT silos, which is fundamentally another benefit that you derive from an architecture like ACI.

Erik Suppiger - JMP Securities

So, talk a little bit about the 9000. I think you just recently launched it.

Soni Jiandani

Yes.

Erik Suppiger - JMP Securities

What are some of the initial customer feedback? Are you in production networks? What kind of initial success have you seen?

Soni Jiandani

We have been placed with the success of the Nexus 9000. We got awarded, the fastest 40 gig router stature in performance numbers were we have seen very good traction in our cloud provider's customer base. We are also starting to see our enterprise customers across various sectors, including financial services that are looking to embrace this portfolio in some of the most mission critical part of the networks, and commercial customers. So we are seeing service providers, enterprises, commercial customers. And we see strength across all major geographies.

We have pipeline of over 500 customers that we are tracking on the 9K and ACI already, and very strong acceptance across all major geos. Of course, U.S. being the lead geo for us, but we’re very pleased that we have customers in Europe that are deploying the product, as well as Asia Pacific and in Latin America. So we’re seeing geographic strength and acceptance of the product. And it’s now been close to a quarter and a half that we have been shipping the portfolio. We continue to enhance it. And the APIs that we are making available are being used by a lot of cloud provider customers already. So they are taking advantage of the versatile API and the programmability attributes of the platform as we have rolled it out.

Erik Suppiger - JMP Securities

Can you talk a little bit about why does it need to be integrated in hardware? Why does it – the world has talked about software defined networking, people have talked about Cisco’s approaches with hardware defined networking. Why do you have to need this embedded into silicon? What aspect, what advantage does that bring to Cisco?

Soni Jiandani

So, give me an opportunity, first to talk about some of the use cases. Because when I talk about the use cases and then I talk about why the approach had to be taken in a different manner, you will get a better appreciation. I get asked this question a lot. What are you going to -- what does your customer perceive as the biggest used cases for ACI.

At the least common denominator level, the way our customers plan to exploit ACI, is to have the ability to define the connectivity policies for their applications, their virtualized applications and their bare metal applications so that you can come to one place and define, what are the quality of service attributes for this application, what are the security attributes for this application? What are the firewall rules allowing and disallowing the communication of this application tier, the web tier to the database, the database to the application tier and so on and so forth. And that’s at the least common the dominator, where they have a combination of virtualized applications and bare metal applications and they want to define in one location the policies for connectivity.

Now let’s go for the second use case. The second use case is a more real-world use case, as well. It’s as real world is the first use case I just articulated. In any network I happen to have lots of security appliances like firewalls, intrusion prevention systems; to protect not only the boundary of my data centers but also to keep my production traffic separate from my development traffic and to prevent from any intrusion into my internal data center assets.

Today the process of connecting a policy of security into a network policy is a manual task. So if an application team needs to rule out an application, they have to span seven to eight different organizations. And the process of commissioning applications takes weeks, if not months, because it is a mundane process that spans multiple siloed organizations that don’t have any automation and built into them across the IT silos.

If today I look at what are some of the most vivid, what are most applicable used cases for security with ACI, this is the most common used case, is that I can some to one place, the controller and I can define all my security policies for my compliant applications and my non-compliant applications through the controller and make it application centric, and drive a level of fine grained and security in that network that I would been able to get and try to now apply that with a pure software overlay model. You still have the constraints of the underlying infrastructure. Right?

Erik Suppiger - JMP Securities

Yes.

Soni Jiandani

You still have to deal with all your physical appliances whether they come from checkpoint, whether they come from another vendor, another supplier. You still are manually tying all of this together, because 99% of all your appliances are physical, they are not virtual. Right?

Erik Suppiger - JMP Securities

Yes.

Soni Jiandani

Therefore through seven services. How do you automate it if you don’t have mechanism to tie in existing third party physical appliances into this automation and this policy framework, which the APIC controller can deliver day one to our customers, and automate it across their networking teams and the security teams? Today what is a manual process will now be automated. You’ll have the ability to accommodate your security policies and map them into the policy model of APIC and ACI as is.

