Todays Leading CIO minds talks with host Steve Ginsberg
Mark has held IT executive and CIO roles at Ericsson, Apcera, HP, Gilead, VMware, and Switch SUPERNAP. Mark Thiele’s successful career in IT spans 30 years and has focused on both operating roles and on driving IT innovation, efficiency and cloud adoption across enterprises of all sizes. Mark speaks globally at leading industry events on topics like cloud adoption, data center, IT organization, Edge Computing and more.
Mark has deep industry experience and extensive knowledge of the requirements for building IT solutions that optimize for value and speed, while enabling innovation. In his current role at Ericsson he is providing strategy and engineering for edge cloud in the network and telco vertical.
He is the president and founder of Data Center Pulse, an organization created to promote best practices in the data center industry. He has also contributed to nonprofit industry groups, including the Chair of the IDCA Technical Committee, the Cloud Native Compute Foundation, The Green Grid and Infrastructure 2.0.
Steve Ginsberg: Hi I'm Steve Ginsberg. My guest today is Mark Thiele, an expert in enterprise data center strategy and operations, cloud and edge computing. In this episode we'll get Mark's take on what edge really means and how enterprises should be viewing both the challenge and the opportunity. Mark, thanks for joining us.
Mark Thiele: Thanks for having me. Appreciate it Steve.
So we've talked a lot about the multiple definitions of ‘edge.’ How do you look at the terminology?
Yeah, I think the simple reality around edge is that it means all things to all people and rightfully so. Right? So if I'm the data center community, edge to me is: what kind of data center do I build and how close can I get that data center to more people from a latency standpoint? Or the right kind of buying audience from a latency standpoint. From a from a user of the edge... so the complete other side of that coin is: am I using the edge because I need... or am I benefiting from the edge because I'm using some content that's been staged there? Or am I benefiting from the edge because I'm able to do some quick analytics off real time data coming off of some drones or coming off of facial recognition or something like that?
And then within that, the gamut of opportunity of what the edge could be, runs any number of a hundred different scenarios. And so the way I like to picture or frame the edge environment is: that it's in effect, cost effective ability to deliver high quality experience for the customer wherever the customer happens to be, right? So it could be that the customer is 50 milliseconds away and for that solution, that's the edge. That's fine. But for solutions that benefit from having low latency, ten milliseconds, five milliseconds, whatever it happens to be, then that's potentially the edge. But what is successful is what you can provision that supports that solution cost effectively. In other words it has a return on investment.
And with that you mentioned drones for example. What applications do you think are really moving to edge locations today?
Yeah it's interesting, there are a wide range of applications that I've seen people deliver applications on, basically application distribution networks where they're putting applications in containers and distributing them out based on a pre-set determination of latency and failover capability etc. Those applications could be almost anything that would benefit from lower latency or some form of data sovereignty or privacy or something like that. And then I've seen... I mentioned drones are one example that I actually worked directly with. Somebody was trying to run drones over, or wanted the ability to run drones over almost anywhere in the country. But they wanted to extend the distance and coverage and provide for better real time analytics of what was seen and how it was interpreted, in order to make patterns more efficient and identification more quickly of subject matter, whether that subject was a stolen vehicle or a loose animal or a runaway child, whatever it happens to be.
And so the combination of the edge network along with localized ability to store and process data and provide feedback from video, allowed for people to coordinate drones over long distances while maximizing their patterns for search and not guaranteeing, but virtually guaranteeing that they didn't show replicated data or mismatched data, in other words, identify things that in fact weren't being searched for.
Well that's really exciting to be able to use it to assist in emergency situations. I'm wondering in that situation, it sounds like there would be centralized data in the cloud and then localized data where needed. Is that the right way to think about it?
Yeah, I mean I think one of the hardest problems for the edge (and you and I might have even had this conversation before) but I think one of the hardest problems for the edge will be how to take the data and how to keep it, and how much of it to use in real time locally versus how much of it to send upstream, and whether that should be batched or sent continuously. I think over time we'll get smarter at it. Either that or we'll just get bigger pipes and deal with it the way we've always done managing data, which is just get bigger hard drives and more space and keep it all.
But realistically there is a huge opportunity still from a problem solving standpoint to help customers that are attempting to deploy at the edge, govern how they keep and identify data that's created at the edge.
How do you think about the unevenness of the likely rollout of edge? What I'm alluding to is, obviously in cities it's likely, as we have seen in the past, and in certain areas in the country like Ashburn, Virginia, you're going to continue to see more data center and more data center locations, even smaller edge locations. 5G I think it will be a part of that. But in America, for example, in the middle of the country it's not that likely I think that you're going to end up with edge data centers every five miles for example. Will applications be as useful given the unevenness of edge?
