 Actually, why don't you come up? You can come up with me and then, OK. So we can get started. I know everybody has a tight schedule, so we'll try to stay on schedule. Just a bit of introductions. My name is Adnan. I work for Dell, and I run the global strategy for the networking BU within Dell, which is a pretty fancy title, which means I just get to go on airplanes to all the fun cities in the world and talk to people. So we have, by the way, Andy here as well. Andy runs our EMEA networking team as well. So I think if any questions, comments, suggestions, we can probably answer those things. So the reason why we come and do this is to give you guys an idea about what we are doing to help change the world, and also get your input and feedback on what you guys think about where we are going, as far as the industry is concerned, as far as some of the things that you guys are doing. So please feel free to stop me, interrupt. I mean, this is a great opportunity for us to learn, and that is, quite honestly, my incentive to come to these forums. It's not for this to be a one-way thing, but essentially engage with you guys. And I'm around. I should have put my email address up there, but if you need to talk in detail, feel free to grab me, and I'll give you my card, and I'm at the summit until Wednesday, so either me or Andy would be happy to have any conversation. So just to give you, get an idea about the folks, how many people here touch networking on a daily basis? OK. Partners, vendors, versus people that actually deploy. So people who deploy and manage networks? OK, fairly good group. OK, so what is open networking? Let's start with that. I think when we were looking at the industry, this word SDN, I think it's the second most abused word in the industry after cloud. And at Dell, we were trying to figure out what is the next big thing as far as networking is concerned. And we didn't have to go back. If you look at what happened on the server side, it gave us the next logical transition in the networking space. So what happened on the server side, if you look at it, the industry moved from the mainframe model. I mean, we used to buy vertically integrated systems, either from our friends at Sun or other vendors. I'm going to try and not take vendor names because I can get in trouble and I have my own opinion. But what happened was if you look at how the application ecosystem evolved, it was because we moved to standard architectures from a chipset standpoint on compute. We gave customers and partners the ability to run their choice of operating system on the compute platform, which actually served as an application host, and then the application ecosystem evolved. And today, as operators or as people that deploy infrastructures, you're forced by the application requirements which are driven by the business requirements to go and deploy an x86-based infrastructure. How many people here still buy and deploy mainframes? How many people work with x86-based architectures? So it happened on the compute side. And then Dell was very integral in that part of the industry. And I think what happens is that is exactly where we think the networking industry is going. And what we did was essentially this. So it's not overly complicated. This is our vision for software-defined networking. And what it means is that we basically started down this journey a few years ago by moving to merchant silicon, open hardware. And then now we have given the customers the ability to run different operating systems on our switching platforms. Because if you look at this, this is exactly what happened on the compute side. And the reason is that every infrastructure has different requirements. So why should we be making the decision as to what operating system you need to deploy? And if you go back and you look at how networking works today, you go, you buy proprietary ASICs. I mean, you basically buy the box from either us or our friends across in the industry. Then you have completely integrated, tightly controlled protocols that run. And the only way you can do SDN is either through OpenFlow or API access or programmability or things that people offer you. But if you go down this direction, which is what we have enabled now, we announced two partnerships. One was Cumulus, which is a complete Linux-based operating system. How many people here are familiar with Cumulus? They actually have a booth down there. So go talk to those guys. It's essentially a full Linux OS. So when you do deploy it, it brings up a Linux batch prompt. So the idea is that why is networking so different than compute? So for that, you have to give people the ability to run standard hardware on the networking devices like they do on the server side. But an operating system that can actually serve as an application host. So tomorrow, if you want to go write code that you want to run on switch, because you think you can do things in a better way, you should have that flexibility. You want to run standard L2, L3 protocols? That will work as well. So it gives you a transition path. And that is what we believe is the two definition of SDN, because the first word in SDN is S, which stands for? Exactly. Second partnership we announced was with Big Switch. Essentially, they're tapping solution, because we believe it's a very non-intrusive way to come and give people the ability to run monitoring on their infrastructure. And now we're going to partner with them on the cloud fabric solution as well. But the idea is, from a high level strategy standpoint, this is where we are going. It's not overly complicated. And you guys have the ability to do what you want to do. And that's how Dell networking, we believe, is different than anybody else in the industry. We, by the way, have our own OS. We are continuing to invest in it. But we want you guys to make a choice. Now, if you look at it, what does this do? It actually enables your ecosystem. You, as customers, as partners, as