.banner-thumbnail-wrapper { display:none; }

HPC

Inside story on HPC's role in the Bridges Research Project at Pittsburgh Supercomputing Center

The next BriefingsDirect Voice of the Customer high-performance computing (HPC) success story interview examines how Pittsburgh Supercomputing Center (PSC) has developed a research computing capability, Bridges, and how that's providing new levels of analytics, insights, and efficiencies.

We'll now learn how advances in IT infrastructure and memory-driven architectures are combining to meet the new requirements for artificial intelligence (AI), big data analytics, and deep machine learning.

How modern storage provides hints on optimizing and best managing hybrid IT and multi-cloud resources

The next BriefingsDirect Voice of the Analyst interview examines the growing need for proper rationalizing of which apps, workloads, services and data should go where across a hybrid IT continuum.

Managing hybrid IT necessitates not only a choice between public cloud and private cloud, but a more granular approach to picking and choosing which assets go where based on performance, costs, compliance, and business agility.

Listen to the podcast. Find it on iTunes. Get the mobile app. Read a full transcript or download a copy.

Here to report on how to begin to better assess what IT variables should be managed and thoughtfully applied to any cloud model is Mark Peters, Practice Director and Senior Analyst at Enterprise Strategy Group (ESG). The discussion is moderated by Dana Gardner, principal analyst at Interarbor Solutions.

Here are some excerpts:

Gardner: Now that cloud adoption is gaining steam, it may be time to step back and assess what works and what doesn’t. In past IT adoption patterns, we’ve seen a rapid embrace that sometimes ends with at least a temporary hangover. Sometimes, it’s complexity or runaway or unmanaged costs, or even usage patterns that can’t be controlled. Mark, is it too soon to begin assessing best practices in identifying ways to hedge against any ill effects from runaway adoption of cloud? 

Peters: The short answer, Dana, is no. It’s not that the IT world is that different. It’s just that we have more and different tools. And that is really what hybrid comes down to -- available tools.

Peters

Peters

It’s not that those tools themselves demand a new way of doing things. They offer the opportunity to continue to think about what you want. But if I have one repeated statement as we go through this, it will be that it’s not about focusing on the tools, it’s about focusing on what you’re trying to get done. You just happen to have more and different tools now.

Gardner: We hear sometimes that at as high as board of director levels, they are telling people to go cloud-first, or just dump IT all together. That strikes me as an overreaction. If we’re looking at tools and to what they do best, is cloud so good that we can actually just go cloud-first or cloud-only?

Cloudy cloud adoption

Peters: Assuming you’re speaking about management by objectives (MBO), doing cloud or cloud-only because that’s what someone with a C-level title saw on a Microsoft cloud ad on TV and decided that is right, well -- that clouds everything.

You do see increasingly different people outside of IT becoming involved in the decision. When I say outside of IT, I mean outside of the operational side of IT.

You get other functions involved in making demands. And because the cloud can be so easy to consume, you see people just running off and deploying some software-as-a-service (SaaS) or infrastructure-as-a-service (IaaS) model because it looked easy to do, and they didn’t want to wait for the internal IT to make the change.

All of the research we do shows that the world is hybrid for as far ahead as we can see.

Running away from internal IT and on-premises IT is not going to be a good idea for most organizations -- at least for a considerable chunk of their workloads. All of the research we do shows that the world is hybrid for as far ahead as we can see. 

Gardner: I certainly agree with that. If it’s all then about a mix of things, how do I determine the correct mix? And if it’s a correct mix between just a public cloud and private cloud, how do I then properly adjust to considerations about applications as opposed to data, as opposed to bringing in microservices and Application Programming Interfaces (APIs) when they’re the best fit?

How do we begin to rationalize all of this better? Because I think we’ve gotten to the point where we need to gain some maturity in terms of the consumption of hybrid IT.

Learn More About

Hybrid IT Management

Solutions From HPE

Peters: I often talk about what I call the assumption gap. And the assumption gap is just that moment where we move from one side where it’s okay to have lots of questions about something, in this case, in IT. And then on the other side of this gap or chasm, to use a well-worn phrase, is where it’s not okay to ask anything because you’ll see you don’t know what you’re talking about. And that assumption gap seems to happen imperceptibly and very fast at some moment.

So, what is hybrid IT? I think we fall into the trap of allowing ourselves to believe that having some on-premises workloads and applications and some off-premises workloads and applications is hybrid IT. I do not think it is. It’s using a couple of tools for different things.

It’s like having a Prius and a big diesel and/or gas F-150 pickup truck in your garage and saying, “I have two hybrid vehicles.” No, you have one of each, or some of each. Just because someone has put an application or a backup off into the cloud, “Oh, yeah. Well, I’m hybrid.” No, you’re not really.

The cloud approach

The cloud is an approach. It’s not a thing per se. It’s another way. As I said earlier, it’s another tool that you have in the IT arsenal. So how do you start figuring what goes where?

I don’t think there are simple answers, because it would be just as sensible a question to say, “Well, what should go on flash or what should go on disk, or what should go on tape, or what should go on paper?” My point being, such decisions are situational to individual companies, to the stage of that company’s life, and to the budgets they have. And they’re not only situational -- they’re also dynamic.

I want to give a couple of examples because I think they will stick with people. Number one is you take something like email, a pretty popular application; everyone runs email. In some organizations, that is the crucial application. They cannot run without it. Probably, what you and I do would fall into that category. But there are other businesses where it’s far less important than the factory running or the delivery vans getting out on time. So, they could have different applications that are way more important than email.

When instant messaging (IM) first came out, Yahoo IM text came out, to be precise. They used to do the maintenance between 9 am and 5 pm because it was just a tool to chat to your friends with at night. And now you have businesses that rely on that. So, clearly, the ability to instant message and text between us is now crucial. The stock exchange in Chicago runs on it. IM is a very important tool.

The answer is not that you or I have the ability to tell any given company, “Well, x application should go onsite and Y application should go offsite or into a cloud,” because it will vary between businesses and vary across time.

If something is or becomes mission-critical or high-risk, it is more likely that you’ll want the feeling of security, I’m picking my words very carefully, of having it … onsite.

You have to figure out what you're trying to get done before you figure out what you're going to do with it.

But the extent to which full-production apps are being moved to the cloud is growing every day. That’s what our research shows us. The quick answer is you have to figure out what you’re trying to get done before you figure out what you’re going to do it with. 

Gardner: Before we go into learning more about how organizations can better know themselves and therefore understand the right mix, let’s learn more about you, Mark. 

Tell us about yourself, your organization at ESG. How long have you been an IT industry analyst? 

Peters: I grew up in my working life in the UK and then in Europe, working on the vendor side of IT. I grew up in storage, and I haven’t really escaped it. These days I run ESG’s infrastructure practice. The integration and the interoperability between the various elements of infrastructure have become more important than the individual components. I stayed on the vendor side for many years working in the UK, then in Europe, and now in Colorado. I joined ESG 10 years ago.

Lessons learned from storage

Gardner: It’s interesting that you mentioned storage, and the example of whether it should be flash or spinning media, or tape. It seems to me that maybe we can learn from what we’ve seen happen in a hybrid environment within storage and extrapolate to how that pertains to a larger IT hybrid undertaking.

Is there something about the way we’ve had to adjust to different types of storage -- and do that intelligently with the goals of performance, cost, and the business objectives in mind? I’ll give you a chance to perhaps go along with my analogy or shoot it down. Can we learn from what’s happened in storage and apply that to a larger hybrid IT model?

Learn More About

Hybrid IT Management

Solutions From HPE

Peters: The quick answer to your question is, absolutely, we can. Again, the cloud is a different approach. It is a very beguiling and useful business model, but it’s not a panacea. I really don’t believe it ever will become a panacea.

Now, that doesn’t mean to say it won’t grow. It is growing. It’s huge. It’s significant. You look at the recent announcements from the big cloud providers. They are at tens of billions of dollars in run rates.

But to your point, it should be viewed as part of a hierarchy, or a tiering, of IT. I don’t want to suggest that cloud sits at the bottom of some hierarchy or tiering. That’s not my intent. But it is another choice of another tool.

