1.1 — Transforming Data Management with John Roese

Technology is shifting to a decentralized model – We’ll talk about what you need to know today, including how data management is transforming and how distributed IT infrastructures is paving the way for a multi-cloud environment.
All The Next Horizon Podcasts

The tech environment cyclically changes from centralized to decentralized – but what is driving that change? In this episode, Bill sits down with John Roese, President and CTO of Products and Operations at Dell Technologies, to figure it out. From their discussion, we learn more about how IT infrastructure is being distributed, how data pipelines are being reconstructed, and how Dell Technologies is poised to be the only company to deliver a multi-cloud system.

Read more about the nine critical technology inflections for enterprises to understand. Download the white paper.

Guest List

  • Bill Pfeifer is a Dell Technologies messaging creator, focusing on the emerging tech landscape and striving to ensure everyone is ready and excited for what’s to come.
  • John Roese is President and Chief Technology Officer of Products and Operations at Dell Technologies and is responsible for establishing the company’s future-looking technology strategy and fostering innovation to make sure Dell Technologies is at the forefront of the industry.

Bill Pfeifer:                         Hello and welcome to The Next Horizon, a Dell Technologies podcast. I’m Bill Pfeifer, and together we’ll be talking about emerging technologies, their potential to impact society, and what you need to know today. Let’s start by getting to know one another a little bit.

Bill Pfeifer:                         I’ve been kicking around the technology industry for a little over 20 years now, and I have the gray hairs to prove it. I was a computer guy, a network guy, a security guy. I sold to customers, trained partners, trained SEs. I’ve been local, regional and global in scope. I’ve run technical enablement, created global strategic plans, and designed and built an entire certification program, not by myself, of course. Way smarter people than me did the real work on that one. In short, I’m a technologist and have been for quite a while now.

Bill Pfeifer:                         In my time here at Dell Technologies, I’ve bumped into some fascinating and brilliant people who work in our office of the CTO, which we internally refer to as OCTO, so you’ll probably hear us use that acronym here and there in these conversations. Now, some of these folks have had amazing things to say about the work they’re doing on technologies that haven’t made it out into the mainstream market, and it’s really changed how I think about what’s coming in a few different areas.

Bill Pfeifer:                         This is the stuff they’re working on all day, every day, and they’re thinking way beyond the typical industry conversations that I hear, so, catching up with the OCTO team always involves a new way to think about something, but, of course, dear listeners, the opportunities to have those conversations can be few and far between, so we set up this series so that we can take part in a shared conversation about what’s coming and how we can start to get ready for it, how we can start to think about it in a meaningful and accurate way so that, as we’re planning the next round of technology designs, policies, processes, we’re ahead of that technology adoption curve rather than just keeping up with what’s trending in mainstream industry conversations at the moment.

Bill Pfeifer:                         I told you I was a technologist, but I’ve also always been a storyteller, and I hope you enjoy the stories that we have for you. For the inaugural session. I have the rare treat of sitting down with John Roese, who is the president and chief technology officer of the product and operations team in Dell Technologies. John spends a lot of his time thinking about, planning for and evangelizing that space where emerging technologies live, so I’m very excited to introduce him and give you the opportunity to hear from him directly.

Bill Pfeifer:                         John, we’re really excited to be kicking off this podcast series with you as our inaugural guest. Let’s jump straight into the technology. There are lots of changes happening in enterprise IT environments. What’s driving that change?

John Roese:                        The biggest change is just a fundamental shift, some big fundamental shifts at the top level of how we think about infrastructure and technology and IT. The biggest is, for the last 10 years, we’ve been thinking that IT is on a centralization journey, that we have been essentially moving everything to some big virtual cloud somewhere else, and maybe 10 years ago and even five years ago, your workloads could do that. Doing email in a factory doesn’t make a lot of sense, so let’s just put that somewhere central, and even ERP and CRM could do that, but, about two ago, people started to realize that their future world of data and interaction with technology wasn’t actually happening in the data center.

