Our CEO, Sam Nagar, was featured in an episode of Latest Shiny Podcast. The podcast is hosted by Stephen Spector and Rob Hirschfeld. They discuss technical aspects of Pixeom technology, value proposition, and other topics. Here we present you the podcast transcript.
Stephen: Welcome to the edition of Latest Shiny Podcast. This is your host Stephen Spector. With me, of course, is Rob Hirschfeld good afternoon Rob.
ROB: Stephen. Good afternoon, So it’s pleasure
Stephen: Always a pleasure in this is our continuing busy week of podcasts. I know these come out once a week, but for us, we’re getting about your podcast has done now before everyone disappears for the summer and takes their long vacations, except me. I don’t think I’m going anywhere this summer, but well, I should take a vacation. I don’t know why.
ROB: A vacation for you would be moving to a new house
Stephen: I’ve done that. This is I’ve now moved two times in the last year. So I’m I’ve told my children I’m not moving anymore. They threatened me. So I move more than another family.
ROB: It’s this the Boise sampler pack, answer this there and you said hey, I think I want to like of the taster menu
Stephen: you’re just running around every couple months. That’s funny what we have. Have a really interesting guest a new company to me. I’m very excited. It’s a pure Edge company which you know, we’re always on the lookout for Edge companies. So let me introduce the co-founder and CEO the company’s name is Pixeom and the gentleman joining us is Sam Nagar, Sam, Did I do it? Right or I probably just did it wrong again.
SAM : No, it’s fine. It’s perfect. Thanks so much for having me on guys
Stephen: Sam just you know, give us a quick overview of yourself and just a high-level overview of what your company does and then Believe It or Not Robble I drill down and we’ll understand what your technology is and go from there.
SAM: Yeah, absolutely. So I’m Sam. I’m the founder and CEO of the company. I did co-found with my sister almost 6 years ago. Now when we started we were doing something different we were trying to make. A better NAS box that had an application platform on it, but we slowly transition into creating an orchestration engine or the edge and kind of being able to deploy containers and large-scale geographically distributed scale. Yeah, and here we are with a hundred people backed by mostly corporate venture capital and dozens of Fortune 500 caliber. companies that are our customers.
ROB: Well, this fantastic. So, I’m interested in your origin story because storage is one of the biggest challenge areas from an edge perspective. So you started from a storage infrastructure.
SAM: Yeah, you know originally I’d been given a gift the Raspberry Pi and this was back in 2013 OpenStack. It was really popular back then and I Appliance OpenStack onto that Raspberry Pi and I thought maybe you know once I’ve gone to now that I’ve you know back then you had to install but in fact multiple nodes, it was really heavy industry might quite I’d streamline quite a bit to get a running on the pi and I thought okay. Well, now that I have data center infrastructure that I’ve you know Appliance into this small footprint device. What else can I do with this? And I thought okay. Well an application that runs an OpenStack should be able to run on this right any kind of data cloud-based application can run here and I wanted to make a platform that could be deployed to these single smaller form factor devices that brand difference, you know despaired applications. So we decided we would you know build that and go after the SMB NAS Market where you know, they were still using. File-based storage and FTP transfers and we thought you know, we could bring object storage to that market and you know Cloud applications and all that and we’re actually going to ship Hardware to and then a full solution right a full nas box better version of a NAS box and then we ended up pivoting pretty shortly after.
ROB: Because you just were what made you think to change out the choice.
SAM: Our first customer and our first investor was Samsung and he had in sedate acquired this company back in 2014 called smart things and this guy, you know, I don’t know how familiar you are with smart things. We put these sensors in the home and the idea is you can kind of as soon as you enter your door all the lights and everything turns on in your home and it’s all automated and you know, they were seeing that there were a Complaints from their customer base. In fact when I was pitching to their Venture arm, I pulled up the website and I showed him I said, look, here’s what some of your customers are saying about the product and there was this really long paragraph remember from like the most like upvoted complaints on the page and this guy described coming home having enough time to go down to his basement getting a beer from his fridge going back upstairs to his living room. And still, have enough time to pick what channel to watch on TV before any of his lights would activate for anything would turn up they were doing all the no doubt up in the cloud. And so Samsung said look, you know, we need to start doing more confused more decision-making on-premise, you know inside the home or inside the facility. Have you ever thought about just licensing your software to us? Because you know, we’re Samsung we can make Hardware. Why are you selling Hardware? We could partner with people like. So that’s actually what made us start pivoting towards a software licensing model. And of course also around this time containers and Docker had you know, they’re starting to reach more of mass popularity. And that’s how we slowly kind of transitioned towards a container-based approach to at the same time.