Let’s be very, very clear about this. There isn’t “a” cloud out there. People talk about the cloud as if it exists as one thing. It does not. Part of the reason hybrid IT is so challenging is you’re not just choosing between on-prem and the cloud, you’re choosing between on-prem and many clouds -- and you might want to have a multi-cloud approach as well. We see that increasingly.

What we should be looking for are not bright, shiny objects -- but bright, shiny outcomes.

Those various clouds have various attributes; some are better than others in different things. It is exactly parallel to what you were talking about in terms of which server you use, what storage you use, what speed you use for your networking. It’s exactly parallel to the decisions you should make about which cloud and to what extent you deploy to which cloud. In other words, all the things you said at the beginning: cost, risk, requirements, and performance.

People get so distracted by bright, shiny objects. Like they are the answer to everything. What we should be looking for are not bright, shiny objects -- but bright, shiny outcomes. That’s all we should be looking for.

Focus on the outcome that you want, and then you figure out how to get it. You should not be sitting down IT managers and saying, “How do I get to 50 percent of my data in the cloud?” I don’t think that’s a sensible approach to business. 

Gardner: Lessons learned in how to best utilize a hybrid storage environment, rationalizing that, bringing in more intelligence, software-defined, making the network through hyper-convergence more of a consideration than an afterthought -- all these illustrate where we’re going on a larger scale, or at a higher abstraction.

Going back to the idea that each organization is particular -- their specific business goals, their specific legacy and history of IT use, their specific way of using applications and pursuing business processes and fulfilling their obligations. How do you know in your organization enough to then begin rationalizing the choices? How do you make business choices and IT choices in conjunction? Have we lost sufficient visibility, given that there are so many different tools for doing IT?

Get down to specifics

Peters: The answer is yes. If you can’t see it, you don’t know about it. So to some degree, we are assuming that we don’t know everything that’s going on. But I think anecdotally what you propose is absolutely true.

I’ve beaten home the point about starting with the outcomes, not the tools that you use to achieve those outcomes. But how do you know what you’ve even got -- because it’s become so easy to consume in different ways? A lot of people talk about shadow IT. You have this sprawl of a different way of doing things. And so, this leads to two requirements.

Number one is gaining visibility. It’s a challenge with shadow IT because you have to know what’s in the shadows. You can’t, by definition, see into that, so that’s a tough thing to do. Even once you find out what’s going on, the second step is how do you gain control? Control -- not for control’s sake -- only by knowing all the things you were trying to do and how you’re trying to do them across an organization. And only then can you hope to optimize them.

You can't manage what you can't measure. You also can't improve things that can't be managed or measured.

Again, it’s an old, old adage. You can’t manage what you can’t measure. You also can’t improve things that can’t be managed or measured. And so, number one, you have to find out what’s in the shadows, what it is you’re trying to do. And this is assuming that you know what you are aiming toward.

This is the next battleground for sophisticated IT use and for vendors. It’s not a battleground for the users. It’s a choice for users -- but a battleground for vendors. They must find a way to help their customers manage everything, to control everything, and then to optimize everything. Because just doing the first and finding out what you have -- and finding out that you’re in a mess -- doesn’t help you.

Learn More About

Hybrid IT Management

Solutions From HPE

Visibility is not the same as solving. The point is not just finding out what you have – but of actually being able to do something about it. The level of complexity, the range of applications that most people are running these days, the extremely high levels of expectations both in the speed and flexibility and performance, and so on, mean that you cannot, even with visibility, fix things by hand.

You and I grew up in the era where a lot of things were done on whiteboards and Excel spreadsheets. That doesn’t cut it anymore. We have to find a way to manage what is automated. Manual management just will not cut it -- even if you know everything that you’re doing wrong. 

Gardner: Yes, I agree 100 percent that the automation -- in order to deal with the scale of complexity, the requirements for speed, the fact that you’re going to be dealing with workloads and IT assets that are off of your premises -- means you’re going to be doing this programmatically. Therefore, you’re in a better position to use automation.

I’d like to go back again to storage. When I first took a briefing with Nimble Storage, which is now a part of Hewlett Packard Enterprise (HPE), I was really impressed with the degree to which they used intelligence to solve the economic and performance problems of hybrid storage.

Given the fact that we can apply more intelligence nowadays -- that the cost of gathering and harnessing data, the speed at which it can be analyzed, the degree to which that analysis can be shared -- it’s all very fortuitous that just as we need greater visibility and that we have bigger problems to solve across hybrid IT, we also have some very powerful analysis tools.

Mark, is what worked for hybrid storage intelligence able to work for a hybrid IT intelligence? To what degree should we expect more and more, dare I say, artificial intelligence (AI) and machine learning to be brought to bear on this hybrid IT management problem?

Intelligent automation a must

Peters: I think it is a very straightforward and good parallel. Storage has become increasingly sophisticated. I’ve been in and around the storage business now for more than three decades. The joke has always been, I remember when a megabyte was a lot, let alone a gigabyte, a terabyte, and an exabyte.

And I’d go for a whole day class, when I was on the sales side of the business, just to learn something like dual parsing or about cache. It was so exciting 30 years ago. And yet, these days, it’s a bit like cars. I mean, you and I used to use a choke, or we’d have to really go and check everything on the car before we went on 100-mile journey. Now, we press the button and it better work in any temperature and at any speed. Now, we just demand so much from cars.

To stretch that analogy, I’m mixing cars and storage -- and we’ll make it all come together with hybrid IT in that it’s better to do things in an automated fashion. There’s always one person in every crowd I talk to who still believes that a stick shift is more economic and faster than an automatic transmission. It might be true for one in 1,000 people, and they probably drive cars for a living. But for most people, 99 percent of the people, 99.9 percent of the time, an automatic transmission will both get you there faster and be more efficient in doing so. The same became true of storage.

We used to talk about how much storage someone could capacity-plan or manage. That’s just become old hat now because you don’t talk about it in those terms. Storage has moved to be -- how do we serve applications? How do we serve up the right place in the right time, get the data to the right person at the right time at the right price, and so on?

We don’t just choose what goes where or who gets what, we set the parameters -- and we then allow the machine to operate in an automated fashion. These days, increasingly, if you talk to 10 storage companies, 10 of them will talk to you about machine learning and AI because they know they’ve got to be in that in order to make that execution of change ever more efficient and ever faster. They’re just dealing with tremendous scale, and you could not do it even with simple automation that still involves humans.

It will be self-managing and self-optimizing. It will not be a “recommending tool,” it will be an “executing tool.”

We have used cars as a social analogy. We used storage as an IT analogy, and absolutely, that’s where hybrid IT is going. It will be self-managing and self-optimizing. Just to make it crystal clear, it will not be a “recommending tool,” it will be an “executing tool.” There is no time to wait for you and me to finish our coffee, think about it, and realize we have to do something, because then it’s too late. So, it’s not just about the knowledge and the visibility. It’s about the execution and the automated change. But, yes, I think your analogy is a very good one for how the IT world will change.

Learn More About

Hybrid IT Management

Solutions From HPE

Gardner: How you execute, optimize and exploit intelligence capabilities can be how you better compete, even if other things are equal. If everyone is using AWS, and everyone is using the same services for storage, servers, and development, then how do you differentiate?

How you optimize the way in which you gain the visibility, know your own business, and apply the lessons of optimization, will become a deciding factor in your success, no matter what business you’re in. The tools that you pick for such visibility, execution, optimization and intelligence will be the new real differentiators among major businesses.

So, Mark, where do we look to find those tools? Are they yet in development? Do we know the ones we should expect? How will organizations know where to look for the next differentiating tier of technology when it comes to optimizing hybrid IT?

What’s in the mix?

Peters: We’re talking years ahead for us to be in the nirvana that you’re discussing.

I just want to push back slightly on what you said. This would only apply if everyone were using exactly the same tools and services from AWS, to use your example. The expectation, assuming we have a hybrid world, is they will have kept some applications on-premises, or they might be using some specialist, regional or vertical industry cloud. So, I think that’s another way for differentiation. It’s how to get the balance. So, that’s one important thing.

