This is a transcription of an interview between Silicon Angle Founder John Furrier and Jean-Luc Chatelain, Executive VP of Strategy and Technology at Direct Data Networks, webcast on Mr. Furrier's new program, Extraction Point.
JF: I’m John Furrier. Welcome to my show, the Extraction Point, where I find the signal amid all the noise. I’m here with Jean-Luc Chatelain, the new executive vice president of strategy and technology at Data Direct Networks. He has just moved over from HP where he was CTO of a major division. Jean-Luc, you were an entrepreneur before landing at HP. So why did you chose to move to another small start-up?
JLC: I’m not getting any younger, but I still enjoy working for a startup, so I want to do it at least one more time. I came to HP with the acquisition of a former start-up. HP has a good group of people, but then again I’ve been observing what’s happening on the market, and I could not pass a fantastic opportunity based on a new way of doing information management and information extraction, and most important, insight. And I am not new to DDN. I’m a long -time personal friend of the founders, and I’ve been on the board since 2006, so I was able to see the growth of the company. And when the time was right to go the next step and help them directly, I signed up.
JF: So at SiliconAngle we’ve been talking about storage being sexy since last year, and recently we saw that in action at the O’Reilly Strata Conference, where we did a live webcast. Data’s at the center of all the conversations around technology and the emerging opportunities. But storage is changing, and there’s a lot of market forces around big data. DDN is not a new company, they are like a grown-up start-up. You guys are certainly well beyond the garage phase with hundreds of millions of dollars in annual revenue. But you’re moving into a phase that’s like a startup. So what’s going on with DDN and how is that related to what you’ve seen in the market for the past eight years?
JLC: DDN is at the stage where they establish a brand and technology. They are well known in the market as a leading supplier. They are the largest privately-held storage company in the world. And now we need to become the largest information company in the world.
We’re entering a new world where information is essentially going to be the currency of the global enterprise. I joke that one day we will see a megabyte of information being counted in Singapore. It is the currency of today’s enterprise, and DDN’s assets are well-positioned to re-leverage the inflection points. Let me give you the really high level view of the interesting trend that is completely changing the IT market picture. Back in early 2000, every single analyst was talking about the information explosion. That was a firecracker compared to the nuclear explosion that we’re about to see. We’ve given voice to the machine; we've finally cracked voice recognition as well as voice synthesis, and we've made computers highly mobile. And they have a lot to say. And today, most enterprise are not ready to listen.
JF: You are talking about the Internet, mobile devices, servers, all pumping out requests?
JLC: That and more. Coke machines. Every time you put a coin in a vending machine it triggers a GPRS radio into action. VTR cameras; you can’t cross the street without some sort of surveillance. Your smart meter at home has lots of say to the energy company. And the IT infrastructure is not ready for the onslaught. I believe that what DDN has built for high performance computing and rich media, and the cloud, their natural market, is well-suited to help the enterprise swallow that information, and do something with it.
So the first big challenge of the immediate future is the rise of machines that are highly- interactive with users, and somebody needs to listen to them, because there's lots of good information in all screaming noise.
JF: JF: And they’re all connected to the internet, so they can be measured.
JLC: Exactly. Now that nuclear explosion of information is also forcing a convergence storage and compute. It will change how we handle data.
For 50 years we have been shipping data from point-to-point on physical media – tape, storage disks, whatever. That’s the old way.
That was fine for moving a a gigabyte of information. But now, in a single day I may capture 10 terabytes of information just using a videocamera. Sending that 10 terabytes from the original storage device to the computer infrastructure supporting the editing or use of those videos is extremely costly. It’s costly in time, it’s costly in dollars because bandwidth is not free, and its costly in storage. So now instead of sending the data to the application, we send the code to the data and do the compute operations inside the store.
So the processing is moving closer to where the data is, and that’s going to be very powerful because we can now start to explore it almost as soon as we receive it. Within 5-10 years that will mean in real time at a very very large scale. So the second point is the convergence of store and compute.
