Podcast #110: A Decade at the Edge

Good day, and welcome to episode 110 of our Digital Industry Leadership podcast, produced by, for, and about Digital Industry leaders. Today I’m pleased to feature someone who has literally been at the edge for a decade, Jason Shepherd, VP of Ecosystem at Edge Orchestration Company ZEDEDA. Prior to joining ZEDEDA, Jason was CTO for the Dell Technologies Edge & IoT Solutions Division. His proven track record as a thought leader in the market is evidenced through his leadership, building up the award-winning Dell IoT Solutions Partner Programme, and establishing the Vendor Neutral Open Source EdgeX Foundry Project to facilitate greater interoperability at the IoT Edge.

Jason is a board member for LF Edge and was recognized as one of the top 100 industrial IoT influencers, in both 2018 and 2019. He holds, and I’m not kidding, 14 granted and 25 pending US patents. So, Jason, welcome to our digital leadership podcast.

Thank you. Thanks Ken, thanks for having me.

Hey, look, anybody who’s got 14 granted and 25 pending US patents…! I always joke that some people might have seen you in the past as more of a marketing guy, because you was always present there at Dell, but man you’ve got some really strong let’s say technical foundations to you, and I hope to review those in the call today. So, let’s start with your professional journey, tell us a bit about your background and how it has informed your views of digital.

Yes, I’ve always been very curious about technology, naturally inquisitive, I always say that I probably annoyed my parents trying to press every button I could find. One day actually I had managed to open up the car door and fell out in an intersection as a kid, though only that once. But always playing with different things, I just like to build stuff, so I play music for fun, we built our house, I’d have some project going, and that’s led me from where I started as a mechanical engineer, and by just being inquisitive surrounding yourself with good people I’ve been able to morph over the years into more of a solution role, of course getting into IoT and then Edge.

So, I’m just really into the blend of form and function, I’m fascinated by the people part of technology, so in many ways the psychology behind how people interface with technology, good technology disappears into the background, and you just go live your life and make things better and whatnot. So, just that blend of people and technology is very interesting to me, so everything that I’ve tried to focus on over the years has been furthering my understanding of that element.

I love the fact you started as a mechanical engineer; I’ll call it truly down to the metal. Your move to embedded whilst not unprecedented is probably unique for mechanical engineers I must say, typically who would see electrical engineering background in that. What was your particular inspiration to move –  I’ll say, up the stack into embedded?

 It’s funny, first up the stack into embedded is a bit ironic, but yes, up the stack. So about 10 years ago I enjoyed doing what I was doing, I was doing mechanical engineering at Dell, leading a team, the precision workstation team, but in our free time I started leading the team in various different concept development efforts, and was for all the reasons I just mentioned just really interested in a variety of different solution aspects. So, we started going down that path, I always say the best way to get a job is to already be doing it, that’s what led me into leading a solution development team, like a concept development team, doing R&D across a variety of things there at Dell.

And so, the inspiration was that interest in broader solutions, how do you drive outcomes for end-users, for customers new experiences. And as I got more and more into solutions became more and more clear, and this is 2010-12 timeframe, that the whole mobile trend that’s evolved is going to turn into even more of an embedded trend. I started saying fixed is the new mobile, somewhere around that time, and you’re just going to see more compute everywhere, and of course that led into the IoT raise that started about 2014. But the inspiration really was just how do I take more of an outcome-oriented approach with the folks that I work with, versus a bunch of widgets. There’s a lot of solutions looking for problems out there, so I think it’s important to start with the, what are you doing, and why, and then back into the technology.

You were at Dell at a pretty eventful time as I remember, going from number 51 on the Fortune 500, up until Michael Dell’s leverage buy back at the company in 2013. Dell of course a world leader in personal computers would soon emerge as a key enterprise computing and edge player, and I’m sure partially due to your work. I noted several of your roles in concept and strategy in the office of the CTO around the same time, tell us a bit about your efforts, and as you said earlier some of the outcomes of those efforts.