vendors, can put your solution together and bring a solution to the market, which is not what is defined by either us or our competitors. So if you look at this, standard A6, our OS, we are going to keep coming out with other choice of operating systems, overlay networks. And then you can do monitoring. Since it's a Linux operating system, you can actually leverage all of Linux's ecosystem out there to deploy any kind of Linux packages that you want. Now, just because, please do remember, Switch does not have the same compute power as a server. So with great power comes great responsibility. So you need to be careful as to what you run at that layer. But essentially, this unlocks a complete ecosystem for you guys to look at your infrastructure in a truly agile way. Second thing is the way you deploy, how many people here run L2 in their networks? Layer two, layer three? So you can still deploy it like that. And when you're ready to move to this direction, you can transition along with Dell. And that's how I believe that Dell is very different than anybody else in the industry. And you are going to see other innovations come up here from our side. And then we are right now working on these ecosystem partners. Basically, with us, you can deploy an end-to-end solution either with VMware or Mitikora. So that's how we are putting it all together compared to a lot of other friends in the industry which are actually still making it a vertically integrated model. So everybody, do you guys think this is a good direction? Do you think this is we're wrong? It would be good to get some feedback. Questions, comments? Okay. So how does this work in this thing? So if you look at, now let's go up and look at the data center. So what is happening in the data center is essentially there is a disaggregation happening across software-defined compute, happened with virtualization, software-defined networking is happening and software-defined storage is the next thing which is coming about. Essentially, the idea is to disaggregate your hardware from the app OS and from the application layer. Essentially, you should build agile infrastructures irrespective of the application that you're running and your infrastructure should be able to serve any application dynamically and adapt. So if you think about this, happened with compute, server virtualization, right? I think everybody here runs virtual infrastructures, I'm assuming. So compute was the first thing that got virtualized. Essentially, one server can become 10 servers, right? And this is where we are taking this. I'm not gonna talk about storage because this is a networking session, but this is what we are doing. Essentially, we are basically going to abstract the physical hardware from the OS and then the application. We think that there is no reason for you to get on the CLI and start making commands. I mean, this is where things are going. So the point is that with us or with Dell, you actually have the ability to go down this journey. Some of our friends across in the industry, it's either a rip and replace because you buy one thing from them or you deploy their infrastructure, it has to be completely vertically integrated. Support, their choice, their OS, their hardware. You can only go down this road if you actually break the model up. And we are giving you the ability to do so at your pace when you're ready because nobody's going to deploy a full SDN infrastructure tomorrow morning. It is all the capital investments that you make. You can slowly start moving in that direction. But this is essentially our thinking as to where we are doing. So compute happened, networking is happening now. We actually announced a partnership with Nutonix on the storage side. I don't know how many people here are familiar with Nutonix. So we have started going down that road on the storage side as well. And that's where we are going with this. So how does this tie into OpenStack? OpenStack is a cloud OS. It actually gives you the ability to manage infrastructure. So you can pick the right OS for you. We'll get into some of those examples, whichever one suits your needs and your application requirements and quite honestly your tastes. At some point it becomes a choice of religion as well. I mean, some people are completely interested in running L2 and don't want to move away from L2. Some people have moved to L3. And I mean, I've had many religious debates with customers across the world on this thing and these are perspectives. So the idea here is that if you have a topology that is running L2 and that is your direction, you have that choice. If you want to go L3, you have that choice. You want big switch. You want to deploy a full STN fabric that is completely controlled by an STN controller. You have that ability. And you can move around, by the way. Can you deploy a different OS on the server? Can you move from Windows OS to Red Hat? How many people here can do that? If I give you a machine, can you switch OSes on it? Yes? Yes, right? It's just setup.exe. Why can't you do that with networking? So the idea is that today, let's say you start with this, which is a traditional OS, and you run your fabric and your infrastructure in a certain direction. Tomorrow or day after tomorrow, you realize you have applications that can optimize and run your infrastructure in a better way. We give you that flexibility to go do that. That is the direction we are going on compared to a lot of other partners out there. Yeah, question? Common? See the idea? We have this huge, like, constraint, but they all point to being not the right one. So our stance right now is that we are basically providing the platform. So what you are saying is more on an application optimization space. I hope that somebody picks this up as the platform and actually develops that on top of it. But it does not exist on the roadmap. And the reason is because we have, I mean, to be very frank, we have limited number of