 Hi everyone, this is Stu Miniman with Wikibon. Thank you for joining today's peer insight. Today is July 30th, 2013, and the topic for today is enhancing cloud storage with technologies such as hybrid storage, a bunch of other technologies we're going to be covering. Our end user guest today is Steve Newell, who's with a cloud services company named Bunami. Bunami is based in Utah. Steve, thank you so much for joining us on the peer insight today. Thanks for having me. Great, so Steve, before you get started, we're going to talk about how your company's been around. Cloud services are kind of new. You recently went through a change, and your storage infrastructure, in my understanding, is you've also made some changes in your networking and really overhauling kind of the substrate of your environment. And that's created some new opportunities, some challenges that you went through going in through your decision-making process. We want to kind of dig into that over the next hour. So if we could start with, if you could tell us a little bit about Bunami and your role there and kind of what led up to this recent set of changes that you went into. Great, thank you. So Bunami, as you said, we are a cloud services company. We're actually a full-service data center. We have two data centers, one in the Salt Lake metro area and one south in the Utah County area. We do everything from traditional co-location to public-private clouds, hybrid clouds. We also do NGU desktop support, full-service IT support. As you mentioned, recently we went through a change in our infrastructure. In order to meet the demands of our growing business, we needed to expand our, initially it was our storage offering. We were using a traditional sand that we were very happy with but needed more capacity and potentially more performance. And that's kind of what led us to the process of evaluating all the different storage options that are available now. And we finally ended up on a 10-gile storage system. Okay, so Steve, can you walk us through a little bit about your profile? You said you've got two primary locations. What's your growth looking like? What's the size of your environment? And how old was the infrastructure that you had that you were then looking to upgrade? Yeah, so we actually started as a data center company in 2007. At that point we purchased our storage systems then. We've upgraded them and made that. As I said, we do have two locations. We have about 11,000 square feet in the Salt Lake data center and about 6,000 square feet here in the Orem data center. Both of those are about 50% capacity right now in terms of square footage. We have multiple bandwidth carriers coming into both locations. We have redundant connections between both of those. So both of those data centers can function as a hot backup to the other. With this, again, back on the storage, our initial storage purchase was in 2007. We made a follow-up purchase in 2009 from the same vendor. Steve, this is Dave Vellante. How you doing? Good, how are you, Dave? Good, thank you. So can you talk a little bit about how much storage, what kind of growth you're looking at and what some of the real challenges were that you were facing? Okay. Well, so, as I said, we were nearing capacity in our storage and began looking for alternatives. We stayed with the existing vendor and we're looking at that, but found that, you know, we were looking to add in the neighborhood of 20 terabytes or so to our storage system and that began to look a little cost-prohibitive. Okay, so who values some other vendors? Steve, can you just, my understanding is, you know, you've had a real good experience with your previous vendor, the original vendor and it's NetApp, correct? Yes, it is NetApp, yes. Okay, and you weren't looking to necessarily, throughout the NetApp, there wasn't any gap in the functionality of the product. It was really that kind of the price per gigabyte that you were looking at and the growth that you had. And you also mentioned performance. You were having challenges there, is that right? In fact, we spent quite a bit of time with NetApp and their engineers looking at additional solutions and things that we could do it and it actually did have a solution that would have met our needs. But again, it was rather expensive. And in order to get the performance we needed, we were gonna have to add a lot more spindles to the shells than what we had originally hoped, which obviously increased the price. So, but we were very happy with NetApp and continue to have NetApp in our environment and continue to use that story. Okay, and it was a performance issue at the end of the day. Great, so Steve, you went through a process with NetApp and you looked at a lot of vendors. Can you walk us through how you handled that process and what you were finding as you looked at a number of vendors? Right, so we spent, as I said, a lot of time with NetApp and explored various options. We looked at adding the flash in front, putting cash in there, mixing and matching the different types of drives, trying to get a price performance point that we were comfortable with. And at the end of the day, we felt like we were just about there, but felt like we needed to spend a little bit of time and look at some of the other vendors that were out there. And so we reviewed four or five different vendors. The last minute we were introduced to Tegile and looked at their offering and found that we were able to get the 20 plus terabytes that we were hoping for, a significantly lower price per terabyte. The IOPS were nearly an order of magnitude more than what we were getting with the NetApp as we had it back. And one of the things that we really like about Tegile is that many of the add-ons that we were paying for, the protocols and the replication and the snapshotty, that were add-ons with NetApp were included with Tegile. And that was kind of the thing that kind of helped push us over the edge there. Okay, so Steve, I