 From the SiliconANGLE Media office in Boston, Massachusetts, it's theCUBE. Now, here's your host, Stu Miniman. Hi, I'm Stu Miniman, and welcome to theCUBE's Boston Area Studio. Happy to welcome back to the program two CUBE alums. To my immediate right is Peter Smales, who's the CMO of Amanant's Data. And joining him for this segment is Zeus Caravalla, who is founder and principal at ZK Research. Gentlemen, thanks so much for joining us. Thank you. Thanks for having me. So, you know, we go out to so many shows. We're talking about massive change in the industry. The last two shows I've gone to, really looking at how hybrid and multi-cloud are shaping up and, you know, change and just the proliferation of options really seems to define what's happening in our industry. And Zeus, I want to start with you because you've got some good research which looks at the data side of it. And of course, you know, I'm an infrastructure guy, but you know, the reason we have infrastructures to run my apps and the other reason we have apps really is behind the data and that transformation of data and data at the core of everything is something that we've loved to cover the last few years. So what's new on your world? Yeah, in fact, the word you said, their change is apropos because I think I've never seen a time in IT and I've been an analyst for 20 years and I was in, I was a CIO for a while, but I've never seen a period of change like this before where digital transformation is reshaping companies as fast as possible. Now, the key to being a successful digital organization is being able to take advantage the massive amounts of data that you have and then be able to use some machine learning or other analytical capabilities to find those nuggets in there to be able to help you change your business process, make people more productive, improve customer service. Whatever you're trying to do, I think it really stems from the analytics that data. Now, what my research has found is that companies are really, and this shouldn't be a big surprise, but companies are really only using a very small slice of their data. Maybe five to 10% of the most of their data. Most data is kept in what's called secondary storage and there what's happening is this concept called mass data fragmentation, where we've always had data fragmentation, but it's becoming worse where data's now being stored not only on local computers and servers, but also in the cloud, on IoT devices, out at the edge, within the organization, and so this concept of mass data fragmentation has exploded and it's hampering companies' ability to actually make critical decisions to be able to move fast and keep up with a lot of the cloud-native counterparts. And if they don't get a handle on this, they're going to want to fall in further and further behind them. I think it's absolutely critical today that this challenge of mass data fragmentation be solved. Yeah, Peter, I want to pull you into this discussion. You talked to a lot of users and we've talked to you at some of the Duke shows. We look at what's happening in the database world and there's so many options. I know our team members that keep up to it, they keep spreadsheets and they're trying to keep up with all of these, but seems like every week there's a new open source this and that that's going to capture this segment of the market. But something that I found interesting from one of the previous interviews we'd done with you and your company is it's not that I took my main vendor of choice and I went to one other, it's that today, the database world is like everything else. I'm using a lot and it is and, and therefore we know that has ripple effects for what I do for security and what I do for things like data protection. Can you give us a little bit of just kind of view as to what customers, why are they going to so many applications? What are some of the leading ones in the space and we know that in IT nothing ever dies and it tends to be additive, so how are they dealing with this? Yeah, and it picks up directly on what this was just saying before around this notion of fragmentation. So Amanis data, the genesis of Amanis data was really around if you look at it in the context of cloud. Cloud 1.0 was I have my, it was essentially, let me take all my legacy applications, lift and shift. Right, let's just take everything on-prem and let's put it in the cloud. Okay, people quickly realized that they were solving the wrong problem. The real answer to the problem was if I want to take advantage of all my data if I want to take advantage of hybrid cloud infrastructure, I've got to move from a traditional monolithic stack, application stack to more of a microservices based architecture. That led to a very rapid proliferation of new database platforms, both on the Hadoop side for big data as well as on the NoSQL side. So the synergy here and why we like this research so much is because Hadoop, the key message is that Hadoop and NoSQL have both become significant contributors to the mass data fragmentation challenge and that's really driven ultimately by digital transformation and organizations desire to move to a true hybrid cloud based infrastructure. Yeah, how does cloud and this data fragmentation, how does this all go together? Oh, a cloud and data fragmentation actually go hand in hand. People thought the cloud was actually solving a lot of the problems but in a lot of ways it contributed to it because as you said, we never get rid of the old. We keep the old around and we add to it. In fact, what I've seen happen is with so many cloud