>> Voiceover: Live, from San Francisco. It's theCUBE. Covering Pure Accelerate 2017. Brought to you by Pure Storage. >> Welcome back to San Francisco, everybody. We're at Pier 70, one of the oldest piers in San Francisco which is not long for this place. It's going to be torn down after Pure Accelerate. I'm Dave Vellante and this is Stu Miniman, my co-host. This is theCUBE, the leader in live tech coverage. Kent Petzold this year is the enterprise storage manager at Intermountain Healthcare and Vic Nagjee is back. He's the CTO of Healthcare for Pure Storage. Gents, welcome to theCUBE. Good to have you. >> Kent: Thanks for having us. >> Dave: You're welcome. So Kent, let's start with you because we talked with Vic a little bit already but tell us a little bit about Intermountain and your role. >> So, Intermountain is the biggest healthcare provider in Utah. We've got 22 hospitals, 185 clinics. My role there is, I manage the storage team. We've got eight petabytes of usable storage that we manage. Do lots and lots of backups. You know, all things data protection is under my purview as well. >> Now, have you always been a healthcare you know, practitioner, or is this relatively new for you? >> I've been at Intermountain for 24 years. >> Okay, so that's enough... To qualify you as knowing a little bit about healthcare, and so, my question is, relative to sort of other industries what's unique about healthcare? I mean, obviously it's highly regulated. You've got serious privacy, but you're dealing with, you know, many businesses are dealing with dollars and cents. You deal with a lot of budget, but you also deal with lives. Talk about some of the differences of healthcare and the particular stresses that puts on I.T. >> One of the big things is just doing updates of your technology. Because we deal with people's lives we have to be careful about when we do updates. You know, we've got to be cognizant of you know, "Is the emergency room full?" things like that, so it puts an extra challenge on us for when we need to take systems down to do updates. >> So that means, yeah because updates means downtime. >> Yeah, in the past, yes. >> That's not the case with Pure? Tell us about that, Vic. >> Kent: (laughing) >> Okay, so. Maybe, actually tell us about that a little bit. So, if you guys make a big deal out of it, last segment I turned it into dollars and cents because, on average, a migration, a RAID migration is a minimum of $50,000, minimum. In healthcare, it could be lives. >> Yeah, I mean in healthcare it's definitely lives but it's also a little bit more expensive because this is specialty data. So, the minimum you're looking at is about $1,000 per gigabyte. >> Dave: Per gigabyte? >> Per gigabyte transitioned over. Depending on the kind of application you're dealing with. In this particular case, you know it's more than just the expenses like you mentioned. It's interruption of care, interruption of service, which is not acceptable. So, the technology that we have and the architecture that we have allows us to go in to healthcare organizations such as Intermountain and say "You know what? You can have an environment that's "going to get better with time, because we're going to be able "to come in and not only upgrade your software, "we're also going to be able to come in "and upgrade your hardware and keep you on the tock cycle "every three years, update your controllers, "and so on and so forth with zero downtime." And what we're seeing is this big shift in the healthcare industry where, you know, Kent can relate to this. Typically we have these updates all teed up and lined up for three o'clock in the morning on some obscure weekend day, right, where if something goes sideways the number of experts you can reach are very very low and now we're seeing a switch with this kind of technology to actually have people say "You know what, two o'clock in the afternoon on Tuesday? "I'm there. I'm doing it." >> Okay, so Kent. Take us through sort of your journey here. Sort of give us the before and after of Pure, what problem you're trying to solve, and how you solve that problem. >> So, we started down that with our insurance arm Select Health. We were getting calls pretty much every week. Sometimes two and three times a week for slow issues, and, you know, we're looking through logs. We're doing our monitoring and stuff and it was continuing and my architect was spending hours and hours every week >> Dave: Fun. >> trying to research this. So, we started looking at flash vendors. Pure was one of the only ones that came in, gave us the documentation we wanted, was able to answer the questions we had about our environment. It was a sybase database. AIX with some kind of weird settings, and we started testing it. We liked what we saw. We moved along, finally put it into production. They haven't called us about slows since we put it into production over there years ago. >> This is three years ago? >> Kent: Yeah. >> So it was really a performance issue you were having with your traditional apps, and you said you dropped in Pure Flash array and the problems just disappeared. >> Yeah, we haven't had any calls about slows since then. >> Dave: And if you had to sort of increase your capacity of the Pure system. >> We'd increase the capacity. In fact, because our three years was up we just did a head swap on them and added a little more capacity, and that went flawless. No