definitely want to come back to some of these enhancements you saw, especially inserting Flash. Can you talk about kind of your business drivers? I understand that really updating your infrastructure allowed you to make some changes and some new offerings into what you had. Yeah, so in conjunction with upgrading our storage, we also upgraded our compute. And so we were able to tie in that new compute platform as an Assistant UCS platform. And we were able to integrate that with our new Tegile storage. And really provide a very, very performant environment. It is primarily our cloud that is hosted on that environment. We were able to, with this new hardware, offer significantly more performance at comfortable prices. So prices really actually didn't go up to our end user in fact in some cases we were actually able to lower the price of some of our cloud offerings and yet give quite a bit more performance. And most of that performance came from the storage subsystem although the compute also gives us a lot more. So Steve, I wonder if you could talk a little bit more about your business. So you're competing in an infrastructure as a service marketplace. Talk about how that's changed over the last, let's say 12 to 24 months. I mean, you've seen much more aggressive Amazon, you've seen responses from a lot of the major system vendors. You've seen this flash and hybrid storage come into play, this whole software defined meme. Talk a little bit about that business in general. How it's changed and how you guys compete and what your unique value is. Right, okay, well, so as I said, originally we were a straight co-op play in a cloud environment about three years ago and we've seen a lot of changes. We're still seeing some changes. As our customers are coming to us, most of the time they're coming to us as a co-op or a shared server customer and we're still having to do a fair amount of sell to get them to move into the cloud. What we have found though is that once we get them into the cloud, all of those things that you talked about, the software defined networking, the hybrid storage, all those things just really make it a really good offering for them. We are finding a lot of competition. Many of our customers are coming to us though from an Amazon or from one of the other public clouds. And I think the thing that we offer that some of those larger players can't use is we actually put a lot of hands on with the customer. We're able to help them understand what it means to migrate to the cloud. What games may need to be made to their infrastructure to get it to play in the cloud properly, as you're aware. We hear you, Steve. Sorry, you still with me? Yeah, we're still with you. Yeah, it's a little background noise from one of the participants it sounds like, but keep going. Okay, one of the things that we found moving applications to the cloud is that too often we think we can just move the app straight into the cloud and not make any changes. And often that's the case, but in order to take full advantage of all the power the cloud offers, sometimes we have to do a little bit of re-architecture or a little restructure. And one of the things that I think Udami offers is that we are able to help the customer do that. We have a very strong software development background and a very strong IT background as well. So we're able to help them understand them and make those changes. So it's fair to say that, relative to the sort of the gorilla of Amazon, you would compete largely on the basis of your service, your belly to belly service, your white glove service, your ability to go in and build a relationship with the customer, do more hands-on stuff than say Amazon is willing to do or it and its partners can do. Is that fair? Absolutely, and I think we also compete very well on performance and price. So it's kind of the trifecta. So how, I mean, you mentioned Teja before using this hybrid storage array presumably or a set of arrays. How has, how have flash economics changed the way in which you're able to deliver performance, cost effectively to clients? Well, particularly in a hybrid storage solution, we're able to take advantage of the flash even though it is still more expensive per gigabyte. But because of the way Teja and other, I assume other hybrid storage makers have architected their offering, it appears to the end user as if it is a fully flash array, even though it's backed by the less expensive SAS drives. So, you know, two years ago, I don't think we would have even considered a flash storage array simply because the economics just made it completely unfeasible. We didn't need quite that performance. It would have been nice, but we certainly couldn't afford a full flash array. So, are you able to, talk about quality of service a little bit and how you're able to, say, pin quality of service, can I do that at an IOPS level, a bandwidth level? How do you, how does QOS play into your service delivery? Well, it's really, honestly, it's less of an issue now than it has been in the past simply because the underlying storage system is so performant that we really don't need to spend as much time worrying about that as we have in the past. Our previous arrays, we had a mixture of SAS and SATA, and so, you know, we had a tiered storage offering and spent a lot of time working with the customers making sure that they were in the right storage tier. With this new hybrid storage system, that's not an issue anymore. All of our, all of our customers are on the gold tier, if you will, they're all accessing the hybrid. The hardware actually deals with the, you know, offloading data out of a flash and onto the slower and SAS drives and pulls it back on demand. So, we honestly have not had to spend very much time at all worrying about QOS issues, at least with the storage. So that's been another win for us. So Steve, I wonder if we could talk a little bit about that interrelation between virtualization and cloud. Most of your