The third most important use case that customers tell us is, if I want to build an on-premise cloud and I want to have an opportunity to automate that and to allow for a common policy model, that can also determine how I would burst it to the hybrid environment, the ACI and the APIC model provide them the technologies to allow them to drive a cloud based deployment model from an automation and a workflow perspective.

So if you look at the three used cases, you will see that a classic network model with a pure software only overlay will not be in a position to meet any of these use cases. A software only model will not allow you to bridge the gap between the physical and the virtual. It’s for a pure virtual play. We’re still managing our physical networks, independent of the virtual networks. A pure software overlay model will not allow you to tie your existing physical layer 4 through 7 appliances. It’s only for virtual. For your bare-metal applications you still left -- you still have to figure out a different way.

And last but not least, you do not have the latitude to go through the process of thinking about again, your mission critical apps and your non-mission critical apps and your bare-metal apps and your secure and PCI compliance apps that should have differentiated granularity and control unless you have not thought about this architecturally bottoms up.

Erik Suppiger - JMP Securities

Is the inability to do that because there isn’t sufficient interfaces performance with merchant silicon to support that flexibility?

Soni Jiandani

And that’s a very good question and for that I will need to get you to a particular point to describe that to you. If you look at the SDN models, that they are attempting to build on top of an existing network, another network, a software overlay network. What are some of the gaps when you start to build another network on top of an existing network? You don’t have -- you have to manage the network in two places, because there's no co-relation of the physical to the virtual. You are managing a physical network and you also have to manage a virtual network.

Now when this virtual network has updates, it is updating -- it’s also those that are running this overlay networks have to keep track of all the tables and this overlay network and if there are any changes -- you are keeping a lot of states in the software. And so your servers which you bought to run your applications is actually running a network. So a lot of the CPUs are getting hogged because now they are supporting multitask and broadcast and they have to keep track of what changes are happening in the virtual machine world, because now I have to map all the changes that’s happening. So if I introduced a new virtual machine, all the end points, all the network end points have to be kept updated with those changes.

It’s more complex and it’s more unreliable, because take an example where I have 1,000 servers and I have 10 virtual machines per physical server. I'm talking about 10,000 virtual machines. And in an environment whether a 10,000 machines, even if there are -- even if you reliably deliver a networking traffic at 99%, for that 1% if you drop, you are talking about on an average, if there a few changes 1,200 times where you could create your dropping packets, because you are maintaining a network and you are trying to run server virtualization at the same time.

In networking it's worse to drop a packet versus deliver a packet to the wrong address. So if you have not kept your network tables updated, you’re in a very bad situation. And now you have a disjoint overlay network and an underlay network. So you have to think about this at a system level. You cannot think about this as a pure software model versus a pure hardware model.

You have to think about it at a system level, where the network has to have the ability to support an overlay environment. The network has to have a certain policy framework. The network has to allow for a certain level of fine grain control and security before you can reliably say, before you can consistently say that I have delivered a system, on top of which you can run your mission critical and your non mission critical apps across. It has to span physical and virtual. It has to give you the visibility of the physical and the virtual. It has to drive simplification. Those are some of the foundational building blocks that we had to approach with.

There’s a blog that was written by our CTO, Tom Edsall and I pulled, I just pulled one table out of that. This is just a basic example of comparing ACI versus a server based SDN model. If you look at a network which is supporting SDN, you have to support an overlay network and you have to support a controller. There are some basic functions a network does. It packets. It does packet forwarding and here with ACI we have just embraced the traditional packet forwarding model, whereas in the case of server based SDN models they have recreated or tried to recreate a full network in pure software.

