 And now, SiliconANGLE TV and wikibond.org, present a focus spotlight. Live from Las Vegas at VMworld 2011, host John Furrier and Dave Vellante, a human being. New models for cloud service providers with support from SolidFire. Alright we're back. This is Dave Vellante. We're live from VMworld 2011, SiliconANGLE's continuous coverage. I'm Dave Vellante at wikibond.org and we've got two great guests here. We're going to talk about the cloud services business, how that's changing. I'm here with Nathan Day, who's the chief scientist and Duke Skarda, who's the CTO of SoftLayer. Very interesting company. Want to hear more about it? Welcome gentlemen. Thanks Dave. Thank you. Start off, tell me a little bit about SoftLayer, who you guys are, history of the company. Take us through that. Sure. So SoftLayer is the infrastructure to service company. We exist mainly to give our customers the ability to get our infrastructure in the terms of cloud computing, regular computing, network appliances, things of that. Trying to deliver self-provision, self-managed compute solutions that live in our data centers. We started in 2006 and we've grown over the years. We've got data centers now in Dallas, Seattle, Washington DC, San Jose and we're actually about to open a data center in Singapore and one in Amsterdam too. So are you compete with the big whale cloud service providers or Amazon or yes? Absolutely. We provide a little different service than you find from Amazon or Rackspace. We kind of fit in the middle of those two guys. Amazon is great at the virtual machine infrastructure to service. Rackspace is great at the managed, dedicated hosting. We kind of fall in the middle where we have the virtualized services but also the dedicated services as well in a self-managed environment. So we fit nicely between those two extremes. So you're a relatively new company. What was the idea behind the company? What was the sort of secret sauce that you were trying to bring to the marketplace? Yeah, well, SoftLayer really began as an outgrowth of a number of people having a lot of experience in the hosting industry and the secret sauce that I would say that was really present from the beginning was automating everything and building on a generic platform. So all of our products and services are really based around an x86 platform. So we try to turn that into as many different products as we can from network appliances, cloud appliances. So all of our dedicated servers and our cloud servers are really based on the same platform. If you're going through our data center, it's rack after rack of x86 boxes that we can turn into just about anything. So what we're doing here at VMworld is we have in the afternoons these in-depth spotlights and you guys, we bring you on as subject matter experts and they're sponsored segments and we help our audience understand, practitioner audience understand the topic, the trends, what's happening. And one of the things that we've talked about is every IT shop that you talk to, it's budgets are flat, we have to do more with less. Now I'm guessing that your business is not flat, it's up and to the right, right? Exactly. We've got great growth going on, we solve the problems you're addressing, basically turn compute into operational expensive capital expense for a lot of people. It's been going great. So the point of that discussion is that what we see is the cloud service provider is actually innovating, they're investing, they're growing faster than the average Joe IT shop, the average Bayer IT shop and as a result, we think that the gap between traditional IT and cloud service providers actually may be widening. But having said that, traditional IT, they got different challenges but they can learn a lot from cloud service providers. So I want to go to school on you guys a little bit. What did you see when you started the company and what do you see now as the cloud evolution? If I can, if I can, everybody talks about the journey, what is that a year journey in your vision that you see? Yeah, I'd say when we first started the company and really dating back to the early 2000s, service providers that were really hosting companies at the time really offered up a dedicated server and it was at the time, very low level, a lot of web services, very sort of simplistic applications on the internet. Over the last several years, a couple of things have really come together to make that a much more powerful platform. The internet itself is so much more robust over the last three years especially. To the house, to our mobile platforms, we have internet access everywhere and it's very powerful. You can get 30 megabits right down to your house. We're all carrying 3G and 4G platforms around with us. That's made access to a centralized platform so much more ubiquitous that things like cloud have really come into their own. So one of the big changes that we've seen that's been brought about by that is the desire to have a cloud service. And there's a couple of different ways you can look at cloud servers. Amazon has really popularized the virtualized cloud model. We look at our dedicated model and we think of that as having been cloud all along. But now what we're trying to bring to the market is the ability to work with dedicated servers in the way that we've all become accustomed to working with those virtualized servers as well. What does the infrastructure look like underneath? I mean you guys, I presume didn't have a ton of leg of the infrastructure, right, a startup. So what's it look like? We basically took an ecstatic server, wrote an automation orchestration engine around it, so we can take any ecstatic server, turn it into any product in our catalog. And one thing that made software different in the beginning was that we built out a dual network for every server, which actually transitioned to every cloud virtual machine as well. That allows our customers to eliminate the problem, reserving racks for growth, because every server from software is connected to each other via private network, as well as your traditional public-facing network. And it gives our customers a lot of flexibility. They don't have to worry about cross-connects. They can have servers in multiple data centers, behind multiple routers, and they still have that back-end connectivity that really allows them to build a scalable application. So here at VMworld, and I remember my first VMworld a few years ago, it hit me right in the face is storage is like the biggest problem that all these practitioners face. And VMware has been working hard in the ecosystem. But where's storage fit in your challenge, pain in the neck pie? It's pretty