customers, you said, starting out in kind of co-location. Are they using VMware today? If they look at going from, you know, just a colo to a hosted, from a hosted environment, I'm sorry, to your services. Are they staying on VMware? You know, what's your relationship with VMware as I guess a good place to start? Okay, yeah, we are, our public cloud currently is hosted on VMware infrastructure. One of the things that kind of the half step that we took to the cloud, I think as an industry was, you know, moving from physical servers to virtual servers, that next step in my mind is the cloud which gives us a little bit more dynamic resources. So we are hosting the majority of our public cloud on VMware and fronting it with vCloud Director, which gives our end users quite a bit of control over their virtual environment. So one of the things that does for us is it kind of eliminates the hands-on that we need to have, gives the customer the ability to adjust their resources dynamically. And I think it's a win-win for everybody. VMware has been a great technology. It integrates well with, you know, all of our systems from storage and networking to the compute level and gives us a great deal of control and gives the end user a lot of control over there. Okay, it's my understanding you're also using network virtualization for VMware. Can you fill us in on some of what you're doing there? I'm sorry? Network virtualization from VMware? I still missed the first part. Steve, network virtualization, then I see your stuff. Thank you. Oh, I hear it was Asian. Yeah, so we do a lot of software-defined networking. We really have appreciated the flexibility that that gives us. So we're able to create virtual data centers for each of our customers and allow them to then manage their own edge gateways in the VMware environment, giving them access to their own firewall, their own VPNs if they need them. And the thing that I think most of our customers are really appreciating is the, you know, VDC networks and the ability to create private networks in their own environment that lets them isolate resources. And we're finding a lot of our customers are taking advantage of that and also taking advantage of the load balancer that's built into the edge gateway. Okay, great. I wanna take this time to open it up to the audience. Sounds like we had quite a few people on the phone. I know David Foyer's phoned in. So let me pause for a second and see if we've got any questions from the community. Sure. If you can please at least let us know your first name and if you're comfortable, you know, what company you're with or what your role is. Absolutely. My name is Dan B. Kraft. I'm a vice president here at Morgan Stanley and I run one of our cloud computing portfolios. And I study a lot about this whole industry thanks to Wikibon and just props out of the day. But quick question for you on companies with a flash drive. So in my area here in Seattle, you know, I cover a lot of Amazon and Amazon's cloud computing services and, you know, our analyst department has said it's kind of a threat to everyone in the industry because of their, you know, their ability to just spend money for something like a whole fusion IO storage facility. What is, how does that, and I noticed I listened to what you said about how you were looking for a cost effective solution to provide that. Is that something you think that you may do in the future where you may look at maybe either a fusion or maybe EMC or STAC or any of those? Do you see your business progressing in that direction or do you see more that you're gonna use some of this SAS technology and hybrid a lot of it out? And I'll take your, I'll mute my phone and listen to your answers. Thanks. Well, you know, as we were doing our analysis, we actually looked at several of those vendors. We lived very closely at EMC. We're very familiar with Fusion IO. And again, for us right now, the price performance was just not there. It wasn't worth it to us to go full flash. That's not to say in the future we may not, but I don't know that at least in the foreseeable future that that's gonna be required simply because of the performance we're getting from the hybrid array that we currently have. I don't know that we would get that much of an increase in performance going to a full Fusion IO type system. Okay, thank you. This is David Foyer. I've got a question about the type of workloads that your clients are using your service for. Could you talk a little bit more about what they, what types of applications they're using, what types of environments they're using your service for? Sure, I can't tell you a whole lot about some of them simply because as it is a cloud environment, we're not intimately involved with what's being run there. However, I do know that we have a fair number of web farms. I know that we've got a fair number of Linux and Apache stacks with either Postgres or MySQL. Some of those databases based on the storage that they're consuming are quite large. I also know that we've got a fair number of Windows web stacks as well. Quite a bit of .NET running in our environment. We also have some desktop, hosted desktops running in this environment as well. So I think we've got a wide range of applications. I know of at least two or three companies that are using our cloud as development environments. I don't know if that's day to day development or if those are kind of dev test environments, but it is a fairly broad spectrum of applications running here. Thanks very much. Do you have any large database type customers or production database type customers in that? Is that an environment that might be more stressful on your current storage environment? Yeah, I believe that we do again. I don't have real good visibility into what each of the virtual machines is running. However, based on some of the storage that's being consumed by some of our customers, I would suspect that they are running large database environments, multi-terabyte