Yeah there is that little EMC acquisition along the way. It was a great time to be there, it’s obviously a great organization, and I guess it was back around 2010 where I started running this concept development team, it had started more hardware centric as you could expect given that I was in the client computing group. So, we were exploring different form factors, getting into the whole tablet space, and different ways to engage with compute back then. But then I started to hire various different software developers, so we could get more into software and solutions, and that led around that timeframe to a bunch of work around session continuity, and this was early in its time back then, but how do you pick-up from one spot to the next when you’re working on say a PC and then a tablet, and then you go into your house and you look at your TV, so that’s turned into things like handoff, you see it fairly widely from a variety of players now.

Then we got into collaboration for a while, which is certainly very relevant these days with COVID, but a number of the patents that I got with the teams there at Dell around how do you make remote collaboration more effective. One of the biggest challenges for remote workers still, is being left out of the whiteboarding, and so there was a number of things that we did for whiteboarding. Stuff like if someone is drawing on a whiteboard digitally, remotely, block out that space where they’re drawing, just like you would be blocking out the whiteboard when you’re physically in the space in a room. Because a lot of times whiteboarding remotely degenerates into a bunch of people drawing over people’s stuff, writing smiley faces and whatever, but when you’re in a room you’ve got those physical boundaries.

We did things around when you get a whiteboard picture, if you weren’t there in the room and you see a bunch of scribbles, you have no clue what one of those scribbles… what was being talked about when that scribble was made, that random drawing and then people draw over it and it’s just a mess at the end of that collaboration session. So, if you clicked on that scribble digitally it would play the audio, and video if you have it, that was done right around that time that it was created. So did that stuff, we did stuff around when you’re working remote and you could have a whole list of contacts in your instant messaging system, or collaboration tool of choice, but you never actually ping somebody, but then the moment you see someone physically, you’re walking through the hallway, you go, “Hey, I’ve been meaning to ask you something for like 3 weeks”, so do you have this concept of walk-byes across your screen and people that sort of… again it’s that physical digital thing that will trigger you to ask questions.

We did all kinds of stuff like that. Now granted a lot of that stuff still is a challenge. We’ve done that, there are certain players that are building their collaboration tools, and just the basics still is a challenge, but it was fun to work on all these different concepts. Anyway, so we did all kinds of different things around that and built-up a patent portfolio and worked with a bunch of different folks. Then in 2014 it was like, “Hey, what do we want to do with this…”, the buzz term “IoT”, just then trying to turn the whole progression from there that we’ll get into more detail. Yeah it was a fun time, my initial days of the decade at the Edge was around how do we drive new experiences, and client.

We were both joking right before this, because we’re using Microsoft Teams for the recording of this, and I couldn’t find the record button, so still we have a lot of ways to go in terms of collaboration! And certainly, the idea of a segmented whiteboard and using that, still seems pretty far-fetched at this point, so you’re well-well ahead of your time my friend!

Your expectations are too high; you like to find… come on!

Exactly [laughter]. So, I believe we first met in 2015 as I remember, as we were pitching PLAT.ONE at the time, one of our portfolio companies which subsequently became part of SAP, its Leonardo now. I think you had just been promoted to Director of IT, IoT Strategy & Partnerships at Dell, what was your key focus here and subsequently as CTO of IoT and Edge computing for Dell?

The focus in that role was really building up our ecosystem. When we started with the concept, early-on we were like, “Hey guys, this is ultimately about software and services”, the whole IoT game really, I think digital in general. Infrastructure is super-important, but it really is about all of the value-add around the chain. That’s actually part of the way we got that initial stuff funded at Dell for the hardware infrastructure, for IoT and Edge was starting to build-up that initial partner ecosystem. So, I would with my team do the Magical Mystery Tour as I would call it, you’re running around looking for the right partners to work with, again this is back in 2015. What we saw back then was that the incumbents were going to spin for a while, try to own everything, just trying to figure out what they do. All major tech transitions happen this way where everyone comes out swinging, “It’s going to be great, we’re going to own everything”, and then after a while they’re like, “Man, this is really hard”, and they kind of settle back into swim lanes.

So, in 2015 it was the let’s own everything, way too many platforms being developed, but my role with the team was to go figure out how does this scale. No, maybe when we met I was probably out there, and that’s when EdgeX got started; I was driving between Bay Area up to San Leandro, and had this epiphany, I was like, “Man, no one’s really architecting for IoT”, I mean Edge wasn’t even the big conversation then properly. And so, I called my team back at headquarters and said, “Hey, if I can get some funding, what if we tried this?” and that’s how EdgeX got started. But in general, it was just how do we grow, how do we scale things, build that ecosystem.