repositories now, users are storing data in the place they were before and they're making copies of it in these new cloud services. And in fact, almost all of the new collaborative applications have their own cloud repositories. So we've gone from an environment where we had a handful of storage repositories to manage to that absolutely exploding. And I think the cloud itself is mature. I think people are now starting to figure out how to really, to your point, use the cloud in a much different way than before. And so they're reliant on it. The companies are dependent on it but if we don't get a handle on where our data is we're going to wind up in a situation where it just becomes unmanageable. Yeah, and just to add to that from additional research is that according to recent research, 38% of interviewed companies had more than 25 databases. 20% of those same companies had over 100 databases. So the point is we've got to, there is a huge fragmentation issue. And if the problem you're trying to solve ultimately is insight to your data and intelligence on your business, you've got to create, you've got to solve this problem of fragmentation because otherwise you're never going to have any economies of scale. You're never going to be able to get visibility to all your data. That's ultimately the problem that needs to be solved. Yeah, it's funny because you talked about early cloud and people thought, oh right, I'm going to move everything there and I'll have one cloud, it'll be the cloud. The cloud. Yeah, things like that. And of course we understand, there's lots of reasons why I'm going to choose multiple solutions. But too many companies I talked to when you figure out how they got there it wasn't like they said, well, this is our strategy and we're going to do this and this and this. It was, well, different business units have different reasons. Just like I would build infrastructure for my various applications, I would have different groups with different needs and then, hey IT, can you help us bring all these pieces together? So how are we doing as an industry for helping customers get their arms around this? Is this just a mess today? Is there a waiver or a trend as to how we put together? Who solves it from a vendor standpoint and who from the customer standpoint kind of has the, is the champion of helping to solve this issue? Yeah, I think one of anything is unrealistic, right? And in fact, customers do want choice and they do an option. So it's not the industry's job to force customers to consolidate to one. In fact, it's better to let them use whatever they want. Now, where it becomes, where the work needs to be done now is creating that middleware layer, if you will, or that management layer that sits above the infrastructure that gives you the common view. So I think this mythical single pane of glass we've been searching for for so long, actually the cloud drives in that direction because we do need something to help us give that visibility. I know one of your partners, Cohesity, does that on the secondary storage side to actually make MDF or mass data fragmentation manageable. And there's other vendors that do that in other areas. But I think the concept here isn't to try and drive customers into selective choices, but it's to allow them to use whatever they want and then create a management layer over top that gives them that visibility to looks like one environment, but in fact, it's whatever they want to use underneath. Yeah, and picking up on that, the notion of, if you look at the, you asked the question about sort of who owns, who owns the mantle of driving all this stuff together. And the answer isn't, you could say, oh, the chief data officer. Certain organizations have gone to the level of saying we have a chief data officer and they're trying to drive towards a consolidated data strategy. That's a great idea, but sort of the federation of how things have evolved is actually, has been a good model. Like a lot of the folks that from an Amanda's data standpoint that we speak to, it's architects, it's developers, it's DevOps. And so from an organizational standpoint, what's happening is you've got to have, over time, you've got to have the application folks, the DevOps folks, the architects, the DBAs, get more closely aligned with your traditional IT and infrastructure folks that's evolving. And to this point there, that's not, you're not going to drive them all to one thing because they have different viewpoints and such, but you need to provide that common layer, sort of let them do their own thing, but then on the back end, be able to sort of provide that common layer to be able to eliminate the backend silos. Okay, and drill us down a little bit. We brought up the notion of management being able to see across these environments is a piece of the solution, but what is Amanda's doing? What are you seeing out there? And our caution, we know a single pane of glass to solve everything is kind of the holy grail, but reality is we need to solve real problems for customers today. Yeah, and our piece of the puzzle, our piece of the puzzle is Amanda's data is enterprise data management for Hadoop and NoSQL. That's where we focus. We're basically delivering industry leading solutions for Hadoop and NoSQL. That has led to a very logical collaboration with Cohesity who is one of the leaders in hyperconverged secondary storage. So they're trying to provide that common layer of infrastructure to address mass data fragmentation. We see that as we're the Hadoop and NoSQL folks, so there's a very logical synergy whereby