outage for the business, and they were very happy about that. >> So as long-time storage practitioner... what's the difference in terms of... What difference does it make to you when you bring in a system like this? >> Some of the older systems to like do the head swap and get the new controller is weeks and weeks of planning and making sure you understand what's on their, what needs to move, what can take down times, what can't. I mean, there's a lot of planning that goes into that when you know there's going to be a disruption. So, with systems like Pure, we don't have to do as much planning. We still do a little bit so that we know what we're getting ourselves into and what's going to be at risk, but it's a lot less. There's no... >> So, Kent, how are you tracked by the business? What are kind of, do you have any measurements or KPI's that they look to you. We talked about uptime before, but, you know, how're you tracked, and how's that changed in say the last few years? >> It's changed quite a bit, cause we're not having to track, especially in our tier one apps that are on Pure we're not having to track the performance as much. So we're able to re-look at what our KPI's are, and come up with ones that are meaningful for us. And really, with the simplicity of it, it kind of helps us to become more of a trusted advisor to our business and be able to help them solve their problems instead of continually pulling knobs and fighting fires. >> Vik, I'm curious. How do you help the storage administrator today? I remember, Pure used to have streaming on its website. Certain data points from customers. What are you seeing today? What's helping them shift what they're working on, get more done with what they're doing? >> Kent: Yeah, absolutely. And just to come back to that and echo the point here Kent just made, essentially we're seeing the successful organizations in healthcare and possibly other verticals too, but I live and breath healthcare, right. So, healthcare. I.T. organizations that are able to make the transition to a trusted advisor, to a partner to the business are really making those leaps ahead. In terms of better patient care outcomes and also cost mitigation. Now, in terms of what we offer, right. So, it's the simplicity that's at the heart of everything. Once you set it up and you basically it's like Ron Popeil used to say. "You set it and forget it." Right? You have that experience. And then, it's not so much about having practitioners say "There's black magic going on "and we're going to just trust it." We have to build a transparency in there, and we have to demonstrate that at a glance, single pane there's answers to all of the questions and more that they might have. The telemetry that we're getting off of these systems allows us to do things with machine learning and AI and a lot of business intelligence the backend to be able to say "Hey, over eighty-some percent of all "of our problem tickets that are ever opened "are opened by Pure on behalf of our customers." And say "Hey, you have something that's demonstrating "a characteristic that is similar to what we've seen "across the world, somewhere else, "and you might run into a problem, "so let's just go resolve it." >> So, Kent, one of the things we've been poking at and they talked about in the key note this morning is how do you get more value out of your data? We talked about in an earlier segment with Vik. How do you look at your data? How are you sharing with other organizations or leveraging data internally better? >> Kent: Umm... >> Or are you? >> We've got quite a bit of data, and we're starting to go down the genomics road, and with that data we've got some good opportunities to be able to make some good advancements in healthcare and how different diseases are treated. So, we're kind of excited about that, and that's one of the areas my team's been really helping out, and being a trusted advisor to our genomics group. To get them set up with the things they need. >> You guys are talking on stage today about how backup and data protection is changing. It used to be kind of disk to disk to disk, and then sort of flash to disk to tape. Well, tape is still somewhere in there. You know, whatever, maybe it's the fourth level. You guys are talking flash to flash to cloud. We were talking off camera, Kent. You said "We're kind of looking at where to put "the right cloud workloads." Is backup one of those? >> Backup is possibly one of those. We talked a lot about how we off-site. Right now we still use a lot of tape. One of our key things that we think about when we're thinking about cloud and like off-siting stuff so we want to make sure we put it somewhere that, if we have a disaster, we can spin it up in that place. We're not trying to bring it back and bring it somewhere that is impossible during a disaster. So, we want to put it somewhere, and we want to be able to use it there and not just have it sit there and say "Yeah, we've got data protection. "It's right there, but we can't use it." >> Dave: Yeah, yeah. Can't recover. But, I mean, tape is still pretty prevalent in healthcare, right? It's a compliance issue,right? >> Vik: Very much so. >> I mean, your auditors aren't going to let you just throw away tape, right? >> Vik: Yes and no. I think it's just more of the "It's worked for so many years." Now, the problem that we run into is with the things, and we touched a little