So that was the initial effort from the client side, so when we met I was still part of the client team that had kind of gotten IoT started. Then around late 2017, this was after the EMC acquisition, we decided that we needed to create a business unit that was more Dell Technologies focused, versus one of the BU’s, and so I helped build up the team there working with a bunch of folks, and that was a virtual business unit, the first of its kind that spanned the portfolio, actually was led out of VMware. So, when I was CTO I worked for VMware leadership, and it was a mix of VMware and Dell resources and worked across the various teams. So, as the CTO for that org represented Dell Technologies as a portfolio, doing thought leadership, hired a team of domain experts and product managers so it was not only doing the CTO role working with a bunch of other technology leaders across the company, but also helping to define our solution roadmap.

It’s that intersection between outbound marketing, how do you build an ecosystem and partnerships going out into the field, but then also how do you take that entire back into how you’re building solutions. So, it was a core function within that broader virtual business unit, it was a lot of fun working with the teams. So, continued to build-up the ecosystem, continued to build-up EdgeX out in the market, but then also the last project that I worked on, actually for a few years, leading within Dell Technologies, was this effort around data trust, data confidence, that turned into a project that’s emerging in Lennox Foundation called Alvarium. Because I think the next big wave in terms of a tech trend, is going to be around data trust.

Sounds like a great topic for another podcast actually! [Laughter]

Yes, yes.

I can have that as one of my topics. But that one, and I fully agree with you, coming from the ag-tech space, working at Syngenta at the time, the idea of trusted exchanges when you deal with futures, crop futures in some sense, created a whole conundrum of who do you trust, and how will people put their information together so we can be aggregated, all the way down of course down to the individual subscriber level. I think we should do another one on that.

Kudos to the program at the time, the partnership, as I remember one of our other portfolio companies, Eigen Innovation, participated, because you guys took originally what I’ll call, industrial PC, put it down in a smart gateway form, in fact we’re looking for companies to really exercise that, and of course they had a very strong edge, vision pattern-matching if you will, machine learning at the edge application, using infrared for manufacturing applications, and they loved the program. They told me about going to your big event, and all the support they got and everything else, so they clearly benefitted, and I know a lot of other companies did as well.

You know, there’s been a lot of hype in terms of describing Edge computing, as any good marketing term everybody lops on. How would you define Edge computing, and what do you see as the key benefits versus as many of us noticed versus cloud computing per se?

First off, you mentioned that you’re kind of lobbing on the various terms. I often joke about… have you seen the movie, My Big Fat Greek Wedding?

I have as a matter of fact yes. Some versions of it if I remember right!

That movie that sprays Windex on everything, is like, “You’ve got a cut. Here’s some Windex”, whatever. And so, I joke that these are all Windex’s of technology, Edge, AI, 5G, digital twin, you name it, all super-important trends and whatnot, but also people overuse the words and do washing of the terms, and it gets confusing, and Edge is definitely, “Hey, we just figured out what the cloud is”, and now you’re trying to forget what the Edge is. In simple terms, so first off, you’ve got to look at it holistically. In simple terms, the way I’ve defined Edge for a while, Edge computing its moving compute as close as both necessary and feasible to the subscribers that need it.

So, the necessary and feasible, you can’t move massive amounts of compute feasibly, very close to users, like right on top of that device in the physical world. Necessary, could include reasons like latency and bandwidth, security, privacy, autonomy, all the very commonly cited reasons of why you would do Edge computing. So, it’s really about a continuum, and instead of thinking of Edge as there’s thin and thick, near and far, and all these different ambiguous terms, it’s what are the inherent technical tradeoffs that will make you need to run workloads closer to the source of data. So, number one would be, is that use case latency-critical? Or is it latency-sensitive? If its latency-critical like your airbag, you will never run it over a wider area network. So that’s why you’re seeing processing happening right inside of vehicles, is that they do assisted or autonomous driving, factory floors and whatnot, all the process stuff is there.

5G you will augment that, but it’s not going to replace the need for doing local compute for latency critical or safety critical. Another key delinear in edge is, is it an a physically secure data center, or is it not in a physically secure data center? Does it not have a network perimeter?  So, anything above a digital on-prem data center up into the regional and access edges in the telco or service provider sense, those are all physically secure datacenters.