databases. That's how much storage they're consuming. And do you have any issues with the noisy neighbors? How do you separate out any noisy neighbors to identify them or do they not cause a problem? We have, one of the advantages of using the VMware technology is that the noisy neighbor problem is minimized and handled automatically using the orchestration piece that VMware has. They will automatically migrate either the noisy neighbor away or the neighbors away on the separate compute node. And that's pretty much handled automatically for us. We have enough capacity that we're not operating at full capacity. I think we try to keep our capacity well below 80% so that we've got enough compute to be able to isolate those noisy neighbors when possible. The effect that's, I've got several of our customers have come to us simply because they had noisy neighbor problems in the previous cloud provider and needed that assurance. And since they've been here, we've been able to eliminate that problem for us. Excellent, thank you. Steve, I'm wondering, you mentioned that you've got a lot of your public cloud environment on VMware. You've got some customers that are using Microsoft environments. Can you just tell us from your perspective, what's your look at things like Hyper-V, Microsoft Azure and OpenStack? How do they play into what you're doing and what your customers are asking for? Thank you. OpenStack is a big part of what we're doing. We have created some private clouds for customers using OpenStack. Those private clouds are hosted both in our data center and at the customer's location. I can see OpenStack becoming a more integral part of our environment as we go forward. We do also have Hyper-V running in some areas. We have not spent as much time in that simply because, again, the VMware and the OpenStack solutions are sufficient for our needs. We are, however, very pragmatic. So if it becomes very evident that Azure and or Hyper-V are required for us to do a good job for our customers, we'll certainly implement that. But to this date, VMware and OpenStack have met all the needs that we've had. Great. We have customers, however, let me just say, we do have customers that I know are running Hyper-V in virtual machines and are VMware Cloud. And we have success. Great. Sorry, sounds like there's a question on the phone. Please go ahead. Yeah, this is Scott. Well, I've got a question about some of the friction points you might be seeing from people. As they consider moving some of their infrastructure into Cloud, what are some of the concerns that you're getting from CIOs? Are you getting any? What kind of business follows they're trying to solve as they move to Vennabu? Well, you know, we have a lot of customers who are coming to us now that currently have two, three, four racks of gear. And in some cases, they're daging. They don't want to continue to support that and they're wondering. I think the biggest question that we have or the biggest concern we have as people are coming to the Cloud is simply lack of understanding and lack of experience with the Cloud. A lot of the people that are coming to us, their Cloud experience is limited to small instances in Amazon or in some cases to burgeoning instances in Amazon where they started small and Amazon has grown nearly out of control and their cost has just gone through the roof. So I think the two major concerns we have with people moving to our Cloud are number one, will it perform the way my existing infrastructure does and can I get away from managing the multiple racks of equipment that I have? And then number two, how can I control my cost moving to the Cloud? And so one of the things that I think that our offering gives those people is that with the VMware environment that we have installed, we give you full access to your environment. So you can create, you can control the number of virtual machines that you're running and the networking between them which gives a lot more control over that. The other thing I think that we're trying to address is just simply the lack of understanding of the Cloud. Too often, we're still viewing the Cloud I think as if it were a traditional rack of servers and I think as we understand the virtual environment a little bit better and the flexibility that it allows us, I think we'll start seeing more and more applications that are architected to take advantage of that rather than architected in the traditional end-tier environment and we'll start to see elastic applications, not just elastic Clouds. Can you add some detail to that, Steve? I mean, we've often said on the cube at SiliconANGLE Wikibon, we feel like we're entering this new wave of Cloud adoption. The first wave was sort of the test and dev experimentation and then when the economy turned down 2008, 2009, there was a big push to transfer CAPEX to variable expense and then coming out of the recession, there seemed to be a lot of line of business movement to the Cloud and now it seems like CIOs are increasingly embracing the Cloud. We know a number of folks within the Wikibon community particularly in mid-sized companies that are really getting aggressive about the Cloud but specifically not doing a my mess for less doing, trying to transform their business and do a deeper business integration. Can you talk about how, if that's a reality yet and our people, I mean, you sort of touched on it just now but are people actually going deeper? Yeah, it's becoming a reality. I think, but I think it's going to require a little bit of a paradigm shift for us. We're still, excuse me, I think we're still locked into the lamp stack mentality where we've got Apache up front and we've got MySQL in the back and PHP or something else in between to manage that but these monolithic apps are just not really what the Cloud is all about. We need to start building a little bit more modular applications and it's going to require somebody at each organization to spend a little time and understand