The other key element that you have to think about when you’re running an overlay network is a mapping database, where you are looking up; you are creating another overlay network to run, to give you the flexibility to place your applications anywhere. In this case you will be heavily relying on your server CPUs and it’s questionable whether you can scale a network, a distributed system purely based on your servers and the CPUs and the impact that you will have in terms of trying to troubleshoot a network, which is not plausible because you are now dealing with two separate networks, vis- á-vis with ACI where, we’re taking an approach where you’re managing your physical and your virtual networks in one place, having no performance penalty because you need to have ASIC. You’re running your servers at 10 gig speeds; you’re running your backbones at 40 gig speeds, right. So you need the system approach of hardware and software coming together. And you’re maintaining a distributed database in hardware versus purely relying on a lot of servers that have to keep track of this database and its entries.

And last but not least which has got to do with the heart of reliability; synchronization. When there are so many changes happening in the network and in the endpoints do you really, are you buying your servers to keep track of all these updates, or are you buying your servers to run your applications? And again what happens when you deliver the packets to the wrong location because you’re synchronization is not reliable. Even in a 1% failure case, right, your error rate goes up tremendously, which are all elements that we get held accountable for from an SLA perspective by our customers, right? I built a network for what? Reliable delivery. If I can’t even assure that, then I don’t have a network as a starting point, let alone SDN.

Erik Suppiger - JMP Securities

All right, I’ll keep going. So when do we start seeing these architectures gain some broader adoption? You say you have 500 customers in the pipeline. That’s probably the most I’ve heard of anybody that’s doing that. It’s a pretty modest penetration at this point. What is the timing in terms of market adoption for this and does this replace -- ultimately does this replace kind of the tiered architectures that we’ve seen in data centers in the past?

Soni Jiandani

So we see SDN as one key market transition point. Another key market transition point for us is the servers are moving towards 10 gigabit Ethernet on the motherboards. So there is an inherent transition happening in the compute market where 70% to 80% of what we have seen through calendar year ’13 will be migrating from selling one gig to 10 gig in the server access space.

When you and because you will always build a hierarchical network, whether it’s a two tier network or a three tier network model, the minute you drive a 1 gig to a 10 gig transition in the server access layer, the backbone will move towards 40 gig. Both of those trends will drive the next refresh in the data center switching marketplace.

Now we have to be cognizant of the fact that almost 65% of all servers out there are running bare metal applications. The world is not 100% virtualized. So the network has to accommodate your Big Data application, has to accommodate your mission critical Oracle SAP applications and has to accommodate your virtualized environments. It has to be a collection of all of that. So that network opportunity for Cisco translates into a transition that is happening in spite of SDN or the rate of SDN adoption, of migration from 1 gig E to 10 gig E in the server access layer and the compounded growth rates and the average selling price increases associated with that, alongside with a transition from 10 gig to 40 gig and possibly also 100 gig in certain cores of the network, right; allowing for an overall ASP trend which is heading north of where it has been.

And as you can see, the lower chart that you see up there, by far the growth is projected to happen in the 10 gig space in the 40 gig market in the next two to three calendar years, right? Now if you add on top of that what ACI can do, it would drive an incremental 20% more software attach rate across a percentage of that market that would have done the transition, okay. So those are the two biggest elements that we see; one, that will happen in spite of the rate of SDN option; and another, that future proofs that infrastructure and provides on top of that an incremental 20% revenue based on the attached rates you can drive to that underlying infrastructure refresh.

Erik Suppiger - JMP Securities

Okay. We are open to questions. I’ll keep going but certainly invite you to feel free to ask. So we’ve talked about kind of the software overlay. Who -- how do you see the switching market? There is the white box under that’s one choice. And then there are other switching vendors that have support for datacenter orientation players like Arista. Let’s start with the white box. How often do you compete with the white box solutions out there?