And the third point, which is really what’s going to make the difference for the line-of-business is the analytics that can be applied not just to traditional structured data but to tht video and other unstructured data as well. Analytics is what changes a business, and if you can do it quickly enough, you can make decisions in real time on changing aspect of your business to capture new customers or retaining more of those you have today. A lot of people are focusing on analytics because that is what creates business value from the data.
JF: Prior to starting the video we were talking about a value chain where data is the currency of the future enterprise. Can you share with the folks out there the value chain you mentioned?
JLC: Data-information-inside-results. Context is what transforms data into information. That data is extremely rich and varied. It’s not just your classic, old school, transactional data but is also instruction data and social data. And from that data you want to construct insight, and from that insight, you want to create new results for your business.
JF: Ok cool. So we’ve been exploring that, and we’re trying to find some proof points in one company in particular called ClickFox that we’ve had on the Cube. Its CEO Marco Pacelli has been talking to a lot of enterprises, but he says many of them aren't there yet. ClickFox has an analytical package that provides insight into all these massive data. It’s primary customers are mobile carriers. He argues that, “If you don’t understand your data, you can’t reengineer your processes.” So he’s kind of bumping up to these big clients. So what’s your opinion on where we are on that value chain for the customers out there? Are they in the ‘just trying to figure it out’ stage? Are there any specific examples that you can share with us on who the leaders in this area are? Who’s using the data for results?
JLC: I’m familiar with Clickfox. It’s a rocking company by the way. They got a great product.
JF: I love them. They’re doing a lot of cutting edge work.
JLC: They are. We are at the start of trying to understand the data. In fact, most of what you see in analytics today, and there are some exceptions, is still focused entirely on transaction data out of operational systems. This has not delivered on its promises because it’s incomplete; it needs to get not only the transactional data but also the other kinds of data. But we’re early in the adaption cycle. We need more proofs that analytics are going to make a difference. But some market segments are quite advanced -- the financial services industry, for instance, makes heavy use of analytics, and they’re doing a pretty good job around it. The hospitality industry is also fairly advanced in using analytics to understand how they can better suit the product to the customer base. And the oil and gas industry is getting there. So what we’re going to see, I think, is that gradually, the proof will start showing up, and more more traditional enterprises will see the value. And to be honest, a lot of work has to be done on a mathematical model for extracting that insight.
So math is involved. Kids, if you go study your mathematics, you’re probably going to get a good job in the world of analytics. People don’t know yet what questions to ask. They have to understand what moves their business, what they can tweak in the business before they can figure out the right question to ask.
JF: Is that because we just never had this kind of access to data in the past? I mean, it seems to me that people are kinda clumsy around the notion of asking questions. The old school executives who're not tech geeks have a back-of-the-envelope kind of mentality, But now we measure everything. It should be that simple, they should know the basic questions. Is it because they never had the chance to do it before?
JLC: They’ve done it in a very backward-looking view of the world. Most of the analytics that we see on the Fortune 500 today are reporting analytics that look backward at historic data. They do that for two reasons: One is the pressure of compliance and governance to get my reports on my quarter for the numbers. The other is that they think that just looking at the past can give you good prediction of the future. But that’s no longer true because we’re in a very dynamic environment where stuff that worked last week may not work this week because of some external event. And there’s some very strong players, but they tend to have a backward-looking view of the world.
JF: Has the time lag in producing the analysis been a big part of that too? It seems to take a long time to generate those reports.
JLC: They are not real-time. They are completely time shifted. You are watching the football game three weeks after it happens. It makes everything that business does reactive, when prediction and business speed is key. I need to extract my information when the event happens so I can predict what will happen next or analyze what to do next.
JF: So that's the opportunity for the folks out there, all the guys you talked to. So tell us what it really takes. There’s an ingestion problem, right? You have machines throwing off massive amounts of information. So talk through that pipe.
JLC: So a few points. Number 1, price should not just be the cost of the drive, it should include how much all the software you need to do something will cost, right? So this 1Tbyte drive is $99 but your real cost is $699. This is why storage is still a good business –it works.
JF: That’s why the cloud storage business is doing pretty well, right?