resources. So if you think about this, let me step back. The reason why this is interesting is because of the following reason. If you are Amazon, Google, you can go do this yourself, by the way, they have done it with ODMs. How do you bring a technology in mainstream enterprise? The only way to do that is if a customer has the ability to pick up the phone and say, I don't need 10,000 switches, I just need two because that is what my infrastructure is. And we are giving them that ability. And if they have a problem, they pick up the phone and call Dell. If they are based in Europe, they pick up the phone and call Andy and scream and shout at him. That's how we, so we are actually bringing this to the mainstream enterprise. Now with that, the first strategy is that all of this requires a lot of resources to certify and get it all working. And that is where we are interested in. On the compute side, we did not take a stance on the layer above the platform. If you look at what Dell did. On the networking side, we are thinking about it. Let me put it that way. But we do think that that problem needs to be solved by the data layer, by the application layer, not in the networking space. Because if you look at it, when software defined storage comes in, I mean if you take Newtonics as an example, are you familiar with Newtonics? Anybody? How many people here are familiar with Newtonics? So I'm gonna use that. So what Newtonics essentially is, because a lot of people are not, is essentially giving you a sand capability on compute. So it's software defined storage. What you do is you have direct attached storage on the server. So it runs the application or the software on it that gives you the ability to replicate your data across so it actually acts like a sand. Now, your storage is actually running on a server which can die any day. So the software has the intelligence to provide. Now, when you look at that concept, it is the software defined storage controller that has the ability to decide where it's gonna move the data. So the idea is by providing a platform in the networking OS, they have the ability to go program the network the way they want to. Right, so that is the reason why we have sort of not crossed that boundary. The other thing is, if you look at it, I mean one of the things we'll talk about is if you build a full class network with lower subscription and the price points on the 40 gig are going down, you, I mean network should be a highway. I mean essentially it's a freeway. So if you bet you should be fine from that standpoint. So point taken, very well valid point, but right now it's not on the roadmap and these are some of the reasons why we have sort of walked that fine line. I mean I get these ideas from my own team that we should go tweak this, we should go tweak this, but I mean right now we are in the business of providing the platform. And let you guys decide how you wanna go do that. So for your health within that one? Yes, so it's integrated in our health reportion. So that's how you provide the overlay. So, yes. So try, basically our moving forward, all of our next generation processors are gonna be x86. The reason being that if you deploy, if you write a monitoring application, for example, right? It's compiled on a server on Linux, it's x86. So it should just run on the switch because it's x86 processor. And the second is we are moving down the Broadcom chipset road. So our platforms are all Broadcom and tried in two supports VX LAN capabilities in the hardware. So it's a hardware feature that needed to be there for us to go support that. So, and then just an additional comment. I mean, basically if you look at on the compute side, right? If you look at VMware that has the intelligence to go do VMotion and all those things, essentially the intelligence is really application. And so that's where we are going. But we do understand that networking is not as trivial. It's a little more complicated. So we need to maybe go make some tweaks and the CTO team actually is working on some of those things. I don't know if Sanjay is in the room, he was supposed to be here, he's from the CTO office, maybe he'll come. So if you look at, we are giving you guys the ability to deploy these open standards and platform-based solutions in OpenStack as well. So if you are an L2 customer, I mean, this is fairly straightforward. You can deploy an L2 domain. The difference between us, I mean L2 is same across the board for most vendors, but the difference is with us you have the flexibility to move in the direction that you want to move in. So it actually completely integrates with your existing L2 networks. There is no rocket science as L2. Personally, I'm not an L2 fan. I don't know how many people here are L2 fans. I don't mean to offend anybody. I think L3 is a much cleaner solution and less issues when you mess up the cabling or something like that. But you can deploy your infrastructure the way you deploy it. You can do subnets, you can do service isolation, all those things the way you do L2 networks today. So it actually completely integrates with your deployment. So that's cumulus plus Dell. I mean, I wanted to use this as an example as to how a hardware from Dell and software from somebody can actually tie into an overall solution, which is the first in the industry by the way. You can't do that with our friends at HP, our friends at Cisco. I hope they're not in the room or if they are, I'm sorry. I mean, you don't have the ability to do that. You can't go and tell these guys that I like your switch, I have bought your switch, but I want the flexibility to run a different OS because it is the direction that I wanna go in. So essentially breaking the model. And I think this is something you were referring to. So full L3 capability. I think the reason why I like this is because you get load balancing via ECMP for free. I mean, it's an easy way to load balance across your network. And you can change