Outside of the datacenter, you’ve got stuff out in the wild. Of course, there’s mobile devices in client, those are solved problems, but then in the IoT sense it’s the Wild West. So, you’ve got to approach security in a different way when you’re outside of a physically secured space, you don’t have a firewall that you necessarily can rely on.

Then in Edge, the last one is – and a lot of people forget about this, is, are the devices capable of running applications, or are they not, in the sense of being abstracted. So virtual machines, containers, you name it, and of course you’ve got IOS and Android in the mobile sense, but it’s again IoT Wild West. That’s a big one because when you look at Edge computing, a lot of Edge talk comes from telco’s IoT-centric companies, this myopic view that all compute is going to be happening in data centers, it’s just the old hammer and nail syndrome; everything’s going to come to me and I’m going to process it. But the reality is, is more processing is happening everywhere, including in constrained devices themselves.

So, one of the big trends that I’ve been tracking is TinyML, so this notion, and granted it’s going to be fixed function AI and machine learning, and stuff like that, but again it’s part of the continuum; you’ve got to look at Edge as a holistic point of view, not from one perspective. And ideally in the end we have abstracted workloads; the infrastructure is consistent, different paradigms, but some more principles across the continuum. You abstract software in the sense that you can transport workloads anywhere along that continuum, based on trade-offs of economics and performance, and things like that.

Then the last thing you want to virtualize is the data itself, the data needs to be transportable, and then of course we talked about trust briefly just a minute ago, focus on trusted data and you win, and then the infrastructure is just that conduit. Then on top of that it’s about domain knowledge, so all of the stuff, Edge computing, cloud computing, AI, 5G, all very-very important trends, but really I think it’s equally important, actually most important, well first to start with outcomes, but also to architect properly because you don’t know all the answers today, so you need to architect now for flexibility, which is why I’ve been so big and open all these years, in this notion of interoperability.

We couldn’t agree more on TinyML, and thus the reason for investing in Edge Impulse, is a new company who I know you’ve talked with not too long ago.

Yeah.

Speaking of open, I know you’re instrumental in EdgeX. Can you tell us a bit about the initiative, and the role you see it playing in IoT and the Edge computing ecosystem?

Yes, as I mentioned there was this epiphany to start EdgeX. I was driving on the Magical Mystery Tour back in 2015, scouting for different technologies, and honestly actually EdgeX was initially about internal to Dell, what’s the right architecture for IoT and Edge computing. This was also back in the time when fog was a term being promoted, and I had a number of people trying to convince me that we should call it FogX, and I’m glad that I stuck to my guns on Edge!

[Laughter] Right above them would be probably be a Flavio Bonomi who I think is attributed with B&B inventor, the term fog computing! [Laughter]

Yes, and then X of course is so that it could be trademarked. Anyway, it initially started as, okay what’s the right architecture, and we quickly arrived, had a good team in Dell CTO architecting a guy, Jim White, who led that team and very quickly arrived at, “Hey, we need to extend cloud native principles”, principles within Cloud Foundry, driving that whole effort at the time. We need to drive that closer and closer to the physical world, anywhere that you can, because you need that flexibility in how you architect, scale out, bring together different components around common API’s, all those principles.

So, we started then, and initially it kind of, “Hey, well is this a Dell platform? Do we create our own platform?” And then we’re like, “No-no-no, we don’t want to be the 401^st IoT platform”, so we decided open source. We built it up for a few years, we leveraged the partner program within Dell that we were growing to build visibility for it, we launched it into Linux Foundation in 2017, and the whole point of EdgeX is to drive interoperability in IoT and Edge computing. Open source is the modern way to drive standards, gone are the days… Of course, standards are important, STO’s and stuff like that, but code brings together people very quickly, and you can run code, you can’t run a document.

So, leveraging open source is a way to drive standards, we recognize back then of course there’s some good standards efforts around different protocols, but there’s never going to be one protocol, there will never be one OS choice, the hardware is inherently fragmented the closer you get to the devices out there. So, EdgeX is basically set up to be an interoperability framework, otherwise the old standards joke is, we’re going to fix the standard’s problem with one new standard, and so the way you fix it is, you bring together standards.