And then, back to what you were talking about, where are those tools? How do you make the right move?

We have to get from here to there. It’s all very well talking about the future. It doesn’t sound great and perfect, but you have to get there. We do quite a lot of research in ESG. I will throw just a couple of numbers, which I think help to explain how you might do this.

We already find that the multi-cloud deployment or option is a significant element within a hybrid IT world. So, asking people about this in the last few months, we found that about 75 percent of the respondents already have more than one cloud provider, and about 40 percent have three or more.

You’re getting diversity -- whether by default or design. It really doesn’t matter at this point. We hope it’s by design. But nonetheless, you’re certainly getting people using different cloud providers to take advantage of the specific capabilities of each.

This is a real mix. You can’t just plunk down some new magic piece of software, and everything is okay, because it might not work with what you already have -- the legacy systems, and the applications you already have. One of the other questions we need to ask is how does improved management embrace legacy systems?

Some 75 percent of our respondents want hybrid management to be from the infrastructure up, which means that it’s got to be based on managing their existing infrastructure, and then extending that management up or out into the cloud. That’s opposed to starting with some cloud management approach and then extending it back down to their infrastructure.

People want to enhance what they currently have so that it can embrace the cloud. It’s enhancing your choice of tiers so you can embrace change.

People want to enhance what they currently have so that it can embrace the cloud. It's enhancing your choice of tiers so you can embrace change. Rather than just deploying something and hoping that all of your current infrastructure -- not just your physical infrastructure but your applications, too -- can use that, we see a lot of people going to a multi-cloud, hybrid deployment model. That entirely makes sense. You're not just going to pick one cloud model and hope that it  will come backward and make everything else work. You start with what you have and you gradually embrace these alternative tools. 

Gardner: We’re creating quite a list of requirements for what we’d like to see develop in terms of this management, optimization, and automation capability that’s maybe two or three years out. Vendors like Microsoft are just now coming out with the ability to manage between their own hybrid infrastructures, their own cloud offerings like Azure Stack and their public cloud Azure.

Learn More About

Hybrid IT Management

Solutions From HPE

Where will we look for that breed of fully inclusive, fully intelligent tools that will allow us to get to where we want to be in a couple of years? I’ve heard of one from HPE, it’s called Project New Hybrid IT Stack. I’m thinking that HPE can’t be the only company. We can’t be the only analysts that are seeing what to me is a market opportunity that you could drive a truck through. This should be a big problem to solve.

Who’s driving?

Peters: There are many organizations, frankly, for which this would not be a good commercial decision, because they don’t play in multiple IT areas or they are not systems providers. That’s why HPE is interested, capable, and focused on doing this. 

Many vendor organizations are either focused on the cloud side of the business -- and there are some very big names -- or on the on-premises side of the business. Embracing both is something that is not as difficult for them to do, but really not top of their want-to-do list before they’re absolutely forced to.

From that perspective, the ones that we see doing this fall into two categories. There are the trendy new startups, and there are some of those around. The problem is, it’s really tough imagining that particularly large enterprises are going to risk [standardizing on them]. They probably even will start to try and write it themselves, which is possible – unlikely, but possible.

Where I think we will get the list for the other side is some of the other big organizations --- Oracle and IBM spring to mind in terms of being able to embrace both on-premises and off-premises.  But, at the end of the day, the commonality among those that we’ve mentioned is that they are systems companies. At the end of the day, they win by delivering the best overall solution and package to their clients, not individual components within it.

If you’re going to look for a successful hybrid IT deployment took, you probably have to look at a hybrid IT vendor.

And by individual components, I include cloud, on-premises, and applications. If you’re going to look for a successful hybrid IT deployment tool, you probably have to look at a hybrid IT vendor. That last part I think is self-descriptive. 

Gardner: Clearly, not a big group. We’re not going to be seeking suppliers for hybrid IT management from request for proposals (RFPs) from 50 or 60 different companies to find some solutions. 

Peters: Well, you won’t need to. Looking not that many years ahead, there will not be that many choices when it comes to full IT provisioning. 

Gardner: Mark, any thoughts about what IT organizations should be thinking about in terms of how to become proactive rather than reactive to the hybrid IT environment and the complexity, and to me the obvious need for better management going forward?

Management ends, not means

Peters: Gaining visibility into not just hybrid IT but the on-premise and the off-premise and how you manage these things. Those are all parts of the solution, or the answer. The real thing, and it’s absolutely crucial, is that you don’t start with those bright shiny objects. You don’t start with, “How can I deploy more cloud? How can I do hybrid IT?” Those are not good questions to ask. Good questions to ask are, “What do I need to do as an organization? How do I make my business more successful? How does anything in IT become a part of answering those questions?”

In other words, drum roll, it’s the thinking about ends, not means.

Gardner:  If our listeners and readers want to follow you and gain more of your excellent insight, how should they do that? 

Peters: The best way is to go to our website, www.esg-global.com. You can find not just me and all my contact details and materials but those of all my colleagues and the many areas we cover and study in this wonderful world of IT.

Listen to the podcast. Find it on iTunes. Get the mobile app. Read a full transcript or download a copy. Sponsor: Hewlett Packard Enterprise.

You may also be interested in:

Inside story on HPC’s AI role in Bridges 'strategic reasoning' research at CMU

The next BriefingsDirect high performance computing (HPC) success interview examines how strategic reasoning is becoming more common and capable -- even using imperfect information.

We’ll now learn how Carnegie Mellon University and a team of researchers there are producing amazing results with strategic reasoning thanks in part to powerful new memory-intense systems architectures.

Listen to the podcast. Find it on iTunes. Get the mobile app. Read a full transcript ordownload a copy. 

To learn more about strategic reasoning advances, please join me in welcoming Tuomas Sandholm, Professor and Director of the Electronic Marketplaces Lab at Carnegie Mellon University in Pittsburgh. The discussion is moderated by Dana Gardner, Principal Analyst at Interarbor Solutions.

Here are some excerpts:

Gardner: Tell us about strategic reasoning and why imperfect information is often the reality that these systems face?

Sandholm: In strategic reasoning we take the word “strategic” very seriously. It means game theoretic, so in multi-agent settings where you have more than one player, you can't just optimize as if you were the only actor -- because the other players are going to act strategically. What you do affects how they should play, and what they do affects how you should play.

Sandholm

Sandholm

That's what game theory is about. In artificial intelligence (AI), there has been a long history of strategic reasoning. Most AI reasoning -- not all of it, but most of it until about 12 years ago -- was really about perfect information games like Othello, Checkers, Chess and Go.

And there has been tremendous progress. But these complete information, or perfect information, games don't really model real business situations very well. Most business situations are of imperfect information.

Know what you don’t know

So you don't know the other guy's resources, their goals and so on. You then need totally different algorithms for solving these games, or game-theoretic solutions that define what rational play is, or opponent exploitation techniques where you try to find out the opponent's mistakes and learn to exploit them.

So totally different techniques are needed, and this has way more applications in reality than perfect information games have.

Gardner: In business, you don't always know the rules. All the variables are dynamic, and we don't know the rationale or the reasoning behind competitors’ actions. People sometimes are playing offense, defense, or a little of both.

Before we dig in to how is this being applied in business circumstances, explain your proof of concept involving poker. Is it Five-Card Draw?

Heads-Up No-Limit Texas Hold'em has become the leading benchmark in the AI community.

Sandholm: No, we’re working on a much harder poker game called Heads-Up No-Limit Texas Hold'em as the benchmark. This has become the leading benchmark in the AI community for testing these application-independent algorithms for reasoning under imperfect information.

The algorithms have really nothing to do with poker, but we needed a common benchmark, much like the IC chip makers have their benchmarks. We compare progress year-to-year and compare progress across the different research groups around the world. Heads-Up No-limit Texas Hold'em turned out to be great benchmark because it is a huge game of imperfect information.

It has 10 to the 161 different situations that a player can face. That is one followed by 161 zeros. And if you think about that, it’s not only more than the number of atoms in the universe, but even if, for every atom in the universe, you have a whole other universe and count all those atoms in those universes -- it will still be more than that.

Gardner: This is as close to infinity as you can probably get, right?