John Roese:                        In fact, almost none of the data they were using and almost none of the actions they were taking were actually happening in the data centers they looked forward. They were happening in smart factories, smart hospitals, smart cities. They were happening out in the real world, and so, very subtly, and I don’t think people even understand this happened, we pivoted from an era that was all about centralization to an era where the enterprise now has to think about putting their IT in the right place, in the right way, in the right time to interact with the real world in real time, and that shift back to a distributed topology is significant because it forces you to rethink where you’ll do IT, how you’ll deliver the technology, how you’ll solve the problem and where the functions will occur.

John Roese:                        There are many things going on, but the fact that, if you’ve been doing this for a long enough time, you know that we go through these phases, we are now entering a phase of distributed IT as opposed to centralized IT, and if you’ve been in the industry long enough, you might remember that we’ve done this a few times before about every decade, and it subtly happened about two years ago, and now it’s starting to become real, and if people don’t understand that and recognize that pivot occurred, they may choose the wrong technology. They may choose the wrong cloud strategy. They may choose the wrong data strategy, and that could be catastrophic, so you have this big seed shift going on, and that should cascade throughout the entire technology ecosystem because, when you enter a new era, suddenly the world is different, and you have to make different decisions.

Bill Pfeifer:                         As we switch back to that distributed model in that ebb and flow of things as they happen, that takes us out to the edge. We’re hearing a lot about the edge, but it tends to be pretty poorly defined. What’s your view on the edge?

John Roese:                        Yeah, I mean, edge is one of those great buzzwords that means something and nothing to everybody, but it’s actually pretty simple. Back to this idea that the data is being produced and the actions are being taken not in data centers, but actually in your automotive environment, in your factory environment, in your healthcare environment, in your home, it tells us that we’re going to have to have an IT footprint out there, and, remember, for the last 10 years, we’ve been pulling all the IT out of the real world. We’ve been taking IT out of factories and out of office buildings and out of branch offices, and that was good. We could do that, but we want to put it back.

John Roese:                        Edge is interesting because you’re not putting it back the same way. You’re putting it back for a very deliberate purpose, and our theory is that there are really two major reasons why you reconstitute an edge, why you put IT back out in a cell site or back out into a factory. The first is that you have a workload or an activity that must be done in real time, and, by definition, real time, you can pick whatever the time value is, but let’s say hypothetically you have 30 milliseconds between the time the data arrives and you have to get the insight and take an action, and that could be steering a car, that could be optimizing a factory production line, it could be reacting to a medical event, but if it’s real time, then, fundamentally, you can’t do that if you have to cross the Internet. It’s just too long to get to the destination to do the processing and get the return path, so any real time functions that are materializing are very likely to be pushed closer to where the activity is just because things like speed of light are a constant, and there are certain limitations of distance and time and, in order to achieve real time, we have to basically push certain IT capability out to this new edge.

John Roese:                        The second reason though has nothing to do with time. Everybody thinks that edge is just real time, but there’s another problem, and that is we are very, maybe unexpectedly, but, surely obvious, we are creating an ability to produce data at a level that we’ve never done before. We are connecting billions and trillions of things potentially, and every one of them is producing data streams. We are no longer buffering them by having slow access links. We now have 10 gigabit access to homes, a hundred gigabit access to enterprises, potentially with 5G, 10 gigabit access over a wireless network, and we’re doing that at national and global scale.