So that was the goal, and we started in 2017, and it just hit I think about 6 million downloads earlier this year, and so definitely we’ve seen a hockey stick grow, I’ve actually been focused a lot more in different projects lately because it’s often living a life of its own. But in terms of the role within IoT and Edge computing, it’s just I think foundationally important to have an open approach to get to the real potential. The real potential of IoT it starts with intranet, it starts with small solutions, of course let’s not get crazy, you’ve got to solve some basic business challenges upfront, but the real potential is all of these different use cases and markets, starting to align and drive new outcomes, new experiences, think of retail crossover into the home, how do you solve problems in the power grid by starting to sync different things together. You won’t get there, one, without some level of opening or interoperability, and two, without trust. That’s the discussion on that being the next big effort to focus on.

So, I think it’s not all about EdgeX, of course there’s a lot of great efforts out there, but the net is pick something whether its EdgeX or otherwise and focus on how you can build a strategy around some sort of snowball effect, leveraging an opensource component that drives a network effect. Trying to go build everything on your own, even these days is just not a good idea, just think about the importance of an open approach to get to the real potential.

You’ve got a phenomenal background, and certainly working from the bare metal all the way up, all of that has converged as of late last year, this great little company called ZEDEDA, as you said an Edge orchestration company. What was your inspiration for joining them, and what problems are they trying to solve?

Yes! I’d known Said and Roman and a handful of folks. Actually, some people say ZEDEDA, some people ZEDEDA (different pronunciations), I think it’s actually ZEDEDA. The way I remember is if you think of a Run DMC song, “My Zededa”, it will stick in your head.

[Laughter] I should have practiced that before the podcast!

Even within our own company we say it different ways. Anyway, I just thought of that random, that’s how I remember it, “My Zededa” (sings). Anyway, when it got to the point Dell Technologies, the level I was doing there, we had a really great team, but I was just like every single one of my blogs that I would write, I could replace Dell with ZEDEDA, and I’m like “Hm”. So, I called up Said our CEO, and I said, “Maybe I should just come work for you”. The reason why I was interested is, one, ZEDEDA as an IoT Edge orchestration company, our foundation is within LF Edge, so it’s actually E, ProjectE which we use as the open foundation is a sister project to EdgeX.

I’ve been very involved in LF Edge, helped to get that bootstrapped with Linux Foundation, we moved EdgeX under that umbrella, there’s other projects within there. So, part of the reason was I could keep working in that community, part of the reason was of course that open philosophy in general. I think what we’re doing at ZEDEDA is foundationally important also getting into that trust conversation, because the last project that I did at Dell and we led the launch into the Linux Foundation into this emerging project Alvarium, is around this notion of data confidence fabrics, and I think that as a general trend is import. So, for all these reasons I felt like it made a lot of sense when I was looking for a change, to come over, joined back in November of last year, so coming up to a year. ZEDEDA’s focus is in a very open way, how do I orchestrate compute distributed outside of a physically secure datacenter? So, lots of great solutions, spinning up existing and also spinning up at the service provider, telco edges, of course there’s a lot of stuff happening in the cloud. But once you get past the point of being inside of a physically secure datacenter, it gets very complex quick. But there’s a range of devices that are between about 256 megs of memory at the lowest end, so a single node with 256 megs of memory, which is the bare minimum these days of being able to abstract workloads with virtual machines and containers, and that greatly simplifies the fragmentation around the hardware.

So, our focus is from that limit, up to maybe a small server cluster deployed outside of a datacenter. It could be on a factory floor, it could be in a retail store etc. There are similar principles to what you would see in data centers, but you need to focus on how you secure them because they’re not physically secured, you don’t necessarily have that network perimeter. The fragmentation is an issue, scale factor, whatnot, but the other big thing that’s a challenge, it’s not just about the technologies, even though they’re similar you need necessarily different tools to address the unique nature of these devices, but also the pricing schemes are different. The datacenter companies that have tried to do this IoT compute Edge, one, they’re trying to adapt datacenter technologies, it doesn’t work, too heavy resource intensive. Two, the pricing models are so different that it breaks down in terms of competing with their own business. And so, it’s just a different paradigm; so, we see data center paradigm, you see this distributed IoT edge compute paradigm, and then of course you see very-very constrained device paradigm.

