 Live from Las Vegas, Nevada, it's theCUBE covering EMCworld 2015. Brought to you by EMC, Brocade and VCE. Welcome back to theCUBE. theCUBE has been called ESP. Live continuous coverage from EMCworld 2015 here in Las Vegas. I'm Stu Miniman, joined by co-host Steve Chambers. We're with Wikibon. Find all of our research on wikibon.com. Having a regular back on theCUBE. Jack Rendoni, who is the VP of Storage Networking. Of course, talking storage ready EMCworld. Jack, welcome back. Thank you. It's great to be here, Stu. I'm glad to see I'm a regular now. That's awesome. Absolutely. I'm moving up in the world. We're going to take a little bit of a different interview with you this time. We've talked a lot about traditionally storage networking. Many people equate to Fiber Channel. That's right. And of course, Brocade's the industry's leader in Fiber Channel. Fiber Channel's doing great. We're going to kind of put that aside for this interview and talk a little bit about the IP networking to start with some of the flash storage that's doing there. So introduce us to what is IP storage networking? I mean, isn't IP just IP? Yeah. So that is the question, right? Because for the last 15 years, that was the story, right? And you hit the nail on the head in the sense that I was in the past, the last 10, 15 years, if you had something that was important, you put it on Fiber Channel. We're going to stay in Fiber Channel. It's going to continue to go along. We're all happy about that. But then what we've seen, and then when it wasn't really that important, you put on your IP storage, right? And think of it as like really mission critical type of work loads. You put it on Fiber, if it's file sharing or something, you can get to it in two or three days. Yeah, you put it on IP, whatever, right? And thus, the network didn't matter because the workload didn't matter. So the change now, what we're seeing, Stu, more and more, is that people are putting high-scale VM stores, they're putting analytics, and those are workloads that are being driven apps that are purely for IP types of storage. So what we're seeing more and more with our customer base is as they deploy these important workloads on IP level storage, they're asking more and more questions about the network. They're starting to realize that the network actually does play a role in terms of the overall level of reliability, performance of the environment. So Jackie, I mean, I lived through this for many years. I mean, one of the biggest challenges you had, originally it was like, okay, do I go Fiber channel or NAS? Yeah. And one of the challenges was if I needed performance, and I went to the network team, it was a three-letter word that was a four-letter word, it was QOS. Oh my God, no, I don't have time to adjust that. I'm just general purpose network and I'm not going to take care of that. So the storage people said, forget it, if you're not going to take care of it, I'll just own it, and therefore the storage network was kind of there. And that exact same dynamic we see happening in the sense that QOS is still a bit of a four-letter word. And when the storage buyers are buying these new arrays, new systems, we're in these workloads, what we're doing as part of our strategy is saying, well, let's give them an IP network that feels like a, it performs like a simple, high-performance storage network. In fact, we're designing features and capabilities that are specifically designated for a storage person to the point where our Ethernet fabric switch, called the VDX6740, has been branded by EMC. And it's positioned by EMC as an IP storage switch. And so the idea then is that dynamic that you're talking about, complexity, politics, whatever it is, you basically give the storage buyer the ability to go, deliver on the SLAs that they're going to be held accountable to and do it within an environment that they don't have to learn protocol exotica. Like if you ever look at Ethernet data sheet, and we make big, beautiful Ethernet switches, so don't get me wrong. But Ethernet needs to do so much. Fiber channel is relatively simple because that's really all it does. And so our IP storage... I'm sorry, Jackie, did you just say that Fiber channel is pretty simple? Relatively speaking, absolutely. I mean, how do you answer the question? I mean, we're not going to talk about Fiber channel, but the general commentary in the industry is that Fiber channel is this dark art that storage people do. It's not like, oh, well, Ethernet's simple. Ethernet's everywhere. It's ubiquitous. Is that just not true? If you want to do a couple ports at five terabytes, you can do whatever you want. But you want to be a petabyte scale, complexity, and variance leads to stability issues, leads to all these problems. Fiber channel from a protocol perspective is very, very lightweight just because it has a single purpose, really. Steve, do you want to chime in on any of that? What's your thoughts? Well, I mean, I think about, you know, the times when I've had to do zoning and... Yeah? We've been talking today about performance and, you know, immediately you think IOPS and things like that, but increasingly we've heard from customers today that performance for them is the time it takes them to actually get something done. Absolutely. And I think once Fiber channel's in place, then, yeah, I get it. But I think what Steve's referring to is, you know, I think there's an impression that Ethernet is completely different. You just plug it in and away you go. But I think you're saying there's a little bit more. You know, it is. And I think, again, when the workloads are important, you're doing it high performance and high scale. The network matters. Fundamentally. I mean, the network does matter. And what we're doing then is with our Ethernet fabric technology, is