Soni Jiandani

So we get asked the questions a fair bit, if you look at the white box models of that have been touted out there as being lower OpEx and lower CapEx models. We did exercises by checking in with our customers and we said so what is your overall IT spend? And we got the following numbers, it’s rule of thumb; about 70% to 73% of our customers spend their IT budgets operating their environments and the datacenters and about 27% is going into CapEx, okay. That’s the rule of thumb that most companies are within the range of, plus or minus 5%, right.

And then we ask them that as you think about a model where you can go with a merchant only product, whether it’s coming from a white box labeled company, or a Dell or Arista one of those merchant only vendors; what will you land up building a network and we did analysis with a few of our customers; a modest 2000 node network. And if you had a three year CapEx amortization period, that network will cost you $14 per 10 gig port per month, okay. It’s over a three year -- that’s a CapEx expense, all in, cabling, and everything, all in.

And then we said if you are to overlay an SDN model coming from a virtualization company like VMware, what would your price be that you would be paying them, not over a three year period but indefinitely because these are perpetual licenses. And we got quoted numbers ranging from $5 to upwards of $20 per virtual machine per month. We said wow, so if you’re running 10 virtual machines on a 10 gigabit Ethernet server, and let’s say you say $10 per VM, that is $100 of virtual machine tax that you are paying per 10 gig server, per month perpetually. And we added the two up. And then we rent to our own IT department and said, if you were bandaging two networks, a physical network from a merchant vendor and a virtual network on top, what will your operational expense look like. And they computed a $40 OpEx model for us.

So if you look at this whole price band, that’s about $154 of total expense, CapEx and OpEx, that you’re looking at. We put alongside it what an ACI network with an integrated model of SDN and a traditional network would go for, 10 gig, 2000 port network, three year amortization, all in cost and then we went to Cisco IT and said now you are managing one network with the automation and the ability to drive through the controller automation spanning your security group, your networking teams and your platform engineering teams and the OpEx number came in at $25. There is a 75% difference in the total cost of ownership. Clearly the biggest gap being that virtual machine tax right for building an overlay network completely running in software. So it therefore becomes extremely important to note that when you think about OpEx and a CapEx model, you keep both of those elements in mind when you think about the evolving SDN models.

Now let’s look at the white box. The white box one can argue Soni, the numbers that Deutsche Bank had reported in September of 2013 are dated September 2013 and let’s even assume that the white box sales, not for 2500 with a cumulus software license at $1,000; let’s assume even if it’s sales at $500 or $200 per box, and you land up saving 25% in CapEx. Let’s even go with that assumption. The Judas Group [ph] report that just got published today says most customers are basically saying to save on the CapEx. My total CapEx is 23% of my budget. My total budget, the OpEx and trying to operationalize a network where I’m buying a white box from [indiscernible]. I'm integrating somebody else’s software and I have to go to Open Source to pick up Layer 3protocols, let alone, Layer 2 is yet to be fully flushed and do you know, at today’s date centers have Layer 2 and Layer 3 in them.

The OpEx associated with trying to integrate open source is so much more greater that this gap of CapEx really is not worth it, let alone the risk associated with the instabilities that I would bring into play and I would raise my OpEx and my overall cost of ownership. Not to worry about maintenance windows and multiple software players, the cost of integration, troubleshooting, and this doesn’t buy you SDN. All the other numbers I gave you for that $100 VM tax, I still would have to tag on to the box price to get to an overlay model. This price with cumulus is just talking about Layer 3 software.

Erik Suppiger - JMP Securities

So, you have 500 accounts. How much are you seeing VMware in the market or how do you look at the competitive environment right now, as this market evolves?

Soni Jiandani

We see about 42% of our customers that are already multi-hypervisor in nature. So many of them are starting to have a combination of either Microsoft and bringing in Microsoft hypervisors, given -- alongside with the VMware environment. So we already are in the place where ACI testing, and with the technologies that we’re bringing in are being welcomed by those customers because it's multi-hypervisor in nature. Another key area as I pointed out to you is that our customers are looking for more fine grain control and security, which an ACI network does bring to them, day one.