the combination of Cohesity solution and Amanda's data solution essentially then provides, ultimately will provide that single pane of glass, but also again at the end of the day provides a common visibility and a common layer to all of your secondary storage, whether traditional relational, VM based, cloud based, whether it's your Hadoop and NoSQL based data. Okay, so bring us back to the customers. We know that simplification is something we want. The cloud world doesn't feel like it's gotten any simpler. So where are we? What needs to happen down the road? What more can you share about customers? Yeah, I think that's fair to say it hasn't gotten more simple. In fact, it's gotten more complicated. Everybody I talked to right now is drowning today in whatever the task is. And I think the point you made of single pane of glass have remained largely myth. I think the focus is wrong. I don't believe we actually need a single pane of glass that can manage, that can see everything. I think what we need are separate panes of glass that let us see what we need to see. And in fact, the way you guys do that for NoSQL and Hadoop makes some sense. Cohesity has their own that looks at things at more of a higher level, data plan. So I think we're really in the early, innings here, Stu. I think over the next few years, we will see a rise in better management tools and things to help us simplify. I know, I just did some research on IT priorities for 2019 and simplification actually is now ahead of even cybersecurity as the number one task for today's CIO. So I think we've gotten to the point where we've consumed so much stuff. Now it's time to simplify it. And that's, there's no one answer for that but I think within the different departments within IT they need to look at what those management tools are to let them do that. Yeah, I mean, going back, I think back to when I first became an analyst about nine years ago, a central premise is that enterprise IT doesn't necessarily have the skill set to go architect it. They're not a Google or a Yahoo. So they will spend money from the vendors and the suppliers to help simplify that for them environment. But Peter, wanna ask you, brought up people are drowning in information. Yes. Definitely, we know that today in 2019 there was more going on than they had a year from now and when we look forward to 2020 we expect that there will be even more. So the answer in the industry is AI and ML are going to come solve some of this for us. So tell us how that fits into these sorts of solutions. Sure, and the answer is machine learning and AI will absolutely need to be. Our view is that they're critical pillars to the future of data management. They have to be because the volume of data and the complexity of the infrastructure within which you're running, you can't as human beings, we are drowning and you need tools, you need help to solve this problem. And machine learning and AI are absolutely going to be key contributors. From an Amanis data standpoint, our approach has been very much about, completely avoiding the whole notion of machine learning, whitewash. Let's talk about the practical application of machine learning. So for example, what we do today is we apply machine learning to do what we call threat sense. So it's very specifically applied to the automation of anomaly detection. Build a model of what normal looks like from a backup and recovery standpoint. Anything that falls outside of normal gets flagged so that administrators can then do something, provide a human feedback loop to that machine learning algorithm so it can get smarter. We also recently introduced something that we call smart policies. That's about the automation of backups. So again, it's not about the holy grail of machine learning. In the case of smart policies, it's instead of creating spreadsheets and having a human being trying to figure out how to address a particular IPO, it's tell us what's your IPO and what data do you want to protect? We'll go build a model and we'll address your RPOs. And if we can't, we'll tell you why we can't. So very practical for today, to the point you made earlier about the fact that we're still in the early innings today. It's about the practical application of machine learning and AI to help people automate processes. I think the fear in dooming, gluing around AI is particularly in the IT circles is completely misguided. I understand why people might think it's going to take their job, but AI and ML is the IT pros best friend. There's so much data today, there's so much to do that people just can't connect the dots between those data points fast enough. Just like you look today, you wouldn't go to a radiologist that doesn't use machine learning to look at your brain science, right? It's getting harder and harder to work, to be a customer of a company that doesn't use AI or ML to analyze your data. It becomes very apparent because they're just not able to provide the same type of service. Yeah, I totally agree. We've done some events with MIT and a couple of the professors there, Erkman Wilson and Andy McAfee, talk about racing with the machines. So the people that can actually harness and leverage that, the challenge is if you're in IT and you're working on stuff that's five to 10 years old and you can't take advantage of those new tools, well, you need to scale up and you need to get ready, but most companies I talk to, it's not that they're looking to cut half the workforce, it's just that they can't add many more people. So most of them can be reskilled or heck, if there's some automation