Sandholm: Ha-ha, basically yes.

Gardner: Okay, so you have this massively complex potential data set. How do you winnow that down, and how rapidly does the algorithmic process and platform learn? I imagine that being reactive, creating a pattern that creates better learning is an important part of it. So tell me about the learning part.

Three part harmony

Sandholm: The learning part always interests people, but it's not really the only part here -- or not even the main part. We basically have three main modules in our architecture. One computes approximations of Nash equilibrium strategies using only the rules of the game as input. In other words, game-theoretic strategies.

That doesn’t take any data as input, just the rules of the game. The second part is during play, refining that strategy. We call that subgame solving.

Then the third part is the learning part, or the self-improvement part. And there, traditionally people have done what’s called opponent modeling and opponent exploitation, where you try to model the opponent or opponents and adjust your strategies so as to take advantage of their weaknesses.

However, when we go against these absolute best human strategies, the best human players in the world, I felt that they don't have that many holes to exploit and they are experts at counter-exploiting. When you start to exploit opponents, you typically open yourself up for exploitation, and we didn't want to take that risk. In the learning part, the third part, we took a totally different approach than traditionally is taken in AI.

We are letting the opponents tell us where the holes are in our strategy. Then, in the background, using supercomputing, we are fixing those holes.

We said, “Okay, we are going to play according to our approximate game-theoretic strategies. However, if we see that the opponents have been able to find some mistakes in our strategy, then we will actually fill those mistakes and compute an even closer approximation to game-theoretic play in those spots.”

One way to think about that is that we are letting the opponents tell us where the holes are in our strategy. Then, in the background, using supercomputing, we are fixing those holes.

All three of these modules run on the Bridges supercomputer at the Pittsburgh Supercomputing Center (PSC), for which the hardware was built by Hewlett Packard Enterprise (HPE).

HPC from HPE

Overcomes Barriers

To Supercomputing and Deep Learning

Gardner: Is this being used in any business settings? It certainly seems like there's potential there for a lot of use cases. Business competition and circumstances seem to have an affinity for what you're describing in the poker use case. Where are you taking this next?

Sandholm: So far this, to my knowledge, has not been used in business. One of the reasons is that we have just reached the superhuman level in January 2017. And, of course, if you think about your strategic reasoning problems, many of them are very important, and you don't want to delegate them to AI just to save time or something like that.

Now that the AI is better at strategic reasoning than humans, that completely shifts things. I believe that in the next few years it will be a necessity to have what I call strategic augmentation. So you can't have just people doing business strategy, negotiation, strategic pricing, and product portfolio optimization.

You are going to have to have better strategic reasoning to support you, and so it becomes a kind of competition. So if your competitors have it, or even if they don't, you better have it because it’s a competitive advantage.

Gardner: So a lot of what we're seeing in AI and machine learning is to find the things that the machines do better and allow the humans to do what they can do even better than machines. Now that you have this new capability with strategic reasoning, where does that demarcation come in a business setting? Where do you think that humans will be still paramount, and where will the machines be a very powerful tool for them?

Human modeling, AI solving

Sandholm: At least in the foreseeable future, I see the demarcation as being modeling versus solving. I think that humans will continue to play a very important role in modeling their strategic situations, just to know everything that is pertinent and deciding what’s not pertinent in the model, and so forth. Then the AI is best at solving the model.

That's the demarcation, at least for the foreseeable future. In the very long run, maybe the AI itself actually can start to do the modeling part as well as it builds a better understanding of the world -- but that is far in the future.

Gardner: Looking back as to what is enabling this, clearly the software and the algorithms and finding the right benchmark, in this case the poker game are essential. But with that large of a data set potential -- probabilities set like you mentioned -- the underlying computersystems must need to keep up. Where are you in terms of the threshold that holds you back? Is this a price issue that holds you back? Is it a performance limit, the amount of time required? What are the limits, the governors to continuing?

Sandholm: It's all of the above, and we are very fortunate that we had access to Bridges; otherwise this wouldn’t have been possible at all.  We spent more than a year and needed about 25 million core hours of computing and 2.6 petabytes of data storage.

This amount is necessary to conduct serious absolute superhuman research in this field -- but it is something very hard for a professor to obtain. We were very fortunate to have that computing at our disposal.

Gardner: Let's examine the commercialization potential of this. You're not only a professor at Carnegie Mellon, you’re a founder and CEO of a few companies. Tell us about your companies and how the research is leading to business benefits.

Superhuman business strategies

Sandholm: Let’s start with Strategic Machine, a brand-new start-up company, all of two months old. It’s already profitable, and we are applying the strategic reasoning technology, which again is application independent, along with the Libratus technology, the Lengpudashi technology, and a host of other technologies that we have exclusively licensed to Strategic Machine. We are doing research and development at Strategic Machine as well, and we are taking these to any application that wants us.

HPC from HPE

Overcomes Barriers 

To Supercomputing and Deep Learning

Such applications include business strategy optimization, automated negotiation, and strategic pricing. Typically when people do pricing optimization algorithmically, they assume that either their company is a monopolist or the competitors’ prices are fixed, but obviously neither is typically true.

We are looking at how do you price strategically where you are taking into account the opponent’s strategic response in advance. So you price into the future, instead of just pricing reactively. The same can be done for product portfolio optimization along with pricing.

Let's say you're a car manufacturer and you decide what product portfolio you will offer and at what prices. Well, what you should do depends on what your competitors do and vice versa, but you don’t know that in advance. So again, it’s an imperfect-information game.

Gardner: And these are some of the most difficult problems that businesses face. They have huge billion-dollar investments that they need to line up behind for these types of decisions. Because of that pipeline, by the time they get to a dynamic environment where they can assess -- it's often too late. So having the best strategic reasoning as far in advance as possible is a huge benefit.

If you think about machine learning traditionally, it's about learning from the past. But strategic reasoning is all about figuring out what's going to happen in the future.

Sandholm: Exactly! If you think about machine learning traditionally, it's about learning from the past. But strategic reasoning is all about figuring out what's going to happen in the future. And you can marry these up, of course, where the machine learning gives the strategic reasoning technology prior beliefs, and other information to put into the model.

There are also other applications. For example, cyber security has several applications, such as zero-day vulnerabilities. You can run your custom algorithms and standard algorithms to find them, and what algorithms you should run depends on what the other opposing governments run -- so it is a game.

Similarly, once you find them, how do you play them? Do you report your vulnerabilities to Microsoft? Do you attack with them, or do you stockpile them? Again, your best strategy depends on what all the opponents do, and that's also a very strategic application.

And in upstairs blocks trading, in finance, it’s the same thing: A few players, very big, very strategic.

Gaming your own immune system

The most radical application is something that we are working on currently in the lab where we are doing medical treatment planning using these types of sequential planning techniques. We're actually testing how well one can steer a patient's T-cell population to fight cancers, autoimmune diseases, and infections better by not just using one short treatment plan -- but through sophisticated conditional treatment plans where the adversary is actually your own immune system.

Gardner: Or cancer is your opponent, and you need to beat it?

Sandholm: Yes, that’s right. There are actually two different ways to think about that, and they lead to different algorithms. We have looked at it where the actual disease is the opponent -- but here we are actually looking at how do you steer your own T-cell population.

Gardner: Going back to the technology, we've heard quite a bit from HPE about more memory-driven and edge-driven computing, where the analysis can happen closer to where the data is gathered. Are these advances of any use to you in better strategic reasoning algorithmic processing?

Algorithms at the edge

Sandholm: Yes, absolutely! We actually started running at the PSC on an earlier supercomputer, maybe 10 years ago, which was a shared-memory architecture. And then with Bridges, which is mostly a distributed system, we used distributed algorithms. As we go into the future with shared memory, we could get a lot of speedups.

We have both types of algorithms, so we know that we can run on both architectures. But obviously, the shared-memory, if it can fit our models and the dynamic state of the algorithms, is much faster.

Gardner: So the HPE Machine must be of interest to you: HPE’s advanced concept demonstration model, with a memory-driven architecture, photonics for internal communications, and so forth. Is that a technology you're keeping a keen eye on?