the business and understand how you can break that into smaller app servers. The advantage of that then is that I can, I can script the expansion and contraction of my Cloud resources based on demand and that demand can either be compute. I have a job that I've got to run that requires more compute resources so I need to expand that or it could be load. I've got more customers hitting my site so I need to spin up more Apache front ends or whatever it may be. Right now most of that is done fairly statically and I think we can do a lot better job managing those resources if we spend a little time on that. So you're kind of proposing a flattening of the Cloud stack, of a Cloud pancake and about a scale out model, am I getting that right? Absolutely, yeah, absolutely. I think it's hard to believe as far as we've come in the last 20 years with technology it's hard to remember that. We're still in the infancy of much of this technology and I don't know that we've explored all the possibilities that are there. I think if we get away from the monolithic app and move over into more actor based potentially or very small modules, I think we can expand this. And the Cloud is perfectly positioned to allow us to do that. If you can define your networking, if you can define your storage, if you can define your compute all within a year from code that gives you a lot more flexibility than we ever had in the past where each of those required a network engineer to deal with or a storage engineer to modify. If I can do it from within my code I've got a lot more, a lot more power. What if we could talk a little bit more about performance too, you were mentioning performance before. How do you measure performance? I mean, is it an IOPS? Is it a latency measurement? Talk about that a little bit. We typically take a look at IOPS and that seems to be the method that is most critical to us. Latency, again, as we've gone to this hybrid storage, latency has kind of gone away. We haven't really had to worry about that anymore. The IOPS, again, have continued to increase. And so can you quantify that at all? I mean, what kind of levels are you seeing? I mean, you're in the hundreds of thousands. I mean, can you put any metrics around that? Yeah, we do. I don't have them on my fingertips. I had scheduled my storage engineer to be here with me today, but unfortunately he had some emergencies last night and it's just now going to bed, so. Yeah, oh, I'm sorry to hear that, but no worries. I mean, David Floyer was pressing you before about database activity where presumably David latency would be more of interest and Dan was sort of asking questions as well. I don't know, David Floyer, if you're still on, I wonder if you could sort of clarify what you're seeing with regard to some of the different performance metrics. I mean, it's interesting to hear Steve talk about, IOPS is really the main metric. You hear other customers talk about latency is the real issue. Can you maybe help us squint through that dissonance? Sure, I can give my perspective on some of the different areas that we're seeing. I think the opportunity of Flash is to separate out IOPS and capacity from each other and obviously IOPS and bandwidth are connected. So that you can charge separately for all those two dimensions. And that's happening in a number of places. And that adds a great deal of flexibility because on the same infrastructure, you can provide low cost, low IOPS for capacity and also high cost, high IOPS as well. Latency is very much an issue for large databases and lower latency, getting it below a millisecond, can enormously increase the throughput through systems, particularly large databases. And that was one of the reasons we're asking about your database environment and obviously, to some extent, that's hidden from you. But the large databases do very well not only with low latency, but very consistent latency. It's avoiding the occasional 200 millisecond response time that you can get in a disk environment, avoiding those and avoiding the timeouts that occur on a whole sequence of, for example, locks. Okay. So we think there's a tremendous opportunity in cloud service to offer a more modular type of approach. I don't know, is that something you're looking for? Well, and yet, I wanted to just ask, Steve, I mean, it seems like the Venami model is really simple. Look, this is the level of service that you're getting. It's the default service, you call it the gold level. It's more performance than you're ever going to need for your class of customer. Am I, we're getting that right, that that's the market that you're going after, that sort of sweet spot in the middle. Right. And that's exactly right. And one of the things that we've found is that before we had the hybrid storage in place, we done a lot with storage and done a lot with the performance of the storage, both latency and IOPS. Since we've installed the hybrid storage, we haven't had to deal with those issues. So as we talked about a minute ago, the hybrid, the flash really does give you both the latency and the guarantee of low latency combined with the high IOPS. So I suspect if we were to have a customer who needed, who had a very large database and needed some guarantees there, we would probably structure that a little differently than just our standard cloud offering. We put some other guarantees around it, some other metrics and some other measurements around it to ensure the performance was there. But as you mentioned, our current customer base is kind of that middle of the road. I need what I need and I need it as fast as I can get it. And what they're getting now is faster than they were getting before and certainly faster than what they're getting elsewhere and they're very happy. So Steve, I'm wondering if we can talk a little bit about just when it was T-Gile that you ended up choosing. My understanding is you looked at a number of companies, both some of the big guys and some of the