Yeah. So there's two different ways to look at that unevenness, Steve, and I'll talk about that, but I'll respond to the question as you answered it first, and then maybe we can expand on that a little bit. From the perspective that you've pointed to, I would say it'll follow a natural progression that most IT solutions do, that are dependent on volume of users in order to provide justification for use. And where there's the most demand and the most existing infrastructure, there's the higher opportunity for return.
What's likely to happen, and so in other words, bigger cities, urban areas are more likely to benefit from more edge oriented services more quickly. But that being said, as services become routinely deployed in cities and as edge technologies become, for lack of a better description, more routine in how they're defined and how they're deployed, then more of those solutions will become justified in smaller communities. Their scale drives down price, the known value of opportunity per person becomes more real, because they can trend the data. And those things mean that over time, more of them will be extended out to more rural communities, but that being said, unfortunately there's no simple answer.
I was just at my brother and sister in law's place in a small town of Roundup, Montana. And they actually live about 14 miles outside of the town of Roundup. And you know I had zero cell phone coverage while I was there, and the best Internet they can get is ISDN. So you know unfortunately, as good as we're getting with the distribution of technology there will continue to be the ‘haves,’ the have nots’ and the ‘have laters.’
Yeah. How do you advise that enterprises go about architecting for edge? Obviously you mentioned access time as a key concern and you alluded to containerization as well. How do you look at the whole picture?
Yeah I think from a from a traditional edge standpoint, I mean the way again, every time we talk about edge, one or more of us are going to make an assumption about who's listening and why what we're saying is appropriate at the time. And so when we talk about edge, the way we've been talking about it, we're talking more of general access to a broad audience, generic audience or end user type audience.
But for the enterprise, edge could also be providing more services for a campus right, per office building, per location of customers, in this case customers being employees. And so there are an enormous number of additional opportunities to do that now that don't require in many cases fancy technology like deploying 5G or something like that, but are effectively prepackaged extensions of modern day cloud environments.
There are opportunities coming from each of the three big cloud providers in North America. There are independent ones that are able to offer a cloud stack that potentially even abstract some of the centralized cloud services so that you can run those in your individual data centers, and in effect run them on existing hardware or in other cases, like with Outpost from Amazon, in theory when it becomes available that might be a prepackaged stack, but there will be a lot more opportunities for that kind of edge in the enterprise.
Now over and above that though if my assumptions about the direction of technology use and the underlying focus of what edge is sort of a symptom of, it's a symptom of the opportunity and demands behind business or digital transformation. And the real demand from my perspective behind digital transformation, business transformation beyond some basic benefits of efficiency and speed of delivery and things like that, maybe supply chain improvement etc., integration with partners better is really the theme of a better customer experience and more customer intimacy.
I could be wrong, but I think it just seems pretty obvious that if you can deliver more capability into the hands of your customer, whether you sell paper towels for a living or sell cell phones for a living or something in between or even all of the above like Amazon, then your ability to provide the customer with a better interaction and to gain greater intimacy with them through that interaction, I believe, has got to be the end result for almost every company as they struggle with becoming technology companies and attempting to exploit that to get a better connection with their customer.
Yeah, agreeing with that focus, a lot of the way I tend to think about it is being a bit analogous to when we saw the move to Google maps with Ajax technology. More and more was just happening at the user level. It was happening in the browser versus all these calls having to go back to farms to provide that data.
I'm wondering if you would like to tell us a little bit about some of the companies that you work with as an advisor or also as an employee?
Well you know it's interesting, there's a few of the companies that I work with that, both on purpose and by accident, are potentially very interesting for edge. I work a lot as an advisor with a company called RackN, [whose] CEO and founder is Rob Hirschfeld. And for an organization that might want to continue to maintain its own infrastructure, or even some of its own infrastructure, and to be able to exploit other provider resources as necessary, or take best advantage of the heterogeneous resources they own internally, its software capability allows for that re-provisioning and patching and automated management of infrastructure environments, deployment of new capacity, etc. that provides better security, provides more rapid rollout, more consistency, thereby more resiliency etc.
So to me you could argue that that's good anywhere. It's good in hyperscale, it's good in a restaurant that's got a rack in the back corner. But when you think about the deployment (and some companies will do this) of infrastructure in many locations to try to attract access to end customers, then having improved remote capabilities that allow you to stay fresh, without taking two or three weeks just to get operating systems patched against a new security threat, that in and of itself, is a huge opportunity. That's one reason why I work with RackN and why I like them.
I've been spending a lot of time with a company called Solecular. And these guys are doing some fantastic work in helping to provide AI as a means to get more value out of the capital expenditures and including the Op expenditures that you have in your data center. So the ability to use power in a rack more effectively and use your space more effectively, potentially using fewer servers and less power. Even at a minimum, it's the potential opportunity to delay the expense and the trauma of expanding a data center or building a new one. And at a maximum, it's saving both time and money in any or all of your locations in a real way.