HPC from HPE

Overcomes Barriers 

To Supercomputing and Deep Learning

Sandholm: Yes. That would definitely be a desirable thing for us, but what we really focus on is the algorithms and the AI research. We have been very fortunate in that the PSC and HPE have been able to take care of the hardware side.

We really don’t get involved in the hardware side that much, and I'm looking at it from the outside. I'm trusting that they will continue to build the best hardware and maintain it in the best way -- so that we can focus on the AI research.

Gardner: Of course, you could help supplement the cost of the hardware by playing superhuman poker in places like Las Vegas, and perhaps doing quite well.

Sandholm: Actually here in the live game in Las Vegas they don't allow that type of computational support. On the Internet, AI has become a big problem on gaming sites, and it will become an increasing problem. We don't put our AI in there; it’s against their site rules. Also, I think it's unethical to pretend to be a human when you are not. The business opportunities, the monetary opportunities in the business applications, are much bigger than what you could hope to make in poker anyway.

Listen to the podcast. Find it on iTunes. Get the mobile app. Read a full transcript ordownload a copy. Sponsor: Hewlett Packard Enterprise.

You may also be interested in:

How IoT and OT collaborate to usher in the data-driven factory of the future

The next BriefingsDirect Internet of Things (IoT) technology trends interview explores how innovation is impacting modern factories and supply chains.

We’ll now learn how a leading-edge manufacturer, Hirotec, in the global automotive industry, takes advantage of IoT and Operational Technology (OT) combined to deliver dependable, managed, and continuous operations.

Listen to the podcast. Find it on iTunes. Get the mobile app. Read a full transcript or download a copy.

Here to help us to find the best factory of the future attributes is Justin Hester, Senior Researcher in the IoT Lab at Hirotec Corp. in Hiroshima, Japan. The discussion is moderated by Dana Gardner, Principal Analyst at Interarbor Solutions.

Here are some excerpts:

Gardner: What's happening in the market with business and technology trends that’s driving this need for more modern factories and more responsive supply chains?

Hester: Our customers are demanding shorter lead times. There is a drive for even higher quality, especially in automotive manufacturing. We’re also seeing a much higher level of customization requests coming from our customers. So how can we create products that better match the unique needs of each customer?

As we look at how we can continue to compete in an ever-competitive environment, we are starting to see how the solutions from IoT can help us.

Gardner: What is it about IoT and Industrial IoT (IIoT) that allows you to do things that you could not have done before?

Hester: Within the manufacturing space, a lot of data has been there for years; for decades. Manufacturing has been very good at collecting data. The challenges we've had, though, is bringing in that data in real-time, because the amount of data is so large. How can we act on that data quicker, not on a day-by-day basis or week-by-week basis, but actually on a minute-by-minute basis, or a second-by-second basis? And how do we take that data and contextualize it?

Hester

Hester

It's one thing in a manufacturing environment to say, “Okay, this machine is having a challenge.” But it’s another thing if I can say, “This machine is having a challenge, and in the context of the factory, here's how it's affecting downstream processes, and here's what we can do to mitigate those downstream challenges that we’re going to have.” That’s where IoT starts bringing us a lot of value.

The analytics, the real-time contextualization of that data that we’ve already had in the manufacturing area, is very helpful.

Gardner: So moving from what may have been a gather, batch, analyze, report process -- we’re now taking more discrete analysis opportunities and injecting that into a wider context of efficiency and productivity. So this is a fairly big change. This is not incremental; this is a step-change advancement, right?

A huge step-change 

Hester: It’s a huge change for the market. It's a huge change for us at Hirotec. One of the things we like to talk about is what we jokingly call the Tuesday Morning Meeting. We talk about this idea that in the morning at a manufacturing facility, everyone gets together and talks about what happened yesterday, and what we can do today to make up for what happened yesterday.

Instead, now we’re making that huge step-change to say,  “Why don't we get the data to the right people with the right context and let them make a decision so they can affect what's going on, instead of waiting until tomorrow to react to what's going on?” It’s a huge step-change. We’re really looking at it as how can we take small steps right away to get to that larger goal.

In manufacturing areas, there's been a lot of delay, confusion, and hesitancy to move forward because everyone sees the value, but it's this huge change, this huge project. At Hirotec, we’re taking more of a scaled approach, and saying let's start small, let’s scale up, let’s learn along the way, let's bring value back to the organization -- and that's helped us move very quickly.

Gardner: We’d like to hear more about that success story but in the meantime, tell us about Hirotec for those who don't know of it. What role do you play in the automotive industry, and how are you succeeding in your markets?

Hester: Hirotec is a large, tier-1 automotive supplier. What that means is we supply parts and systems directly to the automotive original equipment manufacturers (OEMs), like Mazda, General Motors, FCA, Ford, and we specialize in door manufacturing, as well as exhaust system manufacturing. So every year we make about 8 million doors, 1.8 million exhaust systems, and we provide those systems mainly to Mazda and General Motors, but also we provide that expertise through tooling.

For example, if an automotive OEM would like Hirotec’s expertise in producing these parts, but they would like to produce them in-house, Hirotec has a tooling arm where we can provide that tooling for automotive manufacturing. It's an interesting strategy that allows us to take advantage of data both in our facilities, but then also work with our customers on the tooling side to provide those lessons learned and bring them value there as well.

Gardner: How big of a distribution are we talking about? How many factories, how many countries; what’s the scale here?

Hester: We are based in Hiroshima, Japan, but we’re actually in nine countries around the world, currently with 27 facilities. We have reached into all the major continents with automotive manufacturing: we’re in North America, we’re in Europe, we’re all throughout Asia, in China and India. We have a large global presence. Anywhere you find automotive manufacturing, we’re there supporting it.

Discover How the

IoT Advantage

Works in Multiple Industries

Gardner: With that massive scale, very small improvements can turn into very big benefits. Tell us why the opportunity in a manufacturing environment to eke out efficiency and productivity has such big payoffs.

Hester: So especially in manufacturing, what we find when we get to those large scales like you're alluding to is that a 1 percent or 2 percent improvement has huge financial benefits. And so the other thing is in manufacturing, especially automotive manufacturing, we tend to standardize our processes, and within Hirotec, we’ve done a great job of standardizing that world-class leadership in door manufacturing.

And so what we find is when we get improvements not only in IoT but anywhere in manufacturing, if we can get 1 percent or 2 percent, not only is that a huge financial benefit but because we standardized globally, we can move that to our other facilities very quickly, doubling down on that benefit.

Gardner: Well, clearly Hirotec sees this as something to really invest in, they’ve created the IoT Lab. Tell me a little bit about that and how that fits into this?

The IoT Lab works

Hester: The IoT Lab is a very exciting new group, it's part of our Advanced Engineering Center (AEC). The AEC is a group out of our global headquarters and this group is tasked with the five- to 10-year horizon. So they're able to work across all of our global organizations with tooling, with engineering, with production, with sales, and even our global operations groups. Our IoT group goes and finds solutions that can bring value anywhere in the organization through bringing in new technologies, new ideas, and new solutions.

And so we formed the IoT Lab to find how can we bring IoT-based solutions into the manufacturing space, into the tooling space, and how actually can those solutions not only help our manufacturing and tooling teams but also help our IT teams, our finance teams, and our sales teams.

Gardner: Let's dig back down a little bit into why IT, IoT and Operational Technology (OT) are into this step-change opportunity, looking for some significant benefits but being careful in how to institute that. What is required when you move to a more an IT-focused, a standard-platform approach -- across all the different systems -- that allows you to eke these great benefits?

Tell us about how IoT as a concept is working its way into the very edge of the factory floor.

Discover How the

IoT Advantage

Works in Multiple Industries

Hester: One of the things we’re seeing is that IT is beginning to meld, like you alluded to, with OT -- and there really isn't a distinction between OT and IT anymore. What we're finding is that we’re starting to get to these solution levels by working with partners such as PTC and Hewlett Packard Enterprise (HPE) to bring our IT group and our OT group all together within Hirotec and bring value to the organization.

What we find is there is no longer a need in OT that becomes a request for IT to support it, and also that IT has a need and so they go to OT for support. What we are finding is we have organizational needs, and we’re coming to the table together to make these changes. And that actually within itself is bringing even more value to the organization.