And they are looking to build a framework of a network, their cloud-based environment, that networking build out, which can put them in a position to drive towards a hybrid model. So the network, the cloud build-out will be with the building blocks of ACI, because it gives them all the fine grain control and security that they’re looking for, including for the PCI compliance and non-PCI compliant applications. If you cannot control security at a system level, you don’t have a next-generation IT infrastructure for cloud-based environments. They go hand in hand with each other.

And we’re starting to also see our customers deploying the 9K as a classic best-of-breed 10 gig - 40 gig switch router for their classic networking needs; whether they are cloud providers and they have a develops integration model where they want to manage it, like through the Linux development community. They want to manage their networks and their servers in a consistent manner. Or whether they are large customers that want to put it in a mission critical part of their network, to augment their existing Nexus investments, or whether they’re looking as a Cat6k refresh cycle. So we are seeing a broad base of customers on both the 9K and ACI for us. I think there was a question in that end of the row.

Unidentified Analyst

[Indiscernible].

Soni Jiandani

Yes. Our approach from an ecosystem point of view is to have a very broad base of ecosystem partners across the security band, across the applications, delivery controller band, across the multi-hypervisor situation, and across the system management cloud management space, including the application profiles and templates in collaborating with SAP, collaborating with Oracle, collaborating with Microsoft; all the key application companies. So because we have a well published API that allows the grid Es, the ecosystem partners to integrate into this ACI model, without requiring any changes to the application space, any changes to the service space, to the storage space; and we have the ability to try our customers' existing Layers 4 through 7 implementations of security, we are building a best of breed network that allows you to tie in your existing best practices in a very seamless manner into this policy model.

So we are not necessarily going to go into the market and brute force anyone particular technology. Our philosophy is a very open methodology, where the open ecosystem is what will move this industry forward together, with us on ACI. It will not be one only working with another. With these well published APIs, these northbound APIs, we are making it very easy, because now everything is object model for us and they are all accessible as a RESTful API.

So what my GUI developers on the controller have access to, every customer has access to through these RESTful APIs. And we are also taking it a step further of -- all the extensions we are bringing into this space are going to be productized, as the production will be OpenStack extension of Neutron that will ship with the APIC controller. So the customer has a choice; they either use their existing workflows and their existing tools that easily integrate into the ACI framework or if they are banking on OpenStack, then they can utilize all the extensions that we will be shipping into Neutron as a production worthy extension on every controller. Okay?

Erik Suppiger - JMP Securities

With those 500 accounts, I'm curious how of them are Cisco using Cisco security infrastructure versus ecosystem security infrastructure?

Soni Jiandani

I would say that the mix is about a third and two-thirds. A third of them would be with Cisco and two-thirds of them will be non-Cisco.

Erik Suppiger - JMP Securities

Okay. So we talked a lot about technology. Cisco obviously leverages a great reseller channel, a very large certified IT professional community.

Soni Jiandani

Yes.

Erik Suppiger - JMP Securities

How are you leveraging those to get broader deployment of the 9000 and ACI in general?

Soni Jiandani

Yes, it’s a very good question. This is something that we typically say the least about. If you take a look at the trend over the last decade, almost 11 years now, the total number of devices that are connected out there are quickly approaching around 25 billion this year. At the same time the number of certified engineers in the networking business are approaching 5 million people. 5 million people that are certified to operate networks, right.

Now, if you look at the responsibility of a networking person, they continue to grow. It has evolved in the last decade to move from going beyond speeds and users to thinking about how do I do threat mitigation, how do I look at Big Data trends, how do I move towards the cloud model, how do I embrace the internet of things?

These are the level of certifications that we have -- and education programs that we are continuing to evolve across that base of 5 million users and they have skills that go beyond the network. They span compute with UCS and SANs with MDS. They span security with a security portfolio and they span office station tools and controller technology, and hypervisors and analytics tools, right.