John Roese:                        Just in the US alone, we’re going to have probably at least, 250,000 cell sites, and every one of them is now a 10-gig system, not a hundred-meg system, and when that occurs, the amount of data that’s being created, if we didn’t have a place to pre-process it and to reduce it and to manage it before we hit the data centers and the Internet, we’ll simply overrun the Internet, and so the second mission of edge is, yep, got to do the realtime stuff, but also the first line of data processing to basically control the flow of data as it moves through the cloud, and so these two reasons, and so if you had to develop an edge strategy and you’re wondering what should I put there, a good starting point is the function real time and is the activity involved in the creation of a massive set of data that, if you tried to move it across the Internet to actually process it, you’d either overwhelm the Internet or the cost would be too high, and if either of those conditions are true, and probably both of them are true in many cases, then do that at the edge. Build an edge. Do those types of workloads. Other things probably can be done somewhere else.

Bill Pfeifer:                         Makes a lot of sense, so you mentioned the data. We’re moving the data in ways that we haven’t before. That’s presumably going to create some data management challenges.

John Roese:                        Just a few.

Bill Pfeifer:                         Just a few. What do you see emerging in that space, and how do we start to address that?

John Roese:                        Yeah, the biggest thing about data management is, again, a buzzword that is confusing to people is we’re seeing a transformation about how data exists, how it relates to us as industries. The current data pipeline that pretty much every enterprise in the world has is largely a non-real time, fairly aggregated data pipeline. There’s a lot of data, but it sits in databases. It might even in fact go into data warehouses and data lakes and it has ETL functions to move it around, but it’s largely constrained.

John Roese:                        I mean, I used to be the CIO of a company, and I had this map of my data ecosystem. Everybody thought it was terrifying because, you’ll see, every CIO has this, so, “Here’s all my databases and here’s how they talk to each other,” and it looks like a fractal. The reality is even though we thought that was overwhelming, compared to what’s coming at us, that’s manageable because you can actually see centers of gravity for your data. You can decide that maybe I’ll move everything into a data warehouse and process it there, and that’s a perfectly good way to think about data for the non-real time world because the data isn’t really in motion. You create a database. You shove it into a data warehouse and then, at some time in the future, you run analytics around it, and you get a dashboard that tells you how your business is doing. That’s the status quo. That’s what we build today.

John Roese:                        Fast forward, I don’t know, tomorrow and what we’re transitioning into is an environment in which the sources of data are changing. You are now not taking systems of record and aggregating them. You are providing live telemetry and insight and events coming from potentially millions of connected devices, smart things, smart environments, real-time constraints. You’re then putting it not into a data warehouse or a database, but into a data pipeline, and the data pipeline’s job is to move it from those points of origin through the process of organizing, abstracting, enriching and understanding the data.

John Roese:                        Ultimately, the destination of it where its real process is very likely to not be a human being looking at a dashboard, but an AI making a decision, and, more importantly, back to that discussion of edge, many of these tasks have to happen in real time, that the time to move through that data pipeline for data that was just created all the way to insight isn’t 12 days to run the report, it’s 30 milliseconds to get an answer, and so, architecturally, the data pipelines and data structures we’re going to have to build as we go forward don’t look anything like the data pipelines we have today. They are designed to take insight and intelligence and data being created in our real world environments and, ultimately, deliver them to AIs and to machine intelligence that can make decisions and then provide the proper response in the proper timeframe.

John Roese:                        In order to do that, we have to develop a whole new set of tools. We’re using graph databases. We’re using new open-source technologies like Kafka and things that deal with time sequence data. We’re plugging into frameworks like TensorFlow and Caffe and other architectures that are allowing us to develop AI tools, and we’re doing this not in one place like a data warehouse, because you can’t possibly put all this data in one place, you’re doing it as a virtual distributed system that’s covering a multi-cloud environment, so, if that sounds different, that’s because it is, and in the data management world, what we have to recognize is our current data management pipeline and data management structure perfectly suitable, but becomes a point and a subset, a source of information for a much more distributed multi-cloud, distributed real-time, AI-driven data pipeline, and, by the way, if you don’t have one of those yet, that’s because no one has one.