Instead of coming last-minute to the IT group and saying, “Hey, we need your support for all these different solutions, and we’ve already got everything set, and you are just here to put it in,” what we are seeing, is that they bring the expertise in, help us out upfront, and we’re finding better solutions because we are getting experts both from OT and IT together.

We are seeing this convergence of these two teams working on solutions to bring value. And they're really moving everything to the edge. So where everyone talks about cloud-based computing -- or maybe it’s in their data center -- where we are finding value is in bringing all of these solutions right out to the production line.

We are doing data collection right there, but we are also starting to do data analytics right at the production line level, where it can bring the best value in the fastest way.

Gardner: So it’s an auspicious time because just as you are seeking to do this, the providers of technology are creating micro data centers, and they are creating Edgeline converged systems, and they are looking at energy conservation so that they can do this in an affordable way -- and with storage models that can support this at a competitive price.

What is it about the way that IT is evolving and providing platforms and systems that has gotten you and The IoT Lab so excited?

Excitement at the edge  

Hester: With IoT and IT platforms, originally to do the analytics, we had to go up to the cloud -- that was the only place where the compute power existed. Solution providers now are bringing that level of intelligence down to the edge. We’re hearing some exciting things from HPE on memory-driven computing, and that's huge for us because as we start doing these very complex analytics at the edge, we need that power, that horsepower, to run different applications at the same time at the production line. And something like memory-driven solutions helps us accomplish that.

It's one thing to have higher-performance computing, but another thing to gain edge computing that's proper for the factory environment. In a manufacturing environment it's not conducive to a standard servers, a standard rack where it needs dust protection and heat protection -- that doesn't exist in a manufacturing environment.

The other thing we're beginning to see with edge computing, that HPE provides with Edgeline products, is that we have computers that have high power, high ability to perform the analytics and data collection capabilities -- but they're also proper for the environment.

I don't need to build out a special protection unit with special temperature control, humidity control – all of which drives up energy costs, which drives up total costs. Instead, we’re able to run edge computing in the environment as it should be on its own, protected from what comes in a manufacturing environment -- and that's huge for us.

Gardner: They are engineering these systems now with such ruggedized micro facilities in mind. It's quite impressive that the very best of what a data center can do, can now be brought to the very worst types of environments. I'm sure we'll see more of that, and I am sure we'll see it get even smaller and more powerful.

Do you have any examples of where you have already been able to take IoT in the confluence of OT and IT to a point where you can demonstrate entirely new types of benefits? I know this is still early in the game, but it helps to demonstrate what you can do in terms of efficiency, productivity, and analytics. What are you getting when you do this well?

IoT insights save time and money

Hester: Taking the stepped strategy that we have, we actually started at Hirotec very small with only eight machines in North America and we were just looking to see if the machines are on, are they running, and even from there, we saw a value because all of a sudden we were getting that real-time contextualized insight into the whole facility. We then quickly moved over to one of our production facilities in Japan, where we have a brand-new robotic inspection system, and this system uses vision sensors, laser sensors, force sensors -- and it's actually inspecting exhaust systems before they leave the facility.

We very quickly implemented an IoT solution in that area, and all we did was we said, “Hey, we just want to get insight into the data, so we want to be able to see all these data points. Over 400 data points are created every inspection. We want to be able to see this data, compared in historical ways -- so let’s bring context to that data, and we want to provide it in real-time.”

Discover How the

IoT Advantage

Works in Multiple Industries

What we found from just those two projects very quickly is that we're bringing value to the organization because now our teams can go in and say, “Okay, the system is doing its job, it's inspecting things before they leave our facility to make sure our customers always get a high-quality product.” But now, we’re able to dive in and find different trends that we weren't able to see before because all we were doing is saying, “Okay, this system leaves the facility or this system doesn't.”

And so already just from that application, we’ve been able to find ways that our engineers can even increase the throughput and the reliability of the system because now they have these historical trends. They were able to do a root-cause analysis on some improvements that would have taken months of investigation; it was completed in less than a week for us.

And so that's a huge value -- not only in that my project costs go down but now I am able to impact the organization quicker, and that's the big thing that Hirotec is seeing. It’s one thing to talk about the financial cost of a project, or I can say, “Okay, here is the financial impact,” but what we are seeing is that we’re moving quicker.

And so, we're having long-term financial benefits because we’re able to react to things much faster. In this case, we’re able to reduce months of investigation down to a week. That means that when I implement my solution quicker, I'm now bringing that impact to the organization even faster, which has long-term benefits. We are already seeing those benefits today.

Gardner: You’ll obviously be able to improve quality, you’ll be able to reduce the time to improving that quality, gain predictive analytics in your operations, but also it sounds like you are going to gain metadata insights that you can take back into design for the next iteration of not only the design for the parts but the design for the tooling as well and even the operations around that. So that intelligence at the edge can be something that is a full lifecycle process, it goes right back to the very initiation of both the design and the tooling.

Data-driven design, decisions 

Hester: Absolutely, and so, these solutions, they can't live in a silo. We're really starting to look at these ideas of what some people call the Digital Thread, the Digital Twin. We’re starting to understand what does that mean as you loop this data back to our engineering teams -- what kind of benefits can we see, how can we improve our processes, how can we drive out into the organization?

And one of the biggest things with IoT-based solutions is that they can't stay inside this box, where we talked about OT to IT, we are talking about manufacturing, engineering, these IoT solutions at their best, all they really do is bring these groups together and bring a whole organization together with more contextualized data to make better decisions faster.

And so, exactly to your point, as we are looping back, we’re able to start understanding the benefit we’re going to be seeing from bringing these teams together.

Gardner: One last point before we close out. It seems to me as well that at a macro level, this type of data insight and efficiency can be brought into the entire supply chain. As you're providing certain elements of an automobile, other suppliers are providing what they specialize in, too, and having that quality control and integration and reduced time-to-value or mean-time-to-resolution of the production issues, and so forth, can be applied at a macro level.

So how does the automotive supplier itself look at this when it can take into consideration all of its suppliers like Hirotec are doing?

Start small 

Hester: It's a very early phase, so a lot of the suppliers are starting to understand what this means for them. There is definitely a macro benefit that the industry is going to see in five to 10 years. Suppliers now need to start small. One of my favorite pictures is a picture of the ocean and a guy holding a lighter. It [boiling the ocean] is not going to happen. So we see these huge macro benefits of where we’re going, but we have to start out somewhere.

Discover How the

IoT Advantage

Works in Multiple Industries

A lot of suppliers, what we’re recommending to them, is to do the same thing we did, just start small with a couple of machines, start getting that data visualized, start pulling that data into the organization. Once you do that, you start benefiting from the data, and then start finding new use-cases.

As these suppliers all start doing their own small projects and working together, I think that's when we are going to start to see the macro benefits but in about five to 10 years out in the industry.

Listen to the podcast. Find it on iTunes. Get the mobile app. Read a full transcript or download a copy. Sponsor: Hewlett Packard Enterprise.

You may also be interested in:

·       DreamWorks Animation crafts its next era of dynamic IT infrastructure

·       How Enterprises Can Take the Ecosystem Path to Making the Most of Microsoft Azure Stack Apps

·       Hybrid Cloud ecosystem readies for impact from Microsoft Azure Stack

·       Converged IoT systems: Bringing the data center to the edge of everything

·       IDOL-powered appliance delivers better decisions via comprehensive business information searches

·        OCSL sets its sights on the Nirvana of hybrid IT—attaining the right mix of hybrid cloud for its clients

·       Fast acquisition of diverse unstructured data sources makes IDOL API tools a star at LogitBot

·       How lastminute.com uses machine learning to improve travel bookings user experience

·       Veikkaus digitally transforms as it emerges as new combined Finnish national gaming company

 ·       HPE takes aim at customer needs for speed and agility in age of IoT, hybrid everything

How Imagine Communications leverages edge computing and HPC for live multiscreen IP video

The next BriefingsDirect Voice of the Customer HPC and edge computing strategies interview explores how a video delivery and customization capability has moved to the network edge -- and closer to consumers -- to support live, multi-screen Internet Protocol (IP) entertainment delivery. 