startups that were doing flash and it was some of the, tell us kind of why T-Gile stats really made the cut and some of the other ones didn't stack up against the feature functionality that you had before. Right, so what it really boiled down to is a couple of things. Number one, we needed the capacity. We needed to have at least 20 terabytes of usable storage is kind of what we were shooting for. We needed the performance, the IOPS, but we also needed the multiple protocol support. We offer a lot of NFS storage, we offer ICE SCSI. And those were critical to us. We needed the visibility into the infrastructure from a management standpoint so that we could manage our sand and manage the performance and the load and all those things. One of the things that we found as we looked at each of the vendors is one of them could give us what we needed but none of them could give us everything we needed in a way that we wanted them. For example, some of the other hybrid storage offerings are exclusively ICE SCSI. So if we wanted to offer NFS, their solution was that we would spin up a Linux VM, mount the ICE SCSI and then run NFS from there and offer it out which just seemed to be more than we wanted to deal with. Others would offer the protocols but charge us for it so it would be an additional charge. When we got the tech out, we found that we had ICE SCSI and NFS support included in the base offering. Their portal, their management portal gives us all the visibility into the system that we could ever need and gives us the ability to monitor mode and performance and capacity, we can do capacity planning all that stuff. It's integrated with VM 1 so the storage drivers are integrated with the VMware environment as well and we were able to get the capacity that we needed and then them coming in half or a third of the cost of the comparable offering from our previous vendor. And we had the advantage of having the hybrid as well. So how do you see this shaking out? So you've seen, just talk about the players and the market a little bit. You get innovators like a Ted Gile and others coming out, we've got all Flash array guys, you've got some hybrid folks and you've got the big guys now entering the market. You saw EMC big announcement this year, obviously NetApp is trying to get its act together in Flash and has made a number of announcements. So how do you see that shaking out? I mean, do you have an opinion on that? And I know it's just your opinion but in terms of the little guy's ability to stay ahead of the big whales that are now coming into the market as a buyer, how do you look at that? And how do you, you know, because to buy from a smaller company, obviously it's riskier, so they've got to be better. So clearly you saw a better value proposition. I wonder if you could talk about that a little bit and maybe speculate on what you think the market's going to look like a couple of years from now. The big guy's just going to gobble up the little guys or can the little guys keep moving fast? Well, I think, you know, we obviously invent and I think the reason for that is, well, there's actually a couple reasons. Number one, the big guys, they know how to offer stores. They've been doing it for years and they've got a well-established architecture. Unfortunately, sometimes I think that can be a hindrance as we all know, sometimes moving from an existing architecture that has a great deal of market share is difficult if you want to move to a new architecture, especially one that's disruptive. I think that some of these smaller startups are able to take advantage, you know, build on the shoulders of giants, if you will, take advantage of the strides that the NetApps and the EMCs and the other players, the three-parts, the architectural strides that they have made and can take a look at that and say, hey, we can build on that and offer a more economical solution. I'm not sure that, well, I don't believe the EMCs and the NetApps of the world are going away anytime soon. I think they will adapt and adjust and I would suspect that in five years, they'll probably still be owning the lion's share of the market space. However, I see the tedgiles of the world coming in and taking significant chunks out of that market share and I think that will continue to grow and to change. I think that the hybrid storage space is enough of an architectural shift that it's gonna be difficult for the legacy players to jump in quickly. I also think that these smaller guys are nimble enough and agile enough that they can make changes and improvements going forward and can be aggressive enough in their pricing that it makes it attractive to bring it in. I know that there are a lot of people that we've talked to since we adopted tedgiles that are doing the same thing we are. They're an existing NetApp or an EMC shop that are bringing a tedgile in alongside of that. To supplement, augment their existing storage, kind of a tire kicking exercise in some cases, I think. But as we found, and I think others will find as well, that the tedgile and others like that offer enough of a price performance advantage that it's pretty compelling. Okay, so Steve. Five years from now, as we prepare to upgrade this system, five years, two years, whatever it is, when we need to upgrade, I suspect we'll continue the hybrid storage line until we get to a point where flash is cheap enough that we can afford to do all flash. Okay, Steve, I wanna open it up for anybody else on the phone to ask some questions as we're getting closer to the top of the hour. So anybody, if you have a question, please speak now. Yes, it's Josh Kusher. Hi, Josh. Hi. Two remarks. One remark, the last decade is characterized that the innovation is coming from small companies. And these companies, most likely, are purchased by the large one. Like, if you have been, for example, Component by Dell, XIV, Stolvitz by IBM, et cetera, et cetera. Some of these companies manage, in fact, NetApp was also startup something