John Roese:                        The entire industry is now re-vectoring to figure out what it means to build that, but I guarantee you that there will be no connected cars, there will be no smart factories, there will be no real-time world without us building those data pipelines because that’s actually how we connect the source of idea and activity to the source of intelligence to the source of action, and so it’s arguably the biggest activity of our industry for the next decade is building the technologies, [inaudible 00:11:36] a new data pipeline and making this thing operational. Otherwise, what we have is a set of captive data that can’t be acted on in real time, and that does not get us to any outcome that we’re trying to achieve.

Bill Pfeifer:                         With the move to edge and distributing all this stuff and the new data management challenges that we have, we have a new take on old conversation as well, which is security. We need to revisit what it means to secure such a distributed environment, so much data, so much data movement. What do we have coming in that space?

John Roese:                        I mean, here’s the big idea with security. It’s not about a new feature, a new product. There is no market for creating the data management security appliance. That’s not going to be the solution. The reality is we have to start thinking entirely differently about security. We have to think of it as not a set of discrete problems and discrete products that solve those problems even if there are new problems. We have to ask and understand how do we create security as an integral component of the data pipeline, of the cloud infrastructure? How do we connect it to the spines and the backbones and the frameworks that are already coming into existence?

John Roese:                        A good example is, if you build a security product and you try to place it in your infrastructure, in the old world, you could do that because there were physical places you could put it like a DMZ or at the end of the VPN tunnel. In the new world, everything is virtualized, everything is dynamic, everything is across multiple clouds, so there aren’t physical places to put security products. How do you solve that? You make them logical products. You make them microservices. You make them software, and then you deploy them when you deploy the SDN service chain or when you create the virtual abstraction of the end-to-end user experience, and so what really is the biggest shift in security to play in this new world is to rethink what is security. It’s not a product anymore. It’s a feature. It’s a capability, and how is it linked to the rest of the world? It’s not being done in an isolated silo independent of the rest of the virtualized cloud world. It’s actually being driven by the virtual cloud infrastructure that is now being defined by things like SDNs.

John Roese:                        That pivot sounds intuitive, but that’s not how the security world works. That’s not how it was built. That’s not how it’s delivered. That’s not how any of our products in the industry level are typically deployed, so it is a pretty significant seed change. If we don’t do that though, there is no chance of building a coherent security model across this next-generation data pipeline over the multi-cloud, distributed real-time world.

Bill Pfeifer:                         You mentioned the coherent security model that spans multiple segments, multiple clouds, and we talked about edges, and you mentioned multiple clouds a couple different times, so can you talk a little bit about the Dell Technologies’ vision for how to handle all of this distribution, the multiple clouds, data centers behaving like clouds, clouds behaving like data centers, edges popping up all over the place?

John Roese:                        Yeah. Yeah, I mean this is the core of what Dell Technologies is doing. I mean, it may not be entirely obvious, but we do have a strategy. Our strategy is to actually turn the massive distribution of multiple clouds into a coherent system that enterprises can do stuff over without feeling like they’re dealing with 38 independent silos, and, in order to do that, we have a vision that we’ve been executing on for quite some time, and it basically says, “Start with this idea that you can’t unify the multi-cloud world if you don’t participate with those clouds, if you’re not in them,” and so, for the last two decades, we have been deploying our technology as a core component or an element in the data plane of clouds around the world.

John Roese:                        Today, we are in pretty much every data center in the universe for an enterprise. We are in all of the major hyperscale clouds now. We are in 4,500 service provider clouds, and that doesn’t mean we’re just observing them, we’re actually inside of them. Our run-times, our execution environment, our ability to manage our components in those systems, we like to describe those as anchors. If you’re trying to unify the multiple clouds, you ought to, I don’t know, put some anchors out there to start corralling them. Our technology is that anchor because it’s present in all of them.