building the feature sets, right? So you maintain that level of agility that you'd expect from Ethernet, but then you get the performance and the scale of some of the automation that you'd expect, right? Specifically for storage types of workloads. And we're seeing good traction in the market with this. You know, EMC as a partner, you know, they're branding it. Connectrix IP switch, or IP storage switch. And so we're very excited about that in terms of our position within the market to not only build upon our Fiber channel heritage, but address file-based solutions. Even based, you know, object, right? You know, some large object clusters we're looking at, as well as even hyper-converged infrastructure type of solutions there, right? So you get to a larger scale, the network starts playing a more and more important role. So it's very exciting for us. And by the way, the way we're attacking this too, I'm Steve, it's not just a single product. So we're coming in with a, we announced last week a disaster recovery business continuity solution qualified for large scale IP storage environment, right? We already did that in Fiber for years. We're doing it for IP today, right? We're doing it within our management software, right? So we have a single management interface, broken network advisor, whether it's Fiber channel or IP, gives the customer data and information for a storage professional to make important decisions about their storage network. So what you're saying is, you know, right at the top, you were saying that, you know, in the olden days, if you like, you would prejudice your workload on to Fiber channel if it was really important, less important. But I think what you're saying now is, if it's a really important workload, you're still getting all these great enterprise features on Ethernet as well. Exactly, exactly. And in fact, the matter is, you guys are right, a lot of the new workloads are being put on Ethernet, right? And so, and they expect levels of performance and reliability, right? And that's what we're delivering with our IP storage solutions. So, Jack, you know, one thing I want you to dig in here for a second is, you know, it sounds like when we're doing an IP storage network, it's not a general purpose network. It's something that's going to be used, you know, dedicated to that environment. So, in many ways, it's going to look like a Fiber channel network. You know, what I want you to kind of address there is, you know, does that defeat some of the purpose of going to, you know, an IP network? I thought it's supposed to be, you know, same equipment, same skill set, you know, same gear, you know, can I share some boards? You know, utilization is still, you know, really low for most equipment in the data center. Can you kind of address that challenge? Yeah, so I think, you know, a number of companies, so just put brocade aside, right? There's about, you know, 10, 15 companies that we've seen out there, device manufactured, workload that say, basically, if you're running certain levels of scale in I.O. that a dedicated IP network is the best practice. So it is the best practice in the industry. We've talked to analysts, we've talked to big customers, like, we've been doing that for a long time, right? Yeah, absolutely. And I study that was, like, the way we did it going forward. Absolutely. But we can't get beyond that. It's, you know, architecturally, there's not something that we can do. So, no, no, so we're there, right? So you say, now, once you're separate, then the question is, is what does that network look like, right? Does it need to be a, you know, if it's for VPLS, MPLS, and broadcast, multicast, you know, all these types of things? Absolutely. I mean, there's a number of those capabilities within our VDX that needs and switches needs and that switch. But what we're adding onto that are specific storage-related capabilities, right? And FabricVision, for example, is a deep instrumentation level of service that we brought in the Fiber Channel. We're bringing that over to our IP network. The only people that are going to appreciate that are actually storage people, right? Really what it comes down to. So, yeah, it's an ethernet switch. It'll do your ethernet, you know, course stuff. But then the built-in multi-path and the built-in management, you know, we think has a special appeal for storage professionals. So, you know, we saw a really interesting keynote this morning, you know, the beast was unleashed today, right? Yeah. And the performance figures were astounding, and we've heard from some customers today that it just changes the whole equation for them. Yeah. You know, I think Stu's already said to a few people his performance no longer, he's no longer, it just takes it off the table, right? Yeah. But one of the things that stood out to me this morning when we saw the keynote was, you know, you need to keep feeding the beast. And I think, well, if you've sold performance there and this thing is so hungry for data, then is it going to change your network architecture to get to it? Or can you just plug it into the same old thing you had? You know, do you, you know, is there, do you need new features? Do you see a different architecture required to have an old flash array at the end? Do you know how does it affect you guys? Yeah. So, for us on the Fiber side, so most of the connections we've seen coming out of the all flash arrays have been fiber channel. Right. And, you know, our fiber channel products are very low latency. We measure in hundreds of nanoseconds type of stuff. So it's really, those are kind of peas in a pod there, you know, in terms of the perspective being able to handle it. What we do now is we're adding just more of that instrumentation, like I'm saying, to make sure that you can bring these all flash