JLC: And the cloud is helping. But going back to the ingestion process: You need big pipes, right? So buying a bunch of drives by themselves is pretty useless. You need a freeway to bring the information to those drives in parallel. This is what we do at DDN. We build a giant highway that moves massive amounts of information into large pools of storage. And I can say that no one does it better than us from a performance point of view.
But we also need an infrastructure that can bring that compute capability to those drives that are going to capture all that data. So that device has to have virtualized processing capabilities so it can treat that data in real time. That’s how you’re able to handle that stuff. You can do a lot in the pre-processing phase, but you will want to do that in a cloud also, so you need automated cloud infrastructure so you can get that analysis on demand.
JF: The ingestion requires heavy-duty infrastructure. An application like for example Twitter is a firehose of data that you have to put someplace. So how much of the information infrastructure is software-based? If it will be more software than hardware, where’s the IP going to be?
JLC: The IP’s going to be in software, so the hardware must have a set of characteristics that enable that software, right? And there’s a commoditization of hardware. So hardware’s not going to be disappearing but definitely, the trend’s going to be to have an infrastructure where most computers live together and the value’s in the software that can exploit that hardware.
JF: Big data is a trendy term. Most people think it’s the marketing term for the industry. What’s your definition of big data, and what’s your angle on the whole data market, but big data in particular?
JLC: Can I say that you analysts and journalists are making a lot of fuss around that term. Can we find another term?
JF: I actually don’t like big data. Mike Olson, the CEO of Cloudera doesn’t like the name ‘big data’. Tim O’Reilly call his conference Strata, not Big Data. We’re not high on big data as a term because all data is big. If you’re an enterprise or service provider, you have to deal with all kinds of data: big, little, fast, slow –but in aggregate, it’s all data, it’s all to be vetted.
JLC: I think the difference may be that an aspect of big is how rich that data is and how many facets it has that it didn’t have when it was limited to transactional data. So it includes massive amounts of unstructured data. That probably is what defines big data. There's a lot of it, and its different. It’s highly unstructured. And the good point about unstructured data is it’s massive on the value, but that value is very well hidden, and the magic is in how you’re going to extract that value.
JF: Because you’re talking about something like Jodi Foster's movie “Contact” where they are receiving all these unstructured noises, and they're trying to find the signal hidden in all that noise. Now the social networks have created this notion of an activity stream, the river of information. And there is the metaphor about volume coming in. Can you talk about volume and what all these terms mean?
JLC: So the volume is large. That is pretty obvious. What’s interesting is that those new social types of information dangle a lot of handles that we can start grabbing to get at value. I'm a big fan of them because not only are we getting data, but the natural interaction of people with that information is creating metadata. The fact that you like something is pure magic for Facebook, right?
JF: It’s the ultimate gesture point, isn’t it? You’re saying, ‘”I endorse this.”
JLC: Absolutely. But you’re also classifying the information for Facebook, right? So someone posts a video that you would need hours to understand. But five people ”like” that piece of video, and suddenly you have metadata around that video, and you can classify users by whether they “like” that video. When you start getting metadata, you’re half way to extracting your insight.
JF: So the volume is about flow size of information. There’s also velocity, right, which was in real-time in context to that. It creates a lot of noise, right?
JLC: Lots of noise.
JF: You gotta get analytics, you gotta get all these together. You gotta get the Contact movie out there for the folks. Share that story with us. JLC: Yeah, it’s a good metaphor. There’s an observation post somewhere in an island listening to the white noise of space. And the gentle man who happens to have better hearing than normal because he’s blind hears a pattern inside that white noise. And they start isolating the patterns, and they realized that pattern was a recording that was done in the 1940s. But the content of the recording doesn't doesn’t matter, what’s interesting is –
JF: He decoded something encoded the white noise.
JLC: He decoded the white noise. He latched on something specific and was able to pull information out of it. I found that analogy very close to what an enterprise or even a customer is going to have to do?
JF: And this ties to you point earlier about how the unstructured social data is diverse and omnidirectional as it comes in, and you can use it to latch on to things that look narrow. This is that kind of example where it looks like a very small piece of data, but when you unpack it, it provides huge value.
JLC: Exactly.
JF: That’s kind of what you’re getting there.