Listen to the podcast. Find it on iTunes. Get the mobile app. Read a full transcript or download a copy.

We’ll learn how hybrid technology and new workflows for IP-delivered digital video are being re-architected -- with significant benefits to the end-user experience, as well as with new monetization values to the content providers.

Our guest is Glodina Connan-Lostanlen, Chief Marketing Officer at Imagine Communications in Frisco, Texas. The discussion is moderated by Dana Gardner, Principal Analyst at Interarbor Solutions.

Here are some excerpts:

Gardner: Your organization has many major media clients. What are the pressures they are facing as they look to the new world of multi-screen video and media?

Connan-Lostanlen: The number-one concern of the media and entertainment industry is the fragmentation of their audience. We live with a model supported by advertising and subscriptions that rely primarily on linear programming, with people watching TV at home.

Connan-Lostanlen

Connan-Lostanlen

And guess what? Now they are watching it on the go -- on their telephones, on their iPads, on their laptops, anywhere. So they have to find the way to capture that audience, justify the value of that audience to their advertisers, and deliver video content that is relevant to them. And that means meeting consumer demand for several types of content, delivered at the very time that people want to consume it.  So it brings a whole range of technology and business challenges that our media and entertainment customers have to overcome. But addressing these challenges with new technology that increases agility and velocity to market also creates opportunities.

For example, they can now try new content. That means they can try new programs, new channels, and they don’t have to keep them forever if they don’t work. The new models create opportunities to be more creative, to focus on what they are good at, which is creating valuable content. At the same time, they have to make sure that they cater to all these different audiences that are either static or on the go.

Gardner: The media industry has faced so much change over the past 20 years, but this is a major, perhaps once-in-a-generation, level of change -- when you go to fully digital, IP-delivered content.

As you say, the audience is pulling the providers to multi-screen support, but there is also the capability now -- with the new technology on the back-end -- to have much more of a relationship with the customer, a one-to-one relationship and even customization, rather than one-to-many. Tell us about the drivers on the personalization level.

Connan-Lostanlen: That’s another big upside of the fragmentation, and the advent of IP technology -- all the way from content creation to making a program and distributing it. It gives the content creators access to the unique viewers, and the ability to really engage with them -- knowing what they like -- and then to potentially target advertising to them. The technology is there. The challenge remains about how to justify the business model, how to value the targeted advertising; there are different opinions on this, and there is also the unknown or the willingness of several generations of viewers to accept good advertising.

That is a great topic right now, and very relevant when we talk about linear advertising and dynamic ad insertion (DAI). Now we are able to -- at the very edge of the signal distribution, the video signal distribution -- insert an ad that is relevant to each viewer, because you know their preferences, you know who they are, and you know what they are watching, and so you can determine that an ad is going to be relevant to them.

But that means media and entertainment customers have to revisit the whole infrastructure. It’s not necessary rebuilding, they can put in add-ons. They don’t have to throw away what they had, but they can maintain the legacy infrastructure and add on top of it the IP-enabled infrastructure to let them take advantage of these capabilities.

Gardner: This change has happened from the web now all the way to multi-screen. With the web there was a model where you would use a content delivery network (CDN) to take the object, the media object, and place it as close to the edge as you could. What’s changed and why doesn’t that model work as well?

Connan-Lostanlen: I don’t know yet if I want to say that model doesn’t work anymore. Let’s let the CDN providers enhance their technology. But for sure, the volume of videos that we are consuming everyday is exponentially growing. That definitely creates pressure in the pipe. Our role at the front-end and the back-end is to make sure that videos are being created in different formats, with different ads, and everything else, in the most effective way so that it doesn’t put an undue strain on the pipe that is distributing the videos.

We are being pushed to innovate further on the type of workflows that we are implementing at our customers’ sites today, to make it efficient, to not leave storage at the edge and not centrally, and to do transcoding just-in-time. These are the things that are being worked on. It’s a balance between available capacity and the number of programs that you want to send across to your viewers – and how big your target market is.

The task for us on the back-end is to rethink the workflows in a much more efficient way. So, for example, this is what we call the digital-first approach, or unified distribution. Instead of planning a linear channel that goes the traditional way and then adding another infrastructure for multi-screen, on all those different platforms and then cable, and satellite, and IPTV, etc. -- why not design the whole workflow digital-first. This frees the content distributor or provider to hold off on committing to specific platforms until the video has reached the edge. And it’s there that the end-user requirements determine how they get the signal.

This is where we are going -- to see the efficiencies happen and so remove the pressure on the CDNs and other distribution mechanisms, like over-the-air.

Explore

High-Performance Computing

Solutions from HPE

Gardner: It means an intelligent edge capability, whereas we had an intelligent core up until now. We’ll also seek a hybrid capability between them, growing more sophisticated over time.

We have a whole new generation of technology for video delivery. Tell us about Imagine Communications. How do you go to market? How do you help your customers?

Education for future generations

Connan-Lostanlen: Two months ago we were in Las Vegas for our biggest tradeshow of the year, the NAB Show. At the event, our customers first wanted to understand what it takes to move to IP -- so the “how.” They understand the need to move to IP, to take advantage of the benefits that it brings. But how do they do this, while they are still navigating the traditional world?

It’s not only the “how,” it’s needing examples of best practices. So we instructed them in a panel discussion, for example, on Over the Top Technology (OTT), which is another way of saying IP-delivered, and what it takes to create a successful multi-screen service. Part of the panel explained what OTT is, so there’s a lot of education.

There is also another level of education that we have to provide, which is moving from the traditional world of serial digital interfaces (SDIs) in the broadcast industry to IP. It’s basically saying analog video signals can be moved into digital. Then not only is there a digitally sharp signal, it’s an IP stream. The whole knowledge about how to handle IP is new to our own industry, to our own engineers, to our own customers. We also have to educate on what it takes to do this properly.

One of the key things in the media and entertainment industry is that there’s a little bit of fear about IP, because no one really believed that IP could handle live signals. And you know how important live television is in this industry – real-time sports and news -- this is where the money comes from. That’s why the most expensive ads are run during the Super Bowl.

It’s essential to be able to do live with IP – it’s critical. That’s why we are sharing with our customers the real-life implementations that we are doing today.

We are also pushing multiple standards forward. We work with our competitors on these standards. We have set up a trade association to accelerate the standards work. We did all of that. And as we do this, it forces us to innovate in partnership with customers and bring them on board. They are part of that trade association, they are part of the proof-of-concept trials, and they are gladly sharing their experiences with others so that the transition can be accelerated.

Gardner: Imagine Communications is then a technology and solutions provider to the media content companies, and you provide the means to do this. You are also doing a lot with ad insertion, billing, in understanding more about the end-user and allowing that data flow from the edge back to the core, and then back to the edge to happen.

At the heart of it all

Connan-Lostanlen: We do everything that happens behind the camera -- from content creation all the way to making a program and distributing it. And also, to your point, on monetizing all that with a management system. We have a long history of powering all the key customers in the world for their advertising system. It’s basically an automated system that allows the selling of advertising spots, and then to bill them -- and this is the engine of where our customers make money. So we are at the heart of this.

We are in the prime position to help them take advantage of the new advertising solutions that exist today, including dynamic ad insertion. In other words, how you target ads to the single viewer. And the challenge for them is now that they have a campaign, how do they design it to cater both to the linear traditional advertising system as well as the multi-screen or web mobile application? That's what we are working on. We have a whole set of next-generation platforms that allow them to take advantage of both in a more effective manner.

Gardner: The technology is there, you are a solutions provider. You need to find the best ways of storing and crunching data, close to the edge, and optimizing networks. Tell us why you choose certain partners and what are the some of the major concerns you have when you go to the technology marketplace?

Connan-Lostanlen: One fundamental driver here, as we drive the transition to IP in this industry, is in being able to rely on consumer-off-the-shelf (COTS) platforms. But even so, not all COTS platforms are born equal, right?

For compute, for storage, for networking, you need to rely on top-scale hardware platforms, and that’s why about two years ago we started to work very closely with Hewlett Packard Enterprise (HPE) for both our compute and storage technology.