And then another one I'm looking at [is] a company called AdeptDC is using some AI technologies to help find root cause in failure environments etc. So all of these technologies and more to me are indicative of what we just kind of have to do if we're going to be able to effectively exploit an edge environment, because the one thing that I believe... and I think, Steve, you and I even talked about this: is that what I have to believe to be true is that we cannot assume that anything we're doing the way we're doing [it] today, is extensible to the edge, right? So application design, the use of containers, the use of functions as a service, the use of automation, the use of automated remote access, the ability for self healing, the ability to fail in place all of these things times two, times ten, however you want to say it, are going to be required in spades when we're talking about edge deployments.
Yeah I very much agree. And I think you and I see this in very similar terms, which is: today there are multiple edges and there will continue to be. It's in some ways just about how close do you move to your customers. You pointed out well that distinction could be the businesses customers or internal customers like employees or contractors or other kind of factory locations for example.
But in its larger implication, the edge means moving from maybe for some companies, a few data centers if even that, to hundreds of data center locations or even thousands or hundreds of thousands of locations. Therefore this whole wave of: how do I remotely manage and automate basically everything becomes fairly critical.
Yeah I mean Steve, if you don't mind me expanding on it for just a little bit longer, I just I'd like to leave whoever is listening to this and believes this is of any value at all, a point that I think is really urgently required, and I don't see a lot of people thinking this way, is that a lot of us continue to look at what the edge will become, based on what we assume we can extend from what we do today. And so to explain that a little bit more, I talked a little bit about that in my previous points. But the assumption most people have is just at the highest level: “Oh it's you know, it's too expensive to do it that way, it's much more cost effective to just deploy it in the public cloud.” And, okay, under what circumstances?
Well if I have to deploy a thousand instances in edge data centers and I've got to go to 100 cities and deploy 10 instances in every city and I've got to acquire a server and have support and contract with a co-location company or whatever it is, then gonna be 10 times as expensive to run those as it would to put all that in two public cloud locations. Ok?
First problem is if you needed that, then the two public cloud locations won't solve your problem. But second of all, if you continue to assume that the paradigm that you've used to solve the problem in a centralized way is how you will extend that to solve it at the edge, then yes you're right, it won't work. It will be too cost ineffective, and so you have to think about the problem in another way [which] is: where can you borrow? Where can you share? Where can you lease where you would have bought? Where can you multi tenant with partners or friends? Where could you look at more automation? Where could you look at different CPU sets to program against? The list goes on and on.
And it's I think those companies that can exploit the capabilities of somebody in some of the companies we've already talked about and others in the market and can get ingenuous about looking at how they would own and operate their own infrastructure in new ways, as opposed to just throwing more people and hardware at the problem, I think have a real opportunity to position themselves out in front of their competition.
Yeah absolutely. I think for companies who want to be brave and smart about it there'll be lots of opportunities, and for companies who are more conservative about it, I think you’ve talked to the folks a Packet, for example and others, that there'll be virtual platforms that will do a lot of this for enterprises who won't consider this their specialty.
Yeah. Yeah absolutely.
How do you see 5G fitting into all this?
Yeah I mean 5G is an enormous enabler, right? There are a significant number of technologies that are available today to distribute edge capability. Hell, like one of the companies I was talking about is a company called Rafael Systems that can distribute applications based on a latency policy to multiple locations around the world via containers. So you code into one of their containers and it's Docker compliant, and you send it out. And swim.ai, which allows for using data right at the edge. Even though they can collect terabytes in a day, they only use and keep megabytes of that data as they manage trend analysis that can plan for five minutes ahead on the frequency or the speed of change and timing of change of streetlights and stoplights and intersections across the city, potentially speeding access for companies like Uber, Lyft, to get across the city, or FedEx or U.P.S. or something like that.
So a lot of those technologies are available that don't necessarily require 5G. But I do believe that there are an enormous number of opportunities to extend some really, really interesting user experience options to the edge, and user experience sounds like it's all fluff, but it's not, right? I mean there was a recent story about a doctor using 5G specifically to do a surgery from 3,000 kilometers away. And using any other existing connectivity method likely would not have worked in that scenario.
But there are potentially tens of thousands of... everybody talks about automated vehicles or fleets, the amount of data that will be collected in smart cities and will need to be shipped and managed more effectively. I didn't talk to them directly, [but] I was talking to the team that was working on a stadium, looking at how 5G could solve for providing real time video on all sides of the stadium that didn't show delay between when a footballer kicked the ball and when the action was displayed on any or all of the screens.
And then you know you think about that same stadium becoming a smart stadium, and potentially having 500,000 to a million sensors, IoT devices just in that stadium, on top of having 50 or 60 thousand customers in the stadium at any one point in time that are all using some form of data, whether it's their Fitbit or their watch or their phone or some combination of the above. And you get to realize that our existing environments of LTE which have already failed in those situations, would be so overburdened as to become completely worthless.