the hash algorithms depending on how you want to do load balancing within your infrastructure, but using VXLan, you can actually go and deploy this. The reason why this is, we have the chipset name up there is because it's a hardware capability that Broadcom provides. So that's what has to get exposed to basically go support that. So essentially you have this infrastructure and then there are solutions out there that use the overlay technology on top of it and then you can tie those off on top. But again, I wish I had a complicated solution like a lot of people out there. What we are doing is extremely simple. The only difference is we are giving you guys the ability to go and deploy solutions the way you want to. So I showed you an example of an L2 fabric. So if you're an L2 base, it still works. It actually augments to your existing network so there is nothing you have to go change. L3 is much more standardized. So it integrates with your L3 fabric. And if you look at this, this is the third solution that you can deploy today which is with Big Switch. How many people know about the cloud fabric from Big Switch? One, two, okay. So essentially there's no L2, L3 protocols. It's SDN controller. So what they've done is you basically run a controller on a server and that actually figures out how to go route traffic. You can run it on Dell switching platform and you can basically deploy an open stack solution using that. I think it's a very interesting idea. This is where the industry is going. The difference is how it talks to the outside world today is using static routes and we are working with them on dynamic routing because that could be an issue because if you change it out on the outside it has to get propagated within the fabric. But this is not running layer two. This is not running layer three. This is actually, this is just the data plane and the decision on how to flow the packets is actually being made outside, which is SDN controller, yes. But they do have proprietary stuff that they do on top of it because if you look at it, you have to do a lot of things. You have to do load balancing. You have to basically do routing. So open flow is what they use and in a standardized way on their big tap solution which is the monitoring part that replaces the giga mon. But this is using open flow but they've done tweaks, proprietary tweaks that actually make it more optimized on that. So the reason why, let me, sorry, answer that question in more detail. The reason why we did this was the following. If you look at traditional open flow deployments from people, how many people here know open flow? I just want to know, okay. So everybody did open flow because it was a standard, it was the only way to do SDN and good luck trying to get people to agree on a standard and then implement the standard. Open flow was deployed in a way where we used to carve out tables, right? We will keep this much for regularly L2, L3 and we'll carve out a portion of the hardware to basically store the open flow data. The problem we ran into that was that it wasn't scalable. So the reason why we gave Big Switch the ability to have complete control because then they can actually control all of the tried-and-do hardware table sizes. So all the, essentially open flow has the access to complete 16K entries or whatever you want to do on the chip set, whatever the chip set does. So this is using open flow. Essentially that is the standard they follow. They're all there, they're a way Switch Lite actually supports open flow so you can technically tie it to another open flow controller. But for the SDN fabric, they do recommend using their own controller because they do some proprietary fancy stuff to get this working in case of failures, in case of load balancing and things like that. And I think they're also here. I don't know if Pashant is here, their CD is a very, very smart guy, but drop me an email and happy to have a deeper conversation on that. This is completely, yes, simplicity. Can you say anything before? Yes. Within the table? Yes. See, overall, you can. You can, it depends. I mean, at the end of the day, give me an example of an overlay technology that you would use. You can, that's what I said. It is not for, that's, so I'm not. Give me a switch. It's very hard to promise. But the day is a very, very difficult. It's very, very difficult. And then you say, okay, let's go back. Two things in the other, but it's not really an open hardware. It was open, it was a professional made, but then it's the cost was high and the performance was very low. So, then it comes nothing, how do we make these results? Why not make a hardware? It's a cheap, three-core, about support. They're like 10 days. So let me let me let me give you my Dell answer as when I put my Dell hat on and then I love giving my opinions and I'll give you my opinion which is nothing to do with that the fact that both two people here are having a conversation on two different preferences is exactly what we wanted to enable as a company number one second thing is what you are talking about is right here if it works for you great the problem is it depends on your scale if you are going to deploy a 500 port 10 gig chassis then you're sorry what can you give me okay then you can actually deploy deploy a class on the spine and drive the cars down even even further so the latency is in latency is the thing so eventually when you run a data center so let's this is maybe not a maybe doesn't work for people that have an enterprise infrastructure with 50 racks I don't know it's a preference issue at some point people have a preference but if you are a cloud provider that runs 50,000 servers in a physical servers in a data center I personally have seen data centers that have 200,000 servers in one physical location at that point you want a lot more granular control and and run things that have been a very simple way because a route getting a route route flapping happening or