John Roese:                        The second element though is it’s not good enough to just have anchors. You’ve got to attach the anchors to something. You’ve got to organize them, so the second dimension is we’ve realized that the VMware suite of orchestration tools and their vision of, hey, if I have cloud technology in multiple places, wouldn’t it be nice if I had a single operational hub where those various anchors started to talk to each other, and so things like vRealize automation, vRealize orchestration, the next generations of those are giving us an operational hub that says, hey, now you can start to not just observe, but control the technology living in a public cloud and the technology living in your data center and the technology living in your edge clouds because there’s an anchor and there’s a common operational hub.

John Roese:                        Now, all of that, unfortunately, just sounds like plumbing. That doesn’t solve the business problem yet, but if you don’t do that, you don’t have the platform to do the really cool stuff. The cool stuff is now starting to happen because we have those anchors in that operational hub, and that is we can now start to solve horizontal end-to-end problems in the multi-cloud world. In fact, today, we can have an end-to-end networking experience across the multi-cloud world. That sounds like an obvious thing, but we’re the only ones that can actually do that right now, that you can use NSX and our SD-WAN technology. Our application delivery controllers are all software-defined, virtualized and working as a system to literally having network fabric that spans multiple public clouds, your private environment, your edge environment, even your service provider environment across the WAN, across the LAN and the data center.

John Roese:                        If you can do that, now you have something very valuable. You have a platform for connectivity, but then take it a little further. A little while ago, we announced the Project Pacific, which is to take Kubernetes and to make it span the multi-cloud world. Imagine if your Kubernetes clusters were not operating differently, if they were in a public cloud or if they were in your data center or if they were in the edge, but, instead, when you deploy a containerized workload, it can be managed and deployed anywhere in your multi-cloud environment as one coherent system. Go even further, we’re already doing things like vMotion across clouds. It sounds like a simple concept. It is nontrivial. It requires the anchors. It requires the operational hub, but now, as a service, it allows you to have this agile, elastic, flexible experience to consuming VMs.

John Roese:                        You go up a little bit higher, and there’s other things. For instance, our application and developer experience, things like Cloud Foundry and Bitnami aren’t designed to run in only one cloud. They’re designed to give application developers a platform to experience the multi-cloud world without having to be intimately aware of it. Whether it’s the repositories or the application deployment capabilities, those now start to become characteristics of this experience, and then even if you go up into management plane, if you take something like Wavefront, which gives you telemetry, it’s not telemetry for your private cloud or your public cloud or your edge cloud, it’s telemetry for your multi-cloud because it’s part of the system, and then even things like CloudHealth, which initially would give you a view of a single cloud’s economic experience, now can give you the economic visibility about where is the most cost-effective place to run that workload in the multi-cloud environment.

John Roese:                        We’re just beginning, but you can’t even have that third set of discussions, the horizontals that make the multi-cloud system useful unless you have those anchors, unless your technology is pervasive, unless you have a common operational hub, so the security discussion we just had can’t be solved in a vacuum, but it can be easily solved if it’s just another layer, a horizontal layer of that multi-cloud experience that we’re attempting to make real in the industry.

John Roese:                        By the way, we’re not the only ones with this vision. That’s the vision the customers have. That’s the vision people talk about. We just happen to be fortunate because we have by far the most pervasive presence in clouds around the world that we get a good starting point. We’re already there and the anchors already exist and the operational hub is already real, and now we’re just making it work maybe more effectively across this new multi-cloud environment, but now our focus is on let’s turn those horizontals on, let’s make networking pervasive, application delivery pervasive, developer experience pervasive. User experience, management experience, economic experience, that’s what you get from a multi-cloud system, and if you ever wondered what we’re up to, it’s let’s make that real for our customers so that, on top of it, they can run businesses over the multi-cloud system.

Bill Pfeifer:                         The ending there on the customers, all these changes are… They keep accelerating. We’re starting to go exponential with the rate of change here, but, unfortunately, our customers’ IT budgets don’t seem to be increasing exponentially, so they have to make some pretty smart decisions. They have to be able to look forward a little bit, plan now for where they’re headed. If you were speaking directly to our customers, what would you suggest their priorities are for getting ready for the next, say, three to five years and all this change?