I've had some good conversations about the stadium and I think it's thought provoking in and of itself, and I think it gave me maybe a little insight of at least a profile which, as you mentioned, [is] obviously high demand and sort of multiple access, multiple technologies wanting to... There's a need to use multiple technologies in the same place.
And I guess the other thing I would say is it’s a highly financed environment, so it's an area where they can invest and it's essentially set up a data center at the stadium where multiple carriers, the telcos, will want to be there: the AT&T's, Verizon, Sprints of the world will want to be installed there to provide... and then obviously there's enough kind of financial interest from the customers and the sponsors to get applications written for that high density environment.
That's right. That's right.
You mentioned smart cities. I'm curious how you see that playing out from what you've seen so far?
Yeah, it's kind of hard to say, but smart cities have a significant number of opportunities of course, so everything from being able to manage traffic patterns better to parking, to informing people that are entering the cities of what's going on in what part of the city, whether the surf's up or the ethnic food party in one part of town is happening and whether it's crowded or not. So you know there's any one of a million different opportunities for exploiting a smart city.
But I think realistically it will start much more mundane. It'll start around things like more efficient delivery of services, more efficient management of things like water and even irrigation for keeping flower beds watered and things like that, to light management to intersection management etc. I think those are all likely first steps, and then as those become successful and well understood and hopefully secure, you begin to get better opportunities to integrate with more of the city's inhabitants and integrate with homes and integrate with offices and really kind of the sky's the limit. I mean to me frankly, it's both really interesting to think about what it might mean to be living in a city that you're literally connected to, and it's scary at the same time.
Yeah I agree and have had some similar conversations about one might not think of a smart parking lot as being an exciting edge opportunity to use 5G and that type of thing. But as you pointed out, it's a great starting place for a community to get an idea of what would it be like to make that digital and to get that integrated in our environment and then to move on to the next projects from there.
How about smart buildings? Have you seen anything there? That's one of these cases I thought was rather interesting, and like a lot of them it seems to be kind of evolution of things that were already in place. You've already got an air conditioning system with a timer. One could call that a smart building, but then things kind of progress from there.
Yeah I mean the smart buildings are really kind of fascinating and to be honest, I feel like even though I may know you know as much or more than the average audience member, I feel like I'm still a novice. But as I've looked at examples or dove in here and there to do a little bit of research, it's really fascinating how many opportunities each layer of intelligence, each layer of visibility potentially provides.
So how people move through the building could be important in a number of ways as one thing. How people move through the building could dictate planning for where elevators are. It could dictate for planning for how the air conditioning moves; it could dictate where staff members should be. It could actually be an indicator of where staff are wasting too much time because of their travel patterns or because they have to go to the same destination a lot but they don't take very much with them when they go (you know, something stupid, it could be anything), to tying that information to a security system, to providing a route, or not route but preventative analysis of equipment throughout the facilities, to have a better understanding of when an air conditioner might be in need of repair, to when an elevator might be starting to cause problems, to providing real time feedback from customers as they experience aspects of the building. I mean seriously Steve, the list just goes on and on. Frankly I think you and I won't recognize a smart building when we finally get into one, as a traditional building you know five or six years from now. I mean the opportunities are immense.
That's some thought provoking stuff about how it might evolve. With all this we've kind of covered a lot, and maybe as a final question to wrap it up: how would you advise enterprises to be planning for these changes?
Yeah I mean first of all, technology is not the answer. Technology might solve a problem for you or create an opportunity for you, but you have to define what the need is first, and we all love to chase the latest shiny thing. And I sympathize. I'm an IT guy. I love to get my hands on the biggest drives and the fastest servers and the biggest Fibre Channel array and putting in more virtualization and building cloud and data. I love all that stuff. No denying it, but edge is a perfect example where it would be really easy to overextend yourself if you don't have a really clear vision for what the ‘Why’ is for pursuing the activity in the first place.
So if you're not sure what Mark means when he says what the Why is, I could try to give you a definition, but I think I would butcher what Simon Sinek has done so well. I would urge you to go out on YouTube and take a look at Simon Sinek and look at how he defines identifying the Why for any project, and I think that that little step as it may seem, is vital, because if you have the right Why, the bottom line is, today in my mind, you can find the technology to solve the problem. So you don't need to spend a lot of time now going out and searching for technology and then attempting to create a problem to solve. Find the Why and then find the technology, it's there, and partners, they’re there.
About really connecting to the business value or the additional value of a project before you go down that path.
Great, well thank you very much Mark, I really appreciate the discussion and thank you for joining me.
Thank you Steve, I appreciate the invite, I hope to do it again soon.