JLC: And every piece of context associated with that data: who’s listening to it, what they are wondering about it it, what their interest is in it, all creates metadata.
JF: So this comes back down to your point earlier about the role of storage and compute where most analytics have been looking at the rear view mirror and reporting.
JLC: Right. It’s time-shifted. If you can capture and analyze it in real time, it’s really predictive. So the insight results aspect of this value chain is going to come from the new data sources and the ability to act on them. Act on it, and act on it quickly. Let’s take Twitter for example, just an example, you don’t want to accumulate massive amounts of tweet all day and then at 8 p.m. send them all to analysis to find out what people are thinking about your product. You really want to get that information as you are capturing those tweets.
JF: Yeah. Well, we know you're working on that, so don’t give too much detail out to that audience out there. But seriously, you’re working on some pretty cool tech. I mean this is not like I can get a computer science degree and start doing this. You need to have some serious technology to get to this level. What’s your advice to folks out there both on the ‘I want to expand my career on this area from a tech perspective” and also from a customer perspective? How do they jump in? What would you share with them?
JLC: Well, if you don’t know what to major in, I’m preaching about my own church, then get a major on computer science. But you can get a major on math and also be fine. And in fact, you’ll be very complementary.
JF: This new data scientist role is interesting but it’s for math jocks, not so much for programmers. There’s a kind of collision between disciplines now right?
JLC: The math geeks figure out the formula that makes the magic. Generally, they have no clue about how to implement the formula. That’s where the CS guys are coming in. And that’s why I said these two fields are fairly complementary. Now if you’re a business guy, be very aware of your business processes and educate yourself on the turning points of your business so you can translate how you can optimize your business to the math geek and to the computer science geek so they can do the magic behind the scene. A lot of new jobs are going to be created around information, for instance the notion of informational architect.
JF: That was a role of the past around data warehousing, right? A librarian kind of thing?
JLC: The data warehouse architect was creating data boxes, creating schemas. But we're not in a world where you have to create a schem; you have to learn to be extremely lose.
JF: So where this was once librarian-like role, it has evolved into a much more strategic role requiring math, and technical knowledge, a solution architect kind of thing, right?
JLC: Right. Solution architect around the information. The other new career is information steward. This is part of a set of important disciplines around managing information that will include an aspect of governance. People that care for information. They don’t own it, but they care for the information on behalf of the owner.
JF: Let’s talk about you at Data Direct Network. What are some of the cool things you’ve been working on? You guys honestly built a great sustainable business, self-funded. DDN is kind of in a hall of fame category of company in a sense that it started pretty much funded by the founders, and it has grown so big, and it has this massive growth strategy ahead of it. But you must have some tech, and you’re going to have some more tech coming.
JLC: We have lots of tech.
First, we're going to keep on doing what we’ve been doing, but we’re going to do it faster, better, and with more quality than anybody else in the planet. This is the block and tackling. We’re good at what we do.
Then the new effort involves a massive investment on software, right? We’re just opening, enlarging our office here in Mountain View, moving a few miles away. We’re hiring like crazy. We have two other labs: one in Colorado where a lot of our RAID operating system is being done, and the other in Columbia, Maryland. We’re hiring everywhere. In fact, I tweeted last week to keep up with the job page on the DDN/career because it’s a good problem to have, trust me.
JLC: We are a startup. We do a lot of work on, for example, key value stores; a lot of work on the new object file system; we do a lot of call stuff on the cloud.
JF: What’s the most exciting thing that you see in the future coming around the corner that other people might not see?
JLC: Like I said, what you really need is to understand the size of your opportunity. Everywhere I look, no one says this is going to die out, this is not going to last and we better move on. In every market where we play, whether it’s governance, rich media, the cloud, it just doesn’t stop. In fact, our growth is limited by the number of people we can put on the street knocking on doors; and we can sustain growth for a long long time if only we can get the people.