bit on this in the last segment. We talked about security, right? And sort of, in terms of insurance and protection against any of these threats that are malware et cetera, that are coming up, is getting more and more important for folks like Kent to prove to the business that "Hey, we're not only backing this data up "but we're restoring it. "We can restore it, and it's good." And we know how long this takes. So, all your iTell stuff comes into play. You have your SLO's. It's all back on. Try doing that with tape. Try doing that with tape that's been archived off-site. >> Dave: No, you can't. (laughing) >> And so this is why healthcare's actually moving in the direction of saying "You know what, let's just forget about that. "Let's just try to find different, better, faster "cheaper media if we can actually apply all of "the principals from today to do that." >> So you might still have tape, but you just never use it. Or you pray you never use it, just to have it there just in case. It's like that fire extinguisher in your barn that you don't know if it works or not but you have it there. >> Vik: It's there. It's good. It looks good, right? (laughing) >> Okay, and so, if you think about the experience that you've had with Pure. I told you I was going to put you on the spot, so are there things that you would do differently if you had to do it over again? Advice for your peers? Things that are on Pure's to-do list that you'd like them to do that'd make your life easier? >> I mean, yeah there's things that are on their to-do list. I mean, and I think they're announcing some of those today so that's probably pretty good. We want to do more with replication. Obviously, as a data protection, you need that. We'd like the price point of the M's to go down a little bit because there's kind of this misnomer about tier one storage and "Do I put my dev on tier one." Well, there's huge opportunities with cloning and things like that, and some of the partners that Pure has that we can actually bring up dev environments and not use as much storage as what we're using today. >> So that's a data sharing capability that you can give access to current data to your devs and not have to spin up multiple copies and separate infrastructure. And the use case that we talked about before was an enterprise data warehouse, right that you were trying to speed up. How about this, you heard from Scott Dietzen this morning the big push on analytics. Is that something, certainly your industry is pushing it. Is your organization there yet? Have you dipped your toe into the big data lake yet? >> Yeah. We've been doing analytics for a long time in one way or another. It's just, we're just getting more and more pressured to have the data available so they can continue to do that. >> Dave: Are you throwing Pure at that problem or is that... >> We hope to. Over time. We keep adding to our environment. >> Alright, Vik, we'll give you the last word. >> Pure and healthcare. What's the bumper sticker? >> Yeah, before you give me the last word I mean I think Kent's underselling what Intermountain's been doing in terms of analytics >> Yeah, add some color to it. >> over time, right? So, basically, they have been one of the pioneers in terms of really understanding drawing value from data. >> Really? >> Yes. It's been over time. It's been very much so of "I have this old data. I want to go run analytics on it. "Then I want to do some BI on it." And now we're getting to the real-time near real-time insight on data that really matters. And for that, we're hopeful that we're going to have an opportunity to actually participate and help build out those sorts of frameworks. And Intermountain's one of the organizations that's lead the way. A lot of the other organizations sort of following in the same footsteps. And, you know, right at the end, all I have to say is all of the benefits that we've talked about and we've talked about... We talked about across verticals and just horizontally in general that the Evergreen model brings to bare from Pure. I think they're really heightened, in terms of healthcare. So we talked about uptime. We talked about six ninths of uptime across our arrays And we're counting planned maintenance as part of your runtime. We're not saying exclude those, right? Very important. No data migrations. Super important. >> Dave: Downtime is downtime. >> Downtime is downtime. Exactly, thank you. Data migrations are super risky. Not only are they expensive, but they're risky. If you talk to any CMIO or CNIO and you say "Hey, how do you feel about your data being "picked up from here, put over there." See their reaction. >> Dave: It hurts. >> And they're expensive. And then the simplicity aspect of it. The simplicity is sort of at the function of the heart of everything. Its power is through simplicity, really is what it is. Giving him and his team and his organization time back to be able to go back and say to the business "How can we make your life better? "How can we make patient care better, "and how can we improve on resources?" >> Okay, good. Actually, Kent, we're going to give you the last word. Pure Accelerate 2017. Good event. What are you learning? Anything exciting? >> Kent: It's been a great event so far. Love the announcements. I just love being in this type of environment, because there's such a vibe here of wanting to help people do things and it's really great to be in a place like this. >> Dave: Yeah, it's fun too. We've got Snoop and... Snoop with the multi-cloud. That's an inside joke everyone. >> Vik: Multi-cloud. Are you sticking around? Are you sticking around for that tomorrow? >> Yeah, I'll be around. (laughing) Alright, good , we'll leave it there. Thanks you guys. We really appreciate you coming on. Okay, keep right there. This is theCube. We're live from Pure Accelerate 2017 in San Francisco. We'll be right back. (techno music)