that can happen there, there's lots of projects sitting on the table that they've been trying to do for years. I don't find anybody that ever said, hey, if I could give you an extra month in the year that you wouldn't have stuff together. The question is, do you want to be strategic or organizational tactical? And if you want to be tactical, your job's only as long as that tactic, right? Peter, when I was hearing you walk through some of that ML piece, things like security and ransomware kind of popped into my head. Is that part of the solution? Yeah, absolutely. So threat sense is specifically, we talk about it as anomaly detection because overall it really is about, ransomware is essentially about detecting anomalies. So ransomware is an application of anomaly detection. So our threat sense capabilities built into the product, what happens is when we do backups, like I said, we build the model of what normal looks like. And then we flag anomalies. My dataset size, all of a sudden spiked, my data type, all of a sudden I have a bunch of zip files or something, all of a sudden something has changed that's outside of normal. And then we flag that and you can take action against that. So absolutely it is, but the initial application is specifically about ransomware. All right, Zayas, is there advice that you would want to give users or when you're talking to customers, what's the profile of somebody that is handling their data and leveraging it well? I don't know, it's really handling it well. But I think the advice I'd give is, you want to simplify and automate as much as you can and ruthlessly automate. I think if you're trying to do things the old way, you're going to wind up falling behind. And so I suppose to your question, what's the profile of a company that's doing it well? It's one that's actually able to roll out new services quickly. And you see that in a lot of the big name cloud companies, they always have new things coming and new things going and they're able to transform the way they deal with customers and employees. And to me that's the hallmark of a company that's using the data well. Ones that aren't, frankly, we've seen a lot of them go out of business right over the last two years. And so I think from an IT perspective, you want to embrace automation, embrace machine learning, embrace this concept of single pane of glass for your particular domain. Because what it lets you do is, it becomes a tool to help you do your job better. There's certain things people are good at and there's certain things people aren't and connecting the dots and terabits, petabytes of bits of data isn't one of them. So I think from an IT perspective, you want to automate and you want to embrace machine learning because it's going to be your best friend and it's going to help you keep your skillset current. Yeah, and I would just pick up on that and say that the answer isn't constraining. To a large extent it's really embracing data diversity. Like the answer to mass data fragmentation isn't homogenization of your data or eliminating particular data types. The proliferation of different data types is a direct result of organizations trying to be more agile and trying to be more nimble. So the answer isn't sort of constraining data. The answer is making the strategic investments in the right tools and sort of in some of the right policies and governance, if you will, so that you keep everybody strategically going in the right direction in this sort of federated diverse type of environment. Yeah, if you look at any market in IT, well, and really even in the consumer world, where there has been choice, it's created a rising tide for everybody. The question is you can't have it be chaotic. Right, right. And so you're bringing a level of order to a world that was historically chaotic and that untethers people to make whatever choice they want and use the best possible tools. Yeah, yeah. Peter, I go back to the promise of big data was that I was going to turn that proliferation of volume velocity and of data from a, oh my God, that's a problem and flip it on its head and become an opportunity for how we can leverage data. Give you the final word. How do we connect the dots from where that was a few years ago to this mass data fragmentation world today? Yeah, and the answer to that is don't treat, don't make big data sort of the three guys over in the corner who are the data scientists. Embrace big data. You embrace all your data types. So our message is the Hadoop and NoSQL data management folks is simply, look, Hadoop and NoSQL are a key part of your overall data strategy. Embrace those, include those in your overall strategies and make sure you're basically taking the right contextual picture of what you're trying to do. Include all your different data types. Hadoop and NoSQL are contributors to mass data fragmentation, but as part of that, if they're part of the problem then they need to be part of the solution, both from a data standpoint and from a solution standpoint. So that's really the message that we're driving is that embrace all your different data types, put the appropriate systems in place, take the right sort of approach to consolidating and solidifying your overall data strategy. All right, well, Peter and Zayas, thanks so much for sharing the latest update. Absolutely, data at the center of it all and need to embrace those new tools and opportunities out there. All right, I'm Stu Miniman and be sure to check out thecube.net for all of our research and shows that we'll be at and thank you as always for watching theCUBE.