high up there. Whether you're talking about local attach, direct attach storage, and someone trying to do tricks with SSDs to improve performance, putting in RAID cards, going with a memory model, Fuse and IO, so even a local server, there are still huge storage challenges. And then when you bring in the network attach storage, the SAN storage, all kinds of new challenges present themselves. The workloads are different. The ability to fetch small objects, it really, it permeates every application out there because storage is turning into such a huge bottleneck, not only in terms of capacity, but also in terms of the actual performance. Are you able to deliver consistent quality of service for your customers and actually charge for it? Is storage a gate at doing that? Talk about that a little bit. Well, at the storage layer, there's a couple of, as Nathan just walked through, there's a bunch of different options. Each of those options has different levels of service that we can provide different quality of service. One of the things we're excited about with some of the new products that are coming out now, a lot of them based on SSD and Flash, is they're sort of reinventing the way to build a SAN platform. And one of the things that they're able to build in that we're very excited about is the ability to control at a fairly fine grain quality of service. Today, IOPS are really mapped across all of the users of a given SAN. If we can begin to control that, then we can begin to charge for different levels of service. And more importantly, we can guarantee different levels of service for customers. Yeah, so we've been looking into this. David Floyer, one of our analysts has studied this market very closely, and he's sort of laid out the different approaches, you know, the first one being sort of stack drives, let's say in an EMC array, as sort of supporting legacy systems. And then at the top of the spectrum, if you will, you've got Fusion IO server class memory. We're starting to see these new all Flash arrays emerge. And the value proposition behind them is guaranteed quality of service. What we've done is analysis, stacking up a hybrid SAN array with spinning disk and Flash with these all Flash arrays. We found that the all Flash and a lot of applications is less expensive. Trying to see if you can confirm that. Are you seeing that potentially you can get greater utilization out of these devices or do you not have enough experience with that yet? I don't think we have enough experience yet. Right now, we like what we're hearing. I think the models that we're hearing and the architectures that we're seeing look very promising. We're just now beginning to get our hands on these products so we should know soon. We really, all these new products give us an opportunity to get another access to the product line. So instead of just talking about capacity, we can talk about selling a quality of service and guaranteeing performance at certain levels. That's really exciting. So how do you guarantee quality of service today? You just got to throw more hardware at the problem? Is that right? Absolutely, just got to put more hardware behind it. Spindles. And then you got to manage it. That's right. That increases your cost. Will you be able to, let's assume that you'll be able to sort of isolate quality of service by customer, right? That's the nirvana I presume in your business. Nirvana, yeah, that'd be great. How do you charge? You charge cost plus today? Or can you sell the value of that if you can? Absolutely, I believe we can sell the value of that. The other thing we would like to see is the ability to create very high performance cloud products as well. High performance cloud products. Oh, yeah, okay. So there's the ability to sell sand storage as a service on top of our dedicated line as well as blended underneath a large number of cloud servers as well. So we can have multiple levels of cloud products as well. Okay, so let's sort of summarize what we talked about. You've got this cloud evolution going on. You've got flashes, this huge disruptor. And if I understand it, you guys are looking at flash as a way to actually generate incremental revenue and drive value for your customers as well as potentially lower cost. Because right now if I understand it, you're throwing hardware at the problem. And I presume that ripples through to management as well. Management costs? Yeah, the ability to manage any storage environment is going to be critical to this. You're going to have to have the transparency to know what's going on inside the storage. And you're going to have to have the ability to account for that so your application can perform at its best. Do you feel like this will enable new applications? When you think of the cloud, you think of applications like backup, and archiving, CRM. Do you see flash architectures supporting new levels of applications? Absolutely, one of the biggest problems that public clouds have is good storage access and the ability to consistently supply the number of IOPS that our applications want that our clients are using. So if we can do that, we can drive utilization up and the number of applications people can put in there really goes up. It opens up a lot of opportunities for customers. So give us an example. What kind of applications would you consider emerging that you can't support that well and guarantee the quality of service for a cost-effective approach? That you will be able to enable in this new model? I think any application that considers itself social, so any kind of social media, social gaming, those guys have application that really has to have rapid access to data and large quantities at the same time. And right now there's a lot of gyrations going on, memcache, all kinds of technologies used to kind of get around the problem, but if you can get reliable fast storage, you can make the problem a lot simpler for these guys. So if you had to put on your sort of telescope, look to the future, what would you guess would be your breakout between spinning disk and flash from a capacity standpoint? Is it 10% flash, 5% flash, 50% flash? What do you think? I think as far out as you go, that number increases. Whenever we introduce a new flash product today, we can't keep it on the shelf long enough to sell, to reach meet demand. So I think that will only be limited there by supply, I think. Excellent. All right, well, listen, we're out of time. I appreciate you guys coming in and talking a little bit about SoftLayer. Thank you. And sharing your perspectives. Good luck with the company. And thanks for coming on theCUBE. Thank you. All right, we'll see you guys later.