>> Hey, welcome back everybody. Jeffrey here with theCUBE. We're in Palo Alto, California at the Four Seasons Hotel. An interesting event, it's called Security in the Boardroom, and it's part of the security series put on by the Chertoff Group. They do a couple of events a year, and they've returned to the Four Seasons. It's really an interesting twist on the whole security discussion, really elevating it to what's happening in the boardroom. We're excited to be here, we've got some great guests lined up, and we've got our first guest of the day. He's Bob Griffin. He's the CEO of Ayasdi. >> Correct. >> Welcome, Bob. >> Thanks. >> I got the pronunciation right, so. >> You did, indeed. >> For people that aren't familiar with the company, what is Ayasdi all about? >> Well Ayasdi's an artificial intelligence platform manufacturer that builds technologies that allows us to effectively deploy enterprise class artificial intelligence applications. >> For security's specific application or beyond security? >> Yeah, beyond security. We're fundamentally focused in three areas. We're focused in the financial crimes area, specifically around doing things like anti-money laundering, risk and compliance, waste, fraud and abuse. We're focused a lot in the healthcare area, around doing things like, clinical variation management, population health risk, and we've got a very strong focus in the federal government and the public sector, mostly around the intelligence community, DoD and so forth. >> Okay. So, financial institutions, the government, and then who's the purchaser, what's the segment that buys your healthcare focus applications? >> It's traditionally both the payers and the providers. So folks that are looking at, how do we manage costs associated, but how do we make more use of healthcare practices? So, folks like Mercy Hospital, folks like Intermountain, United Healthcare, folks like that. >> So it's interesting 'cause there's a lot of talk of machine learning and AI right now, it's hot, hot, hot like beg-id was a couple years ago. But I think, a lot of people are still confused as to how is it actually being used. Is it actually being used? It's probably affecting them in ways they have no idea. So, how is the adoption of AI progressing from your point of view in these industries, and how is it helping transform them? >> Well, it's absolutely transformational technology. The reality is all applications eventually are going to have to become intelligent or they become obsolete. The biggest challenge with artificial intelligence is that it's moving incredibly quickly. The rate of change, milestones, are daily. So if you're not running to artificial intelligence applications, or developing and deploying those, you're behind the curve. If you're sitting at the stoplight right now, and you're competitors are entering the intersection using artificial intelligence, you're never going to catch up, so you have to move quickly. >> Right. >> The second thing, I think, is that, artificial intelligence now has got an opportunity that can really focus and help with real business problems. Traditionally, what we've done with artificial intelligence is we've parked it in innovation labs, or we've parked it in R&D. It's time to take it out of that and really put it to place, in areas around opportunities we talked earlier about. Anti-money laundering. How do you reduce the number of false positives to make your 5000 investigators more effectively? Artificial intelligence can do that kind of application. >> I was wondering if there's any stories you can share publicly about some of the big impacts or maybe little impacts that people would never have guessed where you can apply this type of technology to positive outcome. >> Sure. So, let's talk a little bit about, let's take anti-money laundering as an example. We have a client that has nearly 7000 investigators. And their challenge is, they're getting almost 98% false positives. They came to >> 98% false positives? >> 98 false positives, I mean think about that. >> Which is crazy. >> Out of every hundred, only two positives are actually effective. Alright so, they came to us and said, look, if we can reduce our false positives by say 3-5%, that's a home run for us, right? What do you think you can do to help us? We took their information, their data, put ourselves within their workflow. And we we're able to give them a 26% reduction in false positives. Well that changes the game for them. Just the economic savings alone is incredible. You're talking nearly 140 million dollars. So, those are real things. I'll give you one more example in the healthcare area. We've been studying type 2 diabetes for nearly 40 years. We took that same data set that people have been studying and working with one of our partners, we were able to very quickly, through our platform, segment up that data set and show that type 2 diabetes really falls into three subsegments. And those subsegments are really indicators of what's likely to happen to patients, but more importantly, they subsegment up into things like, these clients, er these patients that have these conditions are likely to develop cancer. These clients are likely to develop retinopathy, blindness. What that's doing is it's changing the way, not only they're going to prosecute a cure, but also the way they're going to prosecute the treatment of type 2 diabetes. It's changing the game. >> So, it's interesting. You got a technology platform. Do you also deliver the data to scientists? How does it work in terms of, or are you a tool that you hand to data scientists inside the organization, the one you just, given an example of and gives them a different tool, or you also delivering services to help refine and tune? 