like 12 years ago. And this is the only company that grew from startup to be a large company. But as I said, the innovation today is coming from small companies. And most of the cases, they are purchased by large companies. Now, the second one, I listened to the performance. And there is a confusion, because if you look at the disk, there is a performance which is latency and there is throughput with the IO per second. And the IO per second is the maximum, the number of IOs that can be executed before the country goes in saturation. So, both of them are important. And the question is, what do we need? Because if a customer need large throughput, for example, the question is, what is the latency that he need with the large throughput? And you have the graph of the performance is like a goal, so it's normally more or less flat until it goes to saturation. And the flat is the latency time. The second performance measurement, and this is of transaction processing. The second measurement is for sequential operation. And sequential operations are measured by throughput. How many megabytes, gigabyte per second, the contrary it can transfer. So, unfortunately there's a mixture with performance and in fact it should be throughput and not performance. So, two points there Josh, one on innovation and you and I have talked about this a lot, and I wonder if I have a follow up and then the performance discussion, I think you're right, it sort of depends on your workload and it's a balance. But I want to first ask you about the innovation comment and you and I have jostled about this debated et cetera about, I mean you flat out made the statement that large companies don't innovate, they purchase. But to that point, and I hope I'm not putting words in your mouth, but you've been pretty outspoken about that, but to that point you've got large companies like EMC just did a five billion dollar debt offering. It's got 17 billion dollars in cash between its three federated companies. You've got Oracle, IBM, HP paying down its debt and eventually will become an acquirer again. So what do you see as far as these companies, maybe they can't do the organic innovation, but what's wrong with that? If they're in internal R&D is for adding features and functions to existing products to essentially keep existing customers happy and they write checks instead of code to innovate, what's wrong with that model? Does it just create too many stove pipes, too much inconsistencies, lack of integration, or is that a viable model now going forward? What are your thoughts on that? I mean the thing is that I don't say that the large company are not innovating, but they're innovating much less than the small companies and normally they are improving. There are a lot of features which are requested for example for development of those control units. So there's constant development, but if you think about ideas, let's talk for example about the real time compression. Which is Storwitz. It is a company which started in Israel, it was a unique company, and they have several PhDs and the idea came from them. If you think about the duplication for example, nobody spoke about the duplication until 2004, 2005. And then suddenly you have companies, some of them were new companies like Diligent which was purchased by IPM or Data Domain which was purchased by EMC and there was Avamar which was purchased by EMC and et cetera, et cetera. The innovation or startup are people which are normally hungry people which work enormously to, because if they will succeed, the reward is very, very high. The people in established company, large companies, they are getting salary. They, if they will invent something, they may got a bonus, but they will never be a millionaire. You're getting at a boy. I know, I know many millionaires which were from startups very simple. I have people who made four, who started four startups selling each one of them to large companies. Is that answer? Yeah, yeah, yeah, yeah, absolutely. Okay, so I just wanna, as we're getting close to the top of the hour. This is David. Go ahead, thanks, David. I've got one question if I may from Steve. How do you, what's the difference with the Tegile in terms of the support that you need to give it? You've mentioned several times you don't need to spend as much time on that. Could you put some numbers around how that's come down? And how have you integrated that into your orchestration? Has that been an easy thing to do? It has been easy. What we basically have done, and again, most of the storage is consumed by our public cloud. So we're able to do all of the provisioning done either in an automated fashion or when the customer chooses to increase or create storage. In terms of support, the real issue that we found in the past was we spent an awful lot of time, and I suspect it was mostly because of, because our capacity, we were nearing capacity, so we spent a lot of time juggling storage, making sure that it was available. Now that we have more storage, that's not an issue anymore, number one. Number two, we also spent a lot of time in the past balancing the latency in the IOPS for our customers. From our standpoint, the workload is very mixed. We don't have any control over it because it's, it's our cloud, and we don't have any control over the individual workloads. So we would have to kind of play that, that balancing act between latency and IOPS in the past. Because the TedgeOut system is so performant, both from a latency standpoint and an IOPS standpoint, we don't spend any time worried about it anymore. It just performs. How much time did you suspend? Is that a, you had to dedicate to that? Or what sort of size problem? We had, we have an engineer who's, who's among other things, tasked as our storage engineer. I suspect in the past he was spending close to 30 to 40% of his time dealing with storage related issues. Since we've got the TedgeOut installed, I think that's dropped down to about 5% of his