arrays into a shared sand and not affect anything else as well as get the performance you expect. It could be one hell of a noisy neighbor, right? Yeah, right, right. So, that's where we add that deep instrumentation, right? And then on the VDX fabric side, one of the key things we're doing a couple of things, is beyond that deep storage-related instrumentation, we have very low latency as well, in hundreds of nanoseconds type of capability. And then also, we have the ability to go in and take buffers that are associated with a number of ports, pull them together and get a little bit more of a kind of performance optimized for a large packet kind of storage workloads, right? That plays well also within the flash environment. So, it's interesting because, you know, for way too many years, we were arguing over, you know, how much bandwidth I could get for Fiber Channel versus Ethernet, and it's like two-fourths, you know, eight-sixteen versus, you know, one to ten to forty to twenty-five or things like that. Latency is, you know, something that, you know, Fiber Channel's had great latency for years. Absolutely. Can you talk on the Ethernet side? You know, what are you guys seeing? What are you driving? I mean, things like Rocky and IWARP are coming up in a lot of conversations, you know, so often it's Ethernet's doing there. What's your case positioning? Yeah, so from a pure switching latency perspective, like I said, we're in very good shape, right? We're hundreds of nanoseconds type of storage heap, and Infinity Band's always been kind of in that area of ultra-low latency, and then it's got the benefits of the RDMA capability and stuff. I used to work in our Infinity Band a long time ago, right? When it was going to take over the whole kind of, you know, PCI world, parallel PCI, right? So, um... 3GIO, right? Yeah, yeah, exactly, exactly. And then HyperTransport came out and everybody freaked out. So, um, you know, I think what we're going to see with IWARP and Rocky is really take a run in Infinity Band in terms of kind of where it's been able to go and be that kind of back-end cluster kind of interconnect. And so that's really where we see that making probably the most inroads overall, because it's combining not just the latency, but all the off-load type of capabilities as well. So we spoke to a customer from Australia, and he told a great story about having dual data sensors. Yeah. Using your technology, Ethernet and, you know, EMC technologies to have two hot data sensors, and he can just shut one down, and the workloads are always moving between them. He doesn't have to do DR anymore. Is there anything specific in your switches that enable that? I guess that's the first question. The second one is, do you see that happening a lot? I think he said he was the first customer in Australia to do it. I mean, it just sounds so appealing to me, right? Take that headache of DR away, you know? Yeah. And he says that you're a key part of that. Yeah, so we have a couple capabilities in that front, and what we announced last week was, so we build a data replication fabric that's really good for heavy-duty storage, and we do a number of things that optimize those connections between the data centers for large-scale storage. We optimize the TCP windows over storage. We add multi-pathing, excuse me, failover and load balancing into the WAN links, which are notoriously unreliable. Right, okay. And then we even add some kind of WAN testing optimization so you can test your apps when you lose some WAN links and kind of see how they perform, right? That's interesting. And so the interesting thing is, and we're doing that both for Fiber and IP. That's why I'm very clear to say, you know, I'm the storage networking person, right? I don't fiber-tail networking guy. And so that kind of capability gives those customers that ability to go handle it, and the important part is you handle it at scale. Because if you're doing it at large scale, that's when that stuff really starts to matter. You don't want it to be fragile, right? I'm sorry? You don't want it to be fragile. You want it to be reliable. You want it to do it, as you say, at scale. We deal with mission-critical, business-critical type of environments, right? That's where we, you know, that's our forte. All right, so, Jack, to close the interview, you know, here at EMC World, talk a little bit about how EMC's been getting involved in the IP storage networking piece with Brocade. Absolutely. So, of course, you know, EMC is a leader with VNX, Extreme Mile, Icelon, Data, I mean, you name it, right? It's even scale IO, right? And it's solution, you know, for software-defined storage. And really, what EMC did is they looked at our technology offerings, our VDX, Ethernet Fabrics, and said, this solution will couple to their storage, really, kind of like how Fiber Channel makes that total solution better. And basically, what they did is, and you'll see on the show floor, a number of solutions where they've taken our technology branded it themselves, positioned as IP storage networking, right? IPSN, I like to call it. And are showcasing it as part of these solutions, right? Sometimes it's hidden in as part of, let's say, VSpec's blue type of architecture. And then sometimes it's right there, sitting right next to the VNXC. So, we're very excited about this show in terms of what it means for the partnership as well as for Brocade. Alright, well, Jack, we're going to have to leave it there. You have a huge partnership with EMC and Brocade, of course. Diamond Sponsor also, Brocade's one of the lead sponsors for the EMC on the Charity Water, which is happening right over here in the hang space so thanks so much for joining us. We'll be right back with lots more coverage here from EMC World 2015, right after this quick break.