So these are people that have been certifications now around systems, not just around a networking component, because the network is the platform through which you are delivering all of the services. So, therefore you have to have those skills in your portfolio to meet the needs of the different IT organizations that you serve, right, out of the networking teams.

At the same time, you have to think about a network as a system when you are designing it and you are deploying it and you are operationalizing it. So you have to think about all of those practices and then you have to go back and optimize it. How am I optimizing this to run voice over IP effectively? How am I optimizing it for big data? How am I optimizing it with more security? How am I optimizing it for multicast and high frequency trading applications? So it requires a level of knowledge and sophistication that you're constantly evolving, it’s not the static thing. How do I innovate on top of this network now to carry authentication with internet of everything coming on.

So, it’s a constant skill that I am evolving and I'm not statically sitting in one place and managing boxes anymore. On the other hand, our broad channel base has been very sophisticated and has played a huge role in transitioning us into markets like UCS with 30,000 customers. We wouldn’t get there without our channels. Our channels team is now less focused on helpdesk support and network operations and remote monitoring and is now busy building practices around architectural designs. How do I bring in cloud integration services? How do I start to build converged infrastructure with FlexPod and Vblock, how do I make sure that the networking programming skills are accessible to different buying centers, right, within the compute teams, within the networking teams. How do I drive analytics and consumption models for those customers, right? How do I have the ability to come up with more, how to innovate your IT organization, how to automate across your IT organization, how do you break the silos. So they are more consultative practices, right, around IT services. So, we are going to be bringing to bare that 5 million base of certified people and we will continue to evolve their skills to get them ready for ACI.

Erik Suppiger - JMP Securities

Is that 5 million Cisco certified?

Soni Jiandani

They are 5 million through our academy. So we launched an academy almost two decades ago and through that academy we have over 5 million certified engineers out there. And those skills that I talked about are skills that we have been building as part of our certification programs and that then gets propagated through the academy to the whole audience.

Erik Suppiger - JMP Securities

Okay. So, in the past I heard you talked about how Cisco being deployed in nine of the top 10 datacenters throughout the world, nine of the top 10 largest datacenters.

Soni Jiandani

Cloud providers.

Erik Suppiger - JMP Securities

Cloud providers. How do you look at those? And in terms of the customers for ACI what kind of penetration would you expect for Nexus 9000 or ACI amongst those top data centers.

Soni Jiandani

So it’s an absolute accurate statement, that -- what you just, which has 90% of all the largest cloud providers are powered by the Nexus. If you look at what these cloud providers are doing now is that they are moving their infrastructure towards 40 gig in the aggregation and are also looking at 100 gig technologies to drive the connectivity across their data centers. And there we are well positioned with the Nexus 9K. The reason we’re well positioned is because not only does the 9K deliver wire speed routing of 40 gig and is 100 gig capable without running into headroom from a power and in terms of future proof design; but using our architecture, when you contrast it to traditional merchant only architectures, we draw 15% to 20% less power. And that’s a very important metric for these cloud providers. They not only are looking for a highly reliable non-blocking router which has the robust architectures to move towards 100 gig, they also want something that can draw a lot less power for them.

And this strategy, along with the notion of embracing an open API mode, a highly programmatic interface that allows you to decide whether you want to use Puppet or Chef or whether you want to use JSON from an operational model perspective. Running on top of our stable protocols like BGP, OSPF continue to support us in a very robust position with these cloud providers going forward and given that they have tens and thousands of the Nexus products that are deployed, the consistency of the operational model and the speed at which we can drive software velocity with them is a continued proof point that as long as we meet their criteria, they will continue to want to do business with us.

So it’s extremely important to note that this is -- while this is a crucially important market for us, it’s not the exclusive market we play in. There's no particular quarter in which there have been more than 5% or 10% of our overall numbers in the last six consecutive quarters. But what’s important to note is that when you can have networks of this size, you can pretty much fit any network of any other size. It takes a lot to keep networks with million plus nodes running of this nature.