But if I look around the corner I’m really excited with all the work we do around vectorization of compute loads. I want to enable function shipping. I wanted to make it trivial for somebody to take their algorithm and almost transparently distribute the work to multiple devices. That’s a core part. And then around the cloud, the interesting opportunity of leveraging all those technologies that we talked about to focus on hybrid clouds for specific industries to help people do their jobs better and more efficiently, that’s exciting, and it doesn’t have to be a private cloud. It could be an industry-based cloud. But just like as you were mentioning, you can slide your credit card at Amazon and get some storage and get some compute capability. If we ever create a credit card for a given industry, we’re going to be able to slide that credit card in that specialized cloud for the industry and get stuff done.
JF: It’s exciting time. I get intoxicated by the new opportunities that are emerging. I mean we’re on the publishing business. We have SiliconAngle.tv, SiliconAngle.com on the publishing and video side. We see the great possibilities in video. I honestly see business opportunities for guys like us. I could actually expand. We’re self-funded kinda like you guys. A decade or two ago Akamai needed huge amounts of capital. Today I can literally roll out a continental video CDN delivery network for very little capital. Talk about dynamic. I mean that’s not trivial to do, but you can do it.
JF: So for example one of the big things we hear about cloud is that it’s so easy to do and for startups like us with a low startup cost, low validation, and then ship the product. But CIOs are rejecting that because they don’t want to put their critical apps on the cloud. What you guys are essentially talking about is with software I can put the entire production into the cloud. That’s what the private cloud dream is. So where are we at that, what’s your angle on that?
JLC: That’s really what the private cloud is. It has to offer the same level of SLA and QOS and sometimes, security. Although there is too much fear about public cloud security that’s not justified, but I understand why some industries have a heightened sense of security. It’s fairly easy to do now. Other people provide some building-block appliances that are already packaged to provide all that’s necessary to build your cloud, and you can do it almost just-in-time. You don’t have to buy some giant infrastructure on day 1 and wait till you have 99 users to start buying another giant infrastructure for 101 users. You can start small, and as the demand pens up, like in your case, the demand of people watching SiliconAngle.tv, you can add new software and hardware appliances.
JF: You can start by putting your credit card down say at Amazon S3, for example. But once you go beyond a certain size at Amazon it gets expensive, risky, and complex – configuration management, automation, SLAs – the list kinda goes on and on that once you hit a certain tipping point at Amazon.
JLC: Right. Look at storage, for example, because storage has a high TCO, right? You buy into a storage technology with real money, and that technology doesn’t always scale high enough as demand grows. For the storage provider, there's always a point where you can no longer scale for the customers, where it becomes really too expensive for the customers. So for an enterprise it may make sense to run my own storage in my own private cloud, because even including the TCO, it is still going to be less expensive that what Amazon can offer.
JF: TCO is a big issue.
JLC: Yeah and a lot of people don’t buy with TCO in mind.
JF: Just share with us the vision in your mind for the next 5-10 years. Not from a DDN perspective but you know, taking up your experiences as a startup entrepreneur, executive, tech geek, what’s going to be different in the world 5-10 years from now? Our life, our technology, our storage, you know, and all stuff data’s going to drive then? What’s going to be different?
JLC: First, opportunities are limitless. There’s more startups now than there was ever before. And that's because of that information being the currency. So think of massive amounts of oil flowing and we can all play with that , and we can do something with that oil, and we will get wealthy and successful doing that. We can do the same with information. So our life in 5-10 years will be completely driven by how that information is being used. What’s interesting is that today we’re looking for information and looking for insight, but 10 years from now information will find us and insight will find us so it’s no longer the Google model.
JF: Computer science should be working for us.
JLC: Exactly.
JF: Like Star Trek magic needs to come back.
JLC: What’s going to happen is information is going to find us. It will find us in the context that we are. You know, I expect that not long from now, you’ll be landing in Moscow, you’ll be taking money at an ATM machine, and along with the receipt for the rubble that you took will be a map for the nearest McDonalds for example because information knows that you’re a McDonalds lover, and that’s what you want when you land to a foreign country. So that’s a case where information finds the user, and that’s what makes a big big difference. And in fact we are beginning to see that today on smartphones and GPS systems. You buy a GPS and as you’re driving it points to this little icon saying there’s the Starbucks there without you ever asking. That is the beginning of information finding people.
JF: That’s the future