Explore

High-Performance Computing

Solutions from HPE

We develop the software appliances that run on those platforms, and we sell this as a package with HPE. It’s been a key value proposition of ours as we began this journey to move to IP. We can say, by the way, our solutions run on HPE hardware. That's very important because having high-performance compute (HPC) that scales is critical to the broadcast and media industry. Having storage that is highly reliable is fundamental because going off the air is not acceptable. So it's 99.9999 percent reliable, and that’s what we want, right?

It’s a fundamental part of our message to our customers to say, “In your network, put Imagine solutions, which are powered by one of the top compute and storage technologies.”

Gardner: Another part of the change in the marketplace is this move to the edge. It’s auspicious that just as you need to have more storage and compute efficiency at the edge of the network, close to the consumer, the infrastructure providers are also designing new hardware and solutions to do just that. That's also for the Internet of Things (IoT) requirements, and there are other drivers. Nonetheless, it's an industry standard approach.

What is it about HPE Edgeline, for example, and the architecture that HPE is using, that makes that edge more powerful for your requirements? How do you view this architectural shift from core data center to the edge?

Optimize the global edge

Connan-Lostanlen: It's a big deal because we are going to be in a hybrid world. Most of our customers, when they hear about cloud, we have to explain it to them. We explain that they can have their private cloud where they can run virtualized applications on-premises, or they can take advantage of public clouds.

Being able to have a hybrid model of deployment for their applications is critical, especially for large customers who have operations in several places around the globe. For example, such big names as Disney, Turner –- they have operations everywhere. For them, being able to optimize at the edge means that you have to create an architecture that is geographically distributed -- but is highly efficient where they have those operations. This type of technology helps us deliver more value to the key customers.

Gardner: The other part of that intelligent edge technology is that it has the ability to be adaptive and customized. Each region has its own networks, its own regulation, and its own compliance, security, and privacy issues. When you can be programmatic as to how you design your edge infrastructure, then a custom-applications-orientation becomes possible.

Is there something about the edge architecture that you would like to see more of? Where do you see this going in terms of the capabilities of customization added-on to your services?

Connan-Lostanlen: One of the typical use-cases that we see for those big customers who have distributed operations is that they like to try and run their disaster recovery (DR) site in a more cost-effective manner. So the flexibility that an edge architecture provides to them is that they don’t have to rely on central operations running DR for everybody. They can do it on their own, and they can do it cost-effectively. They don't have to recreate the entire infrastructure, and so they do DR at the edge as well.

We especially see this a lot in the process of putting the pieces of the program together, what we call “play out,” before it's distributed. When you create a TV channel, if you will, it’s important to have end-to-end redundancy -- and DR is a key driver for this type of application.

Gardner: Are there some examples of your cutting-edge clients that have adopted these solutions? What are the outcomes? What are they able to do with it?

Pop-up power

Connan-Lostanlen: Well, it’s always sensitive to name those big brand names. They are very protective of their brands. However, one of the top ones in the world of media and entertainment has decided to move all of their operations -- from content creation, planning, and distribution -- to their own cloud, to their own data center.

They are at the forefront of playing live and recorded material on TV -- all from their cloud. They needed strong partners in data centers. So obviously we work with them closely, and the reason why they do this is simply to really take advantage of the flexibility. They don't want to be tied to a restricted channel count; they want to try new things. They want to try pop-up channels. For the Oscars, for example, it’s one night. Are you going to recreate the whole infrastructure if you can just check it on and off, if you will, out of their data center capacity? So that's the key application, the pop-up channels and ability to easily try new programs.

Gardner: It sounds like they are thinking of themselves as an IT company, rather than a media and entertainment company that consumes IT. Is that shift happening?

Connan-Lostanlen: Oh yes, that's an interesting topic, because I think you cannot really do this successfully if you don’t start to think IT a little bit. What we are seeing, interestingly, is that our customers typically used to have the IT department on one side, the broadcast engineers on the other side -- these were two groups that didn't speak the same language. Now they get together, and they have to, because they have to design together the solution that will make them more successful. We are seeing this happening.

I wouldn't say yet that they are IT companies. The core strength is content, that is their brand, that's what they are good at -- creating amazing content and making it available to as many people as possible.

They have to understand IT, but they can't lose concentration on their core business. I think the IT providers still have a very strong play there. It's always happening that way.

In addition to disaster recovery being a key application, multi-screen delivery is taking advantage of that technology, for sure.

Explore

High-Performance Computing

Solutions from HPE

Gardner: These companies are making this cultural shift to being much more technically oriented. They think about standard processes across all of what they do, and they have their own core data center that's dynamic, flexible, agile and cost-efficient. What does that get for them? Is it too soon, or do we have some metrics of success for companies that make this move toward a full digitally transformed organization?

Connan-Lostanlen: They are very protective about the math. It is fair to say that the up-front investments may be higher, but when you do the math over time, you do the total cost of ownership for the next 5 to 10 years -- because that’s typically the life cycle of those infrastructures – then definitely they do save money. On the operational expenditure (OPEX) side [of private cloud economics] it’s much more efficient, but they also have upside on additional revenue. So net-net, the return on investment (ROI) is much better. But it’s kind of hard to say now because we are still in the early days, but it’s bound to be a much greater ROI.

Another specific DR example is in the Middle East. We have a customer there who decided to operate the DR and IP in the cloud, instead of having a replicated system with satellite links in between. They were able to save $2 million worth of satellite links, and that data center investment, trust me, was not that high. So it shows that the ROI is there.

My satellite customers might say, “Well, what are you trying to do?” The good news is that they are looking at us to help them transform their businesses, too. So big satellite providers are thinking broadly about how this world of IP is changing their game. They are examining what they need to do differently. I think it’s going to create even more opportunities to reduce costs for all of our customers.

IT enters a hybrid world

Gardner: That's one of the intrinsic values of a hybrid IT approach -- you can use many different ways to do something, and then optimize which of those methods works best, and also alternate between them for best economics. That’s a very powerful concept.

Connan-Lostanlen: The world will be a hybrid IT world, and we will take advantage of that. But, of course, that will come with some challenges. What I think is next is the number-one question that I get asked.

Three years ago costumers would ask us, “Hey, IP is not going to work for live TV.” We convinced them otherwise, and now they know it’s working, it’s happening for real.

Secondly, they are thinking, “Okay, now I get it, so how do I do this?” We showed them, this is how you do it, the education piece.

Now, this year, the number-one question is security. “Okay, this is my content, the most valuable asset I have in my company. I am not putting this in the cloud,” they say. And this is where another piece of education has to start, which is: Actually, as you put stuff on your cloud, it’s more secure.

And we are working with our technology providers. As I said earlier, the COTS providers are not equal. We take it seriously. The cyber attacks on content and media is critical, and it’s bound to happen more often.

Initially there was a lack of understanding that you need to separate your corporate network, such as emails and VPNs, from you broadcast operations network. Okay, that’s easy to explain and that can be implemented, and that's where most of the attacks over the last five years have happened. This is solved.

They are going to get right into the servers, into the storage, and try to mess with it over there. So I think it’s super important to be able to say, “Not only at the software level, but at the hardware firmware level, we are adding protection against your number-one issue, security, which everybody can see is so important.”

However, the cyber attackers are becoming more clever, so they will overcome these initial defenses.They are going to get right into the servers, into the storage, and try to mess with it over there. So I think it’s super important to be able to say, “Not only at the software level, but at the hardware firmware level, we are adding protection against your number-one issue, security, which everybody can see is so important.”

Gardner: Sure, the next domino to fall after you have the data center concept, the implementation, the execution, even the optimization, is then to remove risk, whether it's disaster recovery, security, right down to the silicon and so forth. So that’s the next thing we will look for, and I hope I can get a chance to talk to you about how you are all lowering risk for your clients the next time we speak.

Explore

High-Performance Computing

Solutions from HPE

Listen to the podcast. Find it on iTunes. Get the mobile app. Read a full transcript or download a copy. Sponsor: Hewlett Packard Enterprise.

You may also be interested in: