 Live from Austin, Texas, it's theCUBE. Covering Dell EMC World 2016, brought to you by Dell EMC. Now, here are your hosts, Dave Vellante and Stu Miniman. Welcome back to Austin, Texas, everybody. This is Dell World 26, Dell EMC World 2016. This is theCUBE, the worldwide leader in live tech coverage. CJ DeScience here is the president of the Emerging Technologies Division at Dell EMC. Good to see you again, CJ, always a pleasure. Good to see you as well. So, well, this is everybody's first Dell EMC World. Yes. Even though EMC World had a flavor of Dell EMC World in the spring, yes. Yes, correct, correct. But how's it going? What are customers saying? What's it all about? First of all, it's going great. We announced a couple of products today, so pretty pleased with that. And what we are finding out, given that the Emerging Technologies team focused on a lot of software-defined storage technologies, now Dell and EMC coming together, we can use those technologies on top of Dell platform. So we are very optimistic as well as giving solutions that are the right fit for customers. So, refresh our memory on the portfolio. Obviously, Isilon, DSSD, Scale IOs in there. ECS. ECS, yes. And then we also have a team EMC code, and I'm sure you'll be speaking to Josh later on. CJ, so we've been hearing a lot from EMC. We've been talking about Flash for many years. Earlier this year, EMC called it the Year of Flash. So in your portfolio, you've got block, you've got file, you've got object. Now you've got a lot of Flash options and all Flash options. Maybe if you could walk us through a little bit kind of the impact architecturally on those various components. Yeah, so let me start with our biggest announcement today. So we announced Isilon Nitro. So we pre-announced it at EMC World in May. And today at Dell EMC World, we really announced it. And the thing that we have seen with Isilon is that unstructured data growth doesn't seem to be slowing down. That's for real. I mean, we thought, okay, it was spiking a few years ago and maybe does it slow down, but it has not slowed down. And the customers who love Isilon platform, you know, scale out NAS, said, hey, you guys do a great job at the capacity tier. Isilon is a great product. One FS is very scalable, but, and the but was can we have higher performance and possibly an all flash configuration. And we used flash before, but it was used more as a cache and things like that. So the team looked at it and figured out that yes, it is absolutely possible. They are also modifying the one FS stack to leverage flash. And we believe whether it's the EDA industry, whether it's media and entertainment or healthcare industry, they would be able to, which are all big Isilon customers, they would be able to take advantage of flash and be able to maybe run applications on it and have very high throughput. So what we announced today is, Nitro is the code name for Isilon, all flash configuration. We are taking orders this quarter, pre-orders because there is a lot of demand from customers. They love this whole idea of all flash Isilon. And in four nodes, you can get 250,000 IOPS and 15 gigabytes per second. But the biggest thing is it can scale up to 400 nodes. So you can go from 924 terabytes in four nodes all the way to 92.4 petabytes. So this is unprecedented for Isilon, very excited. Customers are excited in EDA space as well as in media and entertainment space. We will release the product somewhere in the middle of next year. But we are already working on the beta program this quarter, asking customers to try it out. So that's one I do want to address because you asked around file block and object. Scale IO, it's a great high performance software to find storage product. The acquisition was done, I want to say, three years ago approximately. And what we did is customers, when they evaluate Scale IO, you can download it from emc.com or put it on any type of server. But now with Dell PowerEdge server and Scale IO software, we have ready nodes and this can be consumed easily. You can again, Scale IO is also scale out like Isilon. So we love that. And it also has an all fresh configuration. C.J., just I think one of the things people don't understand about Scale IO is that when they look at the categorizations, it's software defined storage. It's also hyper-converged infrastructure. And it can be in either of those configurations. I can have an all storage configuration or I can have kind of compute and storage together. You know, what do you see from customers? I mean, my understanding is you've got some really big customers doing some really big deployments which kind of blow away what most people think about when they think of either SDS or HCI. Exactly. So I would say our journey started and we were very particular on this journey on Scale IO. When our journey started, it was financial services, service providers, types of customers who looked at Scale IO said, okay, this can support bare metal. This can support many hypervisors, you know, VMware and others. And it supports Docker containers and other things as well. And they said, hey, let me start with tier two workloads in test and dev. And I want to be able to scale out because Scale IO can scale up to thousands of nodes. And some of our large telco service provider customers went to Amazon EBS and few others, looked at the performance and flexibility they have, tried Scale IO and used Scale IO for on-prem configuration. So on the category wise, I'll tell you that we are seeing maximum demand in large financial services, in large telcos, in service providers. And we have seen half and half, so some customers say, I want to use it as a software defined storage layer. So like what you were saying, it's in a two tier server and storage separate. And there are some customers who are using it in hyperconverged fashion as well. We are seeing customers in Japan, some very large ones use it for open stack. So because it's flexibility with operating systems and hypervisors is doing really well. I hope that answers your question. It does, it's a nice bridge actually, everything from kind of small scale to data center scale. And even I can start with kind of solving some of my storage issues and then solve some of my just infrastructure issues overall. And Stu, what we see is, of course VMware is everywhere and we are very proud of it. But customers may be experimenting with something else. They may be looking at Hyper-V or whatever the case might be. And they say, okay, this is a great solution that allows me to scale and I can run mission critical workloads. So we have customers who are running internal cloud services on scale IO because the scale out you keep on adding nodes, it rebalances really well. And talk about they're buying this storage as sort of a new architecture for their new apps. Is it to grow sort of their traditional base? I would say, Dave, it's both. So in some cases we see very clearly, I have a huge ESX farm and this thing spans to hundreds of nodes. There is no cluster size limitation, so this works great. In other cases what we find is, I'm running Cassandra, I'm running MongoDB. I want to use containers platform and persistent storage problem on that container platform is only sold by scale IO. So they are like, this is great, we have certified it, our core team has done a lot of work with scale IO. So overall in a what EMC called it and IDC called it a platform three workloads, we find that scale IO is a perfect fit. How about DSSD? Maybe you could give us the update there. Yes, so DSSD has now been in the market for two quarters. We are, we certified first of all Oracle, so for high performance data warehouse analytics, not at a transactional level, but more on the analytics side. We currently blew away performance benchmark numbers on SAS. So we find that SAS and DSSD is a great fit. And we tried with, in complete candidness, we tried with Isilon and SAS. We struggled on getting the right solution in place, but DSSD and SAS works really, really well. And what we are finding is whether it's in fraud, analytics, government agencies, financial services, they look for analytics application. We have done a lot of very cool work with Cloudera and HBase runs really, really fast on top of Cloudera and DSSD. So it's early days, but we are seeing traction, lot of POCs going on. We closed some very strategic deals over the last couple of quarters with some large names. So very excited. Interesting, HBase. So it wouldn't have been uncommon to see HBase running on Isilon, right? But now you see things like Spark coming in and in memory. And the memory architecture is, yeah. And the storage becomes the bottleneck and DSSD breaks that bottleneck. Yeah, it says close to CPU as possible. And with DSSD, the other advantage is with Dell and EMC coming together, we can create a configuration with Dell servers that are ready to consume DSSD via PCIe and have a rack scale configuration that customers can consume easily. So it was almost like, for whether I look at ScaleIO or whether I look at ECS running on Dell servers or whether I look at DSSD, it's almost like this is something that has been missing and now things come together really well. I mean, as the world becomes software defined, the economics of storage, you know, shift to software obviously, it's non-recurring engineering cost, is the premium that you have to pay for, let's say a DSSD, artificial in that it's more valuable so you can charge more or is it fundamentally more expensive to deliver those services? So, you know, I would tell you, first of all, DSSD's fundamental principle was they are not going to buy off the shelf SSDs and get raw NAND from prominent NAND provider and try to eliminate all system bottlenecks so you can read, write really fast. What we are seeing more than the IOPS, when we say 10 million IOPS or 12 million, people are like, okay, that's a lot of IOPS, maybe some other all flash providers can do that, but it's the bandwidth, right? And the 100 gigabytes per second bandwidth for analytics type of workload where you have lots of data going back and forth and with combined with 100 microsecond latency, that's the sweet spot and that's where we find that we see the traction. So, to answer your question, you know, can we, is the value in the software yes, but there is also value in the hardware because we have very compact NAND and as NAND evolves, we are going to be doubling capacity every year. So, we are starting out at 144 terabytes and I don't see a couple of years from now why we won't be at 500 petabytes in 5U, which is a lot of flash. CJ, there's been a lot of movement in the marketplace in the object space in the last year, so, you know, what about ECS? So, we announced ECS 3.0 today. We shipped it on September 30th, so just a couple of weeks ago. The one thing we have done with ECS is, now what I find, and you know, one of the analyst firms has done even the quadrants based on scale out file and object combined. So, they didn't say, let's just look at object by itself because object is also the eventual destination for unstructured data, right? So, we are finding that customers are using combination of ICELON and ECS. We signed some really nice strategic deals last quarter, where from ICELON, they were using ICELON as a performance tier and then tier the data out via cloud pools to ECS. Very powerful solution. We are the only ones who can do that. Yes, you can tier to other cloud providers, but customers in regulated industries like that control, so that's first. Second is, when I want to use modern applications using S3 or Swift and other protocols, ECS is a good fit. We are seeing slowly, customers say, okay, I am a media and entertainment client. I have lots and lots of unstructured data. I want geo-replication. ECS is amazing for that. And the last thing I want to add is, we announced ECS 3.0. We added a lot of compliance features. We have a lot of loyal center customers who can then migrate easily to ECS. But in addition to that, very dense configuration, where you can use it for tape replacement. So ECS D-series that we announced, in one rack, it is 6.2 petabytes. So it's a very, very dense configuration. CJ, is there any connection between what you do with ECS and the storage cloud that VirtuStream has? Yeah, so, thank you. That's a great question. It would be a miss. Two things we are doing with VirtuStream. One is that VirtuStream storage as a service uses ECS. That's how they store all the data whether it's coming from our VMAX, cloud boost, cloud pools, all of those technologies pump data into VirtuStream cloud. But we also announced with ECS 3.0 a managed offering for ECS. Because you get a lot of efficiencies when you are three data center for ECS. And if customers don't have facilities, they have two data centers or they are maxed out on the floor, then they can use VirtuStream's dedicated ECS service so that they can get maximum advantage. Yeah, in the competitive landscape, I look at IBM has, you know, Cleversafe in SoftLayer now, Oracle has their storage cloud. How do you feel you stack up compared to those guys? So I would say on features, functions, looking at the flexibility of the software, performance, I feel pretty good about it. We have a very competitive solution. As you know that the stack was written from scratch and some of the mistakes the team had made in the past, they were able to correct it. We have very large deployments of ECS now. One of our public references is Verizon, who is using public facing cloud services using ECS. So even though ECS just shipped a couple of years ago, I feel very good about where ECS is going. You mentioned ECS as a tape replacement, potentially. How are you able to make it cost competitive with tape? So ECS in general is very cost competitive because when we compete with ECS, customers are evaluating maybe a public cloud object option. So in general, it's a competitive solution. And in the tape, what we find is customers, and I was just with a very large EDA customer, when they're using Iron Mountain and others, when they send off the tapes, when they need something, it's very logistically inconvenient, which tape to get back. You know, it's a very reliable technology, but which tapes to get back, how to do that. So price point wise, we are very, very competitive as far as tapes are concerned. Yeah, right, I mean, using tape to, you hope you never have to restore it. Yeah, but you have to for regulatory reasons. Yeah, but sometimes you do. Okay, I wanted to ask you, CJ, when Michael was on stage this morning, he showed the Dell Technologies portfolio, there were seven companies there, and he made this statement that it allows us to act like, even with a large company, we can behave like a startup. We can be agile. And it's interesting because you had some companies like RSA up there, and I don't necessarily think of RSA as an agile company, but I do think of your division as an agile division. So it must have been interesting for you to see that and say, well, in many respects, you're like the startup division. You get products, we've talked about this before, to maturity, like you did with Extreme I-O. Yeah, old-scale I-O, ECS, yeah. And then you hand them off to, you know, to guy, go. He takes them, then you jump an S-curve, and is that an accurate description of the role of your organization? It's a great question. And I would say right now, that is my focus. My focus currently is one on the unstructured data and how we solve that problem efficiently, especially when we look at our competitive landscape. But, you know, make no mistake, when we look at this incubation projects, even Scale I-O was an incubation project, ECS, was brand new written when I took over that division, and DSST was pre-alpha stage. So to answer your question, yes, that's the goal. The goal is make the businesses at a certain scale. Do we move them to different organization? Hard to tell, depending on the business, because I'm also focused on platform three workloads and web scale and others. So, depends is the answer. Yeah, so, and the last question I have is, I mean, you're not going to tell us, you know, exactly what you're going to announce in the future. Of course, you can't do that. But one could surmise, okay, if you find a problem, chances are they're going to call you to help figure out how to solve it, either organically or acquiring companies. And one of the problems that you're seeing customers complain about these days, they go to so-called platform three, they might be using, let's say, Amazon even, web services, but there's different APIs for all the different services. You know, whether it's hot data, Kinesis as an API, DynamoDB as an API, S3 Glacier, and there's a lot of complexity. Is that the type of problem that you guys are thinking about in terms of solving in the mid-term or long-term? I would say, we are going into too much details, hybrid cloud and what makes it easy for our customers to consume our technology. Like, we saw this need where customers said, can I tear from my salon to a cloud's provider? And of course, we would like that to be ECS or we would like that to be virtual stream before it goes somewhere else. And then how efficiently we do that, do we do tighter integration between object and file so the hybrid cloud becomes real? So, very focused on adoption of cloud technologies and how we can fit in by using where we came from, which is on the storage side for the most part. And then on the conversion stack, combining with servers so that it's easy for customers to deploy, even if they're doing a private cloud or a hybrid cloud, is the area of focus. Great. All right CJ, thanks for coming back in theCUBE. Always a pleasure to see you. Always a pleasure. I love the cadence of the announcements. I'm sure we'll see you in the spring at Dell EMC World. We'll have more announcements. Thank you. All right, thanks. All right, keep right there everybody. Stu and I will be back with our next guest. This is theCUBE. We're live from Austin, Texas at Dell EMC World and we'll be right back.