Unidentified Analyst

[Indiscernible] What is your expectation with the adoption, do you think there will be customers adopt [indiscernible] and what you think happened there?

Soni Jiandani

We -- the way I think it would play out is that there would be certain environments where there is going to proof of concepts. Everything starts with a proof of concept, any new disruptive technology. But to make the statement that it will scale to meet their data center needs is a bit of a stretch of an imagination because there's no one data center where everything is 100% virtualized with VMware. Any customer that has been around for five years, 10 years has applications that are both physical and virtual.

And the other key important thing to note is that the economics of OpEx and CapEx, where 70% plus is spent operationalizing and 30% is spent on CapEx; it’s not going to change dramatically that equation. And if you pay attention to where majority of the emphasis will be in our industry, with the promises being made in SDN as an industry, it would be towards OpEx. It would not be towards the CapEx reduction.

And from an operational expense point of view, you want to make sure that you have the subject matter experts for networking and computing. You have to retain subject matter experts. You can’t say, if you are a networking person I don’t need you any more tomorrow, somebody who knows server virtualization will run my network. You still have a network, right? You still need to retain that staff, correct? So a lot of emphasis will be around how do I move with agility? How do I cut down my costs of operations, while innovating and move with agility and deliver a service back to my corporation. And we feel that not having an architecture extensible to the application and falling short of thinking it through at a system level will put them -- it’s an architectural challenge for them. It’s an architectural and human resource challenge for them.

Unidentified Analyst

[Indiscernible].

Soni Jiandani

As I said that, you always start with a POD [ph] in any new technology. We do not come across a lot of customers that are saying I am in production.

Erik Suppiger - JMP Securities

Okay. We’re going to have to wrap it up there, again, thank you very much for joining us. This is a major transition in the industry so do appreciate your sitting in and joining us for what was a very insightful and delightful tutorial.

Soni Jiandani

Thank you.

Erik Suppiger - JMP Securities

Thank you.

Copyright policy: All transcripts on this site are the copyright of Seeking Alpha. However, we view them as an important resource for bloggers and journalists, and are excited to contribute to the democratization of financial information on the Internet. (Until now investors have had to pay thousands of dollars in subscription fees for transcripts.) So our reproduction policy is as follows: You may quote up to 400 words of any transcript on the condition that you attribute the transcript to Seeking Alpha and either link to the original transcript or to www.SeekingAlpha.com. All other use is prohibited.

THE INFORMATION CONTAINED HERE IS A TEXTUAL REPRESENTATION OF THE APPLICABLE COMPANY'S CONFERENCE CALL, CONFERENCE PRESENTATION OR OTHER AUDIO PRESENTATION, AND WHILE EFFORTS ARE MADE TO PROVIDE AN ACCURATE TRANSCRIPTION, THERE MAY BE MATERIAL ERRORS, OMISSIONS, OR INACCURACIES IN THE REPORTING OF THE SUBSTANCE OF THE AUDIO PRESENTATIONS. IN NO WAY DOES SEEKING ALPHA ASSUME ANY RESPONSIBILITY FOR ANY INVESTMENT OR OTHER DECISIONS MADE BASED UPON THE INFORMATION PROVIDED ON THIS WEB SITE OR IN ANY TRANSCRIPT. USERS ARE ADVISED TO REVIEW THE APPLICABLE COMPANY'S AUDIO PRESENTATION ITSELF AND THE APPLICABLE COMPANY'S SEC FILINGS BEFORE MAKING ANY INVESTMENT OR OTHER DECISIONS.

If you have any additional questions about our online transcripts, please contact us at: transcripts@seekingalpha.com. Thank you!

Source: Cisco Systems' Management Presents at the JMP Securities' Technology Research Conference (Transcript)
This Transcript
All Transcripts