time. Right, thanks very much. And those are, I know he'll tell me I'm pursuing. Sounds like there's another question on the phone there. Hold on, Steve. Yeah, this is Scott Logan. I have a question about your remote replication agreement with TedgeOut. Can you tell me a little bit about how that's working out and how people are taking advantage of that? Thank you. I'd forgotten about that. You know, one of the things that purchasing the TedgeOut did for us is give us, give us that opportunity. We have a TedgeOut stand in both of our locations. One of the things that TedgeOut builds in is his replication capability. And because of the way they do their technology, all that data that's replicated is deduced data as well. So we're able to take full advantage of that. We have entered into a partnership with TedgeOut to offer off-site replication to their customers. So we have several of their customers now. And it's only been the last, I think, six weeks or so that we've actually productized that. So we're just kind of in the initial ramp-up stage of that. But it really makes a very compelling offering for TedgeOut as they go into some of these local data centers customers to be able to offer a two terabyte replication or a five terabyte replication to our data center. And then we can offer multiple add-on solutions for that as well. So we can encrypt that. If that connection needs to be encrypted, we can create an IP set tunnel between locations. We can do anything from just a cold standby to a full DR solution as well. So if, in one instance, one of the customers is running a VMware environment, so the payload that's getting replicated to our data center are virtual machines. In the event of a catastrophe, if they need to, we can stand those virtual machines up in our cloud environment very, very quickly. And they can be back up and running within a matter of hours, as opposed to days or weeks. Did that come close to Edge? Yeah, do you have people right now that are taking advantage of that service that maybe have their own TedgeOut UV units and they want to use you guys as sort of a DR opportunity? We do, actually. Yeah, we are consuming several different replication points. So that's been a huge win for us as well. Great. David, did you have a follow-up there? OK, Steve, I've got one final question for you. You went through a significant change from an old architecture to a new architecture. What advice would you give to kind of your peers, to companies that are kind of deciding to go internal or to cloud environments, or just looking at those architectural changes from disk to something like a flash environment? What have you learned? What would you do a little bit differently? What advice would you give to your peers? Well, selfishly, I'd say just call Lunami and we'll take care of it for you. But realistically, there are a lot of issues that you need to take advantage of. And nobody knows your environment as well as you do. And so you need to make sure that you understand that. One of the things that we found as we were going through our storage review was that in a couple of cases, the vendor was trying to tell us how our environment worked. And that didn't really fly very well with us. We know it very well. So make sure that you understand your environment. Make sure you know what your pain points are. Make sure that you're evaluating apples to apples. Too often, we get caught up in IOPS, or at latency, or whatever it may be, or it's flash. As technologists, I think sometimes we get too excited about shiny and less excited about what really, really is the solution. So I think as we're going through these processes, make sure you understand what your problem is first of all. And make sure you're solving the right problem. Secondly, make sure you understand the technology you're planning to migrate to so you can take advantage of it. Don't, in terms of the cloud, don't treat the cloud as if it were just another rack. Use it as it's designed. Make sure you take advantage of all the possibilities that are in there, even if it means spend a little bit of time in re-architecting. But I think those are probably the two keys that I would suggest. All right. Steve, thank you so much for sharing your feedback here with the community. We're going to go into wrap up mode now. So a reminder to everyone that the recording of this video, the audio podcast, as well as new research covering lots of different angles from the technology, the CIO, the vendor community, organizational actions, will be up on wikibond.org, usually within 48 hours, definitely within the next week. So we recommend that you go to wikibond.org, check out the content. If you have content that you want to add to it, feel free to add comments, hit edit, participate in the communication. Video will be up on youtube.com slash silicon angle. You're also going to be able to find lots more on cloud. If you follow The Cube, which is on Twitter, just The Cube, and siliconangle.com, a lot of big cloud shows coming up. At the end of August, there will be VMworld. In September, you've got to work a open world. And then a little bit later in the world, Amazon re-invent. Expect to see lots of coverage from us at all of those events and many more. So thank you so much for joining. Thank you to Steve Newell from Vunami. Of course, if you go to vunami.com or Vunami on Twitter, you can find out lots of them. We really appreciate them participating and sharing with their peers what they're doing on there. You can also, you know, contact us. My name is Stu Miniman. I'm at Stu on Twitter. This was D. Volante here with me. I want to thank the people on the call that participated. Dan from Morgan Stanley, Josh, David Fleuer, Scott Lowe, and I think I got everybody there. So this is Stu Miniman wrapping up with today's Peer Insight Enhancing Cloud Services with Hybrid Shored for July 30, 2013. Thank you so much, everyone, for participating. And we will see you next time. Thanks.