happening on BGP can take your control plane down because they have limited processing power right so it depends on the scale this solution if somebody wants to go deploy it is for people that actually want granular control they can do a lot of these things and they can plug into the SDN controller and have more granular control over their fabric for example they can go and say I won't when this happens I want the flows to go this way they it gives you that level of ability to go control your infrastructure essentially it's not I mean vxlan overlay technologies more enterprise play to me if you're running a big infrastructure on the cloud side this could be a good play for that because it gives you a lot more granular controller essentially using the controller you can have the chipset access if you want to do something funky let's say if you want to go touch the the fib on the switch right for whatever reason this gives you that that level of control but the point is from a Dell standpoint we actually wanted to go down this road because we see customers that why is it that one size fits off I mean I for me I always struggle with that and if you're deploying an infrastructure with two racks with ten racks with five thousand racks your requirements and your business needs are very very different as a company I was shoving down one solution down your throat here I'm giving you the flexibility to go down whichever road you want if you're running a thousand racks and you want to go run at that level of control power to you if you want to go run vxlan power to you but me as infrastructure provider sorry Dell not me as infrastructure provider is going to give you the flexibility to do that and transition so let's say you deploy that infrastructure with vxlan and you run into scalability issues because you just got too big or you're doing something crazy I give you the flexibility sorry Dell gives you the flexibility to move from that to this and see what works you know the vxlan is trying to supports vxlan so moving forward all the broadcast basics are on the road map going to support vxlan I get that so we when we when we go to lease a platform we are very cognizant of that if there is an issue like that we will come work with that customer very specifically if there is an issue but the idea is when we move forward we are going to going to give you the ability to switch which means if there is an issue like that some it's not a very at the end of the day when the chipset gets released there are there are updates that come in the STK usually not in the chip respin which gives you the flexibility the next chipset from Broadcom is is Tomahawk right which is the 100 gig platform so there's nothing coming between now and then so it's pretty stable till then but then you're moving to 100 gig now I can't put a 100 gig port in a 40 in a 40 gig chip that's that that is a limitation but today you can move between these two solutions on Dell hardware because again depending on getting on the details chips have revs depending on which rev you latch on to you can you can have that issue so we we are cognizant of that and that is exactly where Dell comes in right I mean if you have an issue like that you can pick up the phone and call Dell that you committed this to me it needs to work compared to going to a white box or or an odm provide but I think I mean I'm interested in in some of the things you talked about so maybe we can catch some one-on-one because I think it's it's very I mean I can I can give you personal examples of a lot of things where we actually ran into scalability issues it's just the size of the infrastructure because no matter what I have it depends on the scale if you start hitting your control plane data plane is not the problem because it's line rate across the board it's when you start hitting control plane that's where the problem comes and that comes at scale and scale is is a very relative term right and then we have our own solution which is Dell IP which is Dell active fabric controller we have a neutron plugin and it's a complete out of box solution end to end Dell so what just to recap what we are doing for STN is the following we are giving customers the ability to choose the right OS that works for them either it can serve as an application host and they can move same thing happened in compute and it I think was pretty successful and we believe this is going to happen to networking and it is happening if you look at the growth numbers the fastest growth in networking hardware was within the odm because people went to odm just because they didn't have anybody else doing this so giving that choice and then one size doesn't fit all if you are an l2 customer base we have a solution that works with the l2 fabric if you are an l3 with overlay it works there if you want to go STN with big switch we support that and moving forward we have our own IP solution as well end to end Dell if you want if you don't want to go through the trouble of putting it all together and you want to pick up the phone and say I want to get this you have a solution from Dell as well so that's the that's the direction we are going I mean just to how many people here are familiar with MPLS okay let me ask a question from my field what do you think about giving an MPLS stack on a 32 port 40 gig switch and you looked up okay why okay that's it's very different you know anybody else thoughts okay so any questions comments pretty straightforward yeah I think it's a so we we were also looking at that solution there was an Intel chip that was working as well I think I know so so the idea here is the one I mean we do have a we did we are going down that road I don't know if you're familiar with our MXL platform so on on Dell Blade servers we actually have a have a switch a 40 gig switch that goes straight into the back of the back of the chassis that actually provides that capability and then I mean you can actually do 40 