There, when you look at orchestration on constrained devices, you’re lucky to get one that max 10 bucks a year for a device. On the other extreme the data center folks, tens of thousands of dollars, no problem, you just scale up into your big data center stuff. And then the IoT Edge is somewhere in-between. So, it’s both technology and the business model that’s different. But really its opened IoT edge orchestration, platform, enabled people to run their choice of hardware apps, and with their choice of cloud, we’re not on the data path at all, that’s important – there’s a lot of people out there that are trying to do a little bit of everything, but when you do everything you rarely do one thing well. So, it’s really just about how do I help you… once you get past a party of a few POC, and you realize, “Man, this is difficult to scale”, that’s where we really come in and help our customers, is how do you scale it.

We had the opportunity at your invitation, so thank you for that, to participate in your ZEDEDA – ZEDEDA (pronunciation), excuse me, transform [laughs]! You got me now! ZEDEDA transform conference you did about a month ago, and all of the topics were certainly near and dear to the Edge, in whether Industrial IoT or mobility applications, and so we were quite pleased to be part of it, but it really was an all-star cast of speakers. So, you’re making an impact relatively quick, in what’s been a relatively quiet company probably up until this point. So, thank you for that.

In closing, can you provide recommendations of books and/or resources that inspire you?

Yeah. It’s interesting, so I get a lot of my inspiration, I read a lot of things, I skim a lot of stuff, I kind of form my own opinions. In terms of books, this is an old one but I often reference “Innovator’s Dilemma” when I worked with different providers, I see a lot of people that are struggling to transform out there, and so I joke that there’s a long list of people that I’m going to send the book, “Innovators Dilemma” to them as a gift. “Start With Why” as a good read, again a lot of the stuff that I gravitate towards is how do you bootstrap innovation.

Last year I read a book called ‘The Right IT’, that one is pretty good, and he had this concept around pretotyping which is I think important, it’s basically before you even get to a prototype, before you even start building something, is really see if someone is willing to buy it. Literally there were all kinds of different ideas around how you can go create test market for something before you’ve even actually built anything. Of course, you’ve got your kick-starters of the world. I think there’s a lot of people that come up with an idea, and they’re so afraid of failing that they just keep driving into that idea before they’re really tested it out, and kind of gotten the honest truth about it, so that was an interesting one.

I follow a bunch of different folks out there, and outside of the tech world I like podcasts like Radio Lab and stuff like that, just various different topics. I think it’s important to get out of your own headspace to inspire you, of course play music for fun on the side. But one that’s stuck with me over the years, and this is completely random, have you ever seen on YouTube a video called Leadership Lessons from a shirtless dancing guy?

[Laughter] I have not, but I’m going to have to look at it now!

So, it’s literally this guy, it starts with this booming voice, kind of narrative, it’s a little hokey but still it’s a video from a concert up on a hill, on the grassy hill at some outdoor venue. It’s this whole group of people, someone’s filming with their camera-phone, and this is probably speaking of a decade of the Edge is probably like 2010, I don’t know. This guy was having the time of his life dancing on the zone, no shirt, just running around doing cartwheels, just on his own. It talks about how movements get started, the booming voice. Then this other guy goes and joins him and they start doing cartwheels and hi-5’s, and just having so much fun, and it basically starts with the lone nut, and then the most important part of a movement is that first follower, because that validates the lone nut, and now they’re no longer that crazy person with an idea.

And then long story short, by the end of the video everybody, there’s like hundreds of people on one side were all on the other side dancing away. So, it’s totally kind of a hokey video or whatever, but it really illustrates how movements get started, and you have to a) not be afraid of being the lone nut, and b) it’s very important to get that first follower.

Excellent. Well Jason, thank you for this insightful interview.

No, thank you, thanks for having me.

So this has been Jason Shepherd, VP of Ecosystem for ZEDEDA. Longtime edge pioneer, and I’ll add in, lone nut! That would make me probably your first or fast follower! [Laughter].

I’m proud to be a lone nut.

There you go. So, thank you for listening, and please join us next week for the next episode of our digital industry leadership podcast series. I guess I’ll have to subtitle this, “The Lone Nut Series” now! Produced, by, for, and about digital industry leaders. Thank you so much Jason.

Thank you.

[End]