John Roese:                        The first is you have a finite amount of resources, and it’s not even the dollars. It’s the people. You have X number of people that can do work. Ask what that work should be, and the reality is we know you’re going to have to build a multi-cloud system. You’re not going to be able to do anything in a single public cloud or just do it all on-prem or just do it all at the edge, so it’s an inevitability that you’ll have to have a multi-cloud system, but you will also have to build AI algorithms. You will have to develop new user experience and new software, and so our message is, look, all of this has to be done. You don’t get to escape any of it. You have to develop and deliver and have a multi-cloud environment. You have to be able to build your applications. You have to be able to manage your data. You have to have business outcomes, so think about where you want to spend your time.

John Roese:                        The reality of it is, in the past, when it came to things like infrastructure, a lot of people felt like, “I’ll spend my time buying and developing and building all the piece parts and, ultimately, I’ll put it together, and maybe I’ll save a little bit of money by owning the components and doing the integration.” Today, you can’t afford to do that. What you need do is be able to consume a multi-cloud system.

John Roese:                        Our vision and what we’re delivering is, and what my goal or vision was for creating Dell Technologies is wouldn’t it be better if all of the best technologies in the world were all in one place? What if you could deal with one vendor that could actually do this for you and, in many cases, do it as a service for you, and then you could, one, get access to the best technology, solve that lower level problem of building the platform and the engine and making that available, but then you could shift the time and energy of your people to doing the things that are more directly connected to your business like developing the AI algorithms and building the better user experience and transforming your digital business up and application transformation or business transformation, and so our message is really simple. All the technical work still has to be done. There is no easy button. You can’t just say a public cloud will do this for you.

John Roese:                        That’s not the way it works these days, but there is an easier path, and that is to, quite frankly, buy into infrastructure architecture, buy into the system design and work with the vendors and the partners that can actually make that real for you and offload much of the integration and maybe even much of the operational tasks so that your people can spend their time on the most high-value activities, which could be low-level functions like developing AI algorithms or they could be high-level things like retraining your workforce or engaging with your customers on a new level, but this is really complex and there is no company in the world that can do it all by themselves, so choosing your partners carefully, and in the case of infrastructure, I am biased, but choosing to work with the largest infrastructure company in the world that has the only capability to actually deliver multi-cloud as a system seems like a pretty logical starting point.

John Roese:                        That’s the invitation to our customers. That’s why people are working with us. That’s why we’ve grown our business almost $20 billion since we put Dell and EMC together, and we think we’re in the right ZIP code doing the right thing, and we’re doing it not just for our benefit. It’s because it accelerates our customers.

Bill Pfeifer:                         Makes good sense. Thanks so much for your time, John. We’ll continue to explore a bunch of these topics as we expand this podcast series, and we look forward to hearing more from you as things continue to develop with Dell Technologies’ office of the CTO.

Bill Pfeifer:                         For those of you who enjoyed this podcast, and, really, who wouldn’t love it, John is always a great speaker, we have a white paper that you can review if you want a bit more detail on John’s answers that we covered in our conversation. You can find it at www.delltechnologies.com/nexthorizon, along with future podcasts and some other great content from the smart folks at our office of the CTO, and, to our listeners, thank you all for listening to The Next Horizon, a Dell Technologies podcast.

Bill Pfeifer:                         We appreciate your time, interest and attention. I hope you’re as excited as we are about the great innovations that are coming out of the Dell Technologies’ office of the CTO, and I hope you’ll join us for the next episode where we’ll be talking with David Frattura about the components of digital transformation.

Bill Pfeifer:                         Be sure to subscribe to the podcast either through your favorite podcast app or through the website at www.delltechnologies.com/nexthorizon so you don’t miss any great new content, and I look forward to seeing you again for upcoming episodes. I’m Bill Pfeifer, and this is The Next Horizon.