gig straight up from there so if you are deploying HPC and you you see I you can I cannot compete with I just cannot compete with infinite band latency I cannot compete with it but if you if depending on how I mean when you look at HPC deployments that it is a function of what your threshold is right what your threshold for latent if the Ethernet falls within the threshold of that then then it's a it's a pretty good solution to go look at at the end of the day I mean a lot of that decision to go to other vendors is driven by cost I mean it's a it's a cost cost decision at that point and and and you have to put it all together so we have a solution we're going down that road we had the first release of MXL I mean it's it's doing great essentially blade server plug it in and you can put multiple I think you can do six right Annie you can do six you can stack up to six what yes so the so we we do two things so there we have a feature we implemented and I don't I can I can get the details I can get your card and Ryan I can send those to you the actual limitation actually comes from the hardware so since we are using the same chipset we have the same hardware limitation the second thing is we do have a feature where we do ACLs on the software to basically extend we carve out and and I don't know if they've told you they have the same feature as well so so you can you can do that my personal opinion now is I mean we have customers I mean ACLs is not a very scalable solution it's very hard to maintain so I yeah it's the question is I don't know I mean you're using it for ice traffic isolation what are you using it for and you control the application stack on the server the users okay then you have then you have no other I mean one of the ideas I have is if you don't I mean this is just I mean we haven't done this if you deploy cumulus no I'm just saying if you look at cumulus you can write a software app Linux app on top deploy it that actually does it for you so instead of instead of saying you can dynamically actually do deny all right when you give it to somebody else so let's say this is where you're running and this is how we believe we are much cooler than the guys that you're talking about I think they they went halfway they didn't they didn't go all the way all the way although their hardware designs on 140 gig are pretty interesting but our philosophy there is small boxes big boxes when they die they take half the infrastructure down with you so we have a 132 port 40 gig in three are you because we believe that you can build a class with that so just just so what you do is if you run cumulus right here right you can actually run a Linux you can write a Linux 50 lines of code that does deny all and allow one and you can change it dynamically when you give your infrastructure to somebody else that I mean it's a creative way to go do it but that's one way you can get me on the ice so because the thing is you program your ACLs once and you don't want to touch them because it's a one to it's a big ACL file right but but that is a good a cleaner fun solution so yeah I think Mina is here yeah Mina is here from cumulus downstairs a few if you have time just just tell her Adnan when I talk I mean she she's a she's a pretty smart person so she could you can actually chat to her about on doing some of these things so okay okay absolutely absolutely no that option what do you mean there are you validating designs with those you are you're you're you're this is a very valid question if you look at the I'll start from the give you a few facts if you look at the one gig decline in price per port the graph was like this okay there was there was a time when people used to pay 30 grand for a one gig switch by the way the 10 gig decline in the industry was like this the 40 gig decline is like this so that tells you one thing that this whole notion of going and charging you $50,000 for a top of rack those days are going away number one customers are getting smart the second thing is customers want a choice if you look at the odm growth you know what an odm is you know they're downstairs quanta and some of our friends there they grew 23 to 43 percent last year in depending on the region you talk about which means that there are customers that want that ability but they have to put it all together now what is del's value in this number one if you have a problem who do you pick up and call and who's going to support the solution number one number two if you are a company if you are amazon or a big cloud provider that buys 50,000 switches a year you can probably go set up your own supply chain that is an operational cost by the way but if you are anybody smaller than that you have you don't have to go set up your own supply chain and you benefit from del's procurement power and the del's solutions team that puts all this together for you and brings it to the market but i wish to god i could come here and say it's overly complicated guys it's not this is happening to networking i was i was at open stack in atlanta and i did not have one customer win because we just announced the solution i think paris is five months later i don't remember when it land i think atlanta was in march if i remember march april we've grown our business by multiple hundreds of thousands of dollars in the last few months just because you can pick up the phone and say listen i want this solution send it to my doorstep i don't want to bother it and this is the tco model that i'm going after so my question to all of you is since it's a del sponsored session with the flexibility if anybody buys a different vendor please come and tell me what is it that we are doing wrong because i really want to learn i you want to know i mean i think we are with all the options the flexibility is worth protecting an investment enough any i know we got into some conversations anybody else has any questions comments in the back okay thank you guys enjoy your day have fun