'Cause obviously it's always implied that these things, not only do you pump the data in, that there's a continuing ongoing process of learning as they, continue to get smarter. >> Absolutely. The answer actually is yes. We provide a platform, and that platform really comes with capabilities to enable our clients to develop artificial intelligence applications in real time or near real time. So, it has things like an SDK, it has REST APIs, but more importantly, it has a tool we build called Envision. And that Envision really allows our clients to very rapidly prototype new artificial intelligence applications and get them into production incredibly quickly. Now to your point, there are, some of our clients that don't have the technological skills or prowess, but yet, need to take advantage of the technology. So we have a professional services capability that will come in. We'll bring in data scientists as required. We'll bring in subject matter experts as needed. We'll bring in program managers and so forth, and we'll take them from kind of, cradle to grave, in helping them build out those applications. As part of that we'll train them, educate them and let them to become self-sufficient. Because, one of the things that I think is incredibly important about artificial intelligence that nobody's talking about, is any machine-intelligent application has to be able to do five things. It has to be able to discover. You know, find out and do observational discovery. What does it not know about itself, What can it learn? And that's important, because if you can do unsupervised discovery, then you can do the next thing, prediction, much more effectively. So it has to be able to discover, it has to be able to do prediction, from the past we can predict the future. It has to be able to do justification, and that's probably one of the most important areas that we talk about. Justification is not necessarily what is it the algorithm did, but why did it do that, why did it take that action? Why did it segment the population to these sizes? What is it that it proved? Why did that sensor go off? And so forth. >> This is really, to kind of, unveil the black box a little bit. 'Cause nobody wants the white box anymore. >> Absolutely. And then lastly, it's got to be able to do two additional things. It's got to be able to act on what it has discovered, what it's predicted, what it's justified. And then lastly, it's got to be episodic, it's got to learn. So what did I learn from the last episode, and how do I apply that back to a new form of discovery, a new form of prediction, the next level of justification and action. >> That's a great summary, Bob. And it's interesting. 'Cause you guys talk a lot about, I was doing some homework before I came in on the justification piece. You got to open up that black box, it's no longer good enough just to kick out an answer. >> Absolutely. And if you can't on it, what's the point, you know? It's kind of more of a science experiment. Before I let you go, we're running out of time, but, the roots of the company, is around this thing called topological data analysis. And you're not a data scientist, nor am I, but conceptually, what was different about that approach, that people weren't doing previously? >> Well so, topological data science, data analysis, is the study of the shape of data. All data comes in shape. The challenge historically is most people apply traditional algorithms to data assuming that it's going to be in a linear fashion, for example. So they'll linear regression analysis. Or if it's clustered data, they'll apply clustering technologies and so forth. The challenge is, what happens if your data is in a flare shape? Or what if it's in a circular shape? Or what if it's time series based and so forth? What we do is, with TDA, the first thing it does, is we understand the shape of the data 'cause the data will tell you a lot about itself and its shape. And from that shape you can start to ask more intelligent questions about the data so you can unlock all of the insight. >> So it's really almost like, a higher order organization if you will. 'Cause we always look for patterns, right? That's what we always do as people. Alright, well Bob, really interesting conversation. >> Thanks. >> I really look forward to the next time we get a chance to sit down. >> I appreciate it. >> We'll have to leave it there for now. >> Alright, appreciate your time. >> Alright, Bob Griffin, he's the CEO at Ayasdi. I'm Jeff Frick, you're watching theCUBE. We're at the Chernoff event, it's called Security in the Boardroom, we'll be right back.