ROB: Makes sense. Okay. So what I want to highlight in this story, right this is sort of the classic. Edge latency, although write large you know if you’re not thinking about doing local processing for systems like that. It’s hugely problematic. The story is awesome because, you know, a lot of times we think about latency, you know, I might have to wait a second before I get a reply. But as these systems scale, it’s a different problem. And I know there’s a ton of developers out there thinking I wouldn’t make those architecture mistakes my whatever yet right. It’s like you’re living it. So the Storage piece started you added the containers in when you think about Edge right is it? What is the edge? Site for you from that perspective and please don’t spend a lot of time to finding Edge we’re going to eat. That’s a reader exercise. It does. I do want to bracket this a bit because you know if you’re there is sort of this micro and Appliance Edge, there’s infrastructure Edge if you’re delivering applications, what’s that edge to you?
SAM: Absolutely Yeah, and you know what in general you can assume that any time I refer to Edge in my contacts and my Paradigm. I’m usually referring to the on-premise of computer systems that are running some applications. Like for example, one of our customers is one of the world’s largest fast-food chains and they have like, you know some area where they put a few systems. They have a few like HP servers and those servers are supposed to be the common infrastructure for multiple disparate applications and workbooks. So things like payment. The same Inventory management, sometimes this fast-food chain, in particular, has smart kiosks in a lot of the restaurants nowadays and you know that has to do some processing of those kiosks and support for those. So, you know, all those different workloads need to run on a common infrastructure. They don’t want to have different vendors come in and say, okay. Well, here’s my box that does you know, data processing and support for this single system that I’m bringing to the table. And so there are usually be a few, you know a few. Systems there, you know think either anything ranging from the kind of a rack server to kind of just a standard HP or Dell kind of power and you know will get installed to that first and then we start orchestrating the different applications on top of that. Sometimes we get installed on top of an operating system. Sometimes we kind of installed all in one.
ROB: so. I want to pause for a second because I love to do you’ve laid out use cases. You got far enough down the stack to sort of, you know, lay out some of it. It would help me. Can you be very specific about what the application is? You know, it’s just sort of enough for me to get my teeth into its container management. It’s a service running into the infrastructure and API driven something. How would you describe the technical side of what the platform is?
SAM: I guess the right word would be we’re an orchestration agent that can be dropped into an existing environment or included in one. We yeah, we would basically be that layer that manages. The intermediary between the resources and the applications and kind of you know helps maintain the harmony of all of them at once.
ROB: So if I wrote an application, I containerize it I’ve got something I need to distribute to all of my outlets right? Let’s assume let’s take the generic Fast Food Outlet. So I’ve got 10,000 stores all of them to have some infrastructure I have menu operation thing that displays menus or something like that. That’s an app running in the store. How do I deliver that to your system in a way that it then knows to distribute it to all of my stores?
SAM: Oh, yeah. So there’s we call it like a back end component is called the portal and the reason I called backend and instead of the cloud is because sometimes even that gets deployed on-premise for some of our customers, but this portal. It is a window for you to see all of your infrastructure all of this all those servers and systems that have our agent running on it and what applications are running on it. So basically you’ll see the kind of if you use the UI you’ll see just a scrolling list of all the systems there and you can select and you click on it. Most of our customers don’t use the UI, but you can click on all of the ones you want to deploy to may be out of you know, 30 40 thousand restaurants worldwide. You only want to test out a new version of the application or some may be new application all together on a thousand of them. And you know with our are the portal is what kind of makes that workflow happen. So you can let’s say you call an API to the portal say, you know, give me a list of all systems available to me that have at least you know, 16 gigabytes of RAM at least free that have you know enough time of you know, this much percentage and maybe in this geography. There’s an API to be calling these things and then you can from that list say, okay. Well, this is a list of the system of the match my criteria submit another API and say, okay push out this application to this list of systems and we make sure that that happens and we orchestrated and sometimes each application is other parameters, which I won’t dive into just yet but there are others and sure primers.
ROB: So in that in that case, you’ve created a multi-cluster or did you consider each Edge location cluster infrastructure? Is it sort of a standalone thing or do you how do you put what’s your sort of mental framework for this type of environment?
SAM: Yeah. So we do support both basically if you take our agent and put it onto a system you can operate as a standalone system or it can cluster together with the other systems too or operate in what’s called alike in our terminology. We refer to it as a tier. We’re like maybe you might have a tier of two systems and a tier of eight systems and it’s your of like 16 systems and each of those represents. Maybe one that has more availability or one that has lower latency because maybe it’s right, you know right next to the data source. But yeah, it’s just it’s configurable to decide. If you want it all to operate as one entity or a few entities or each individual.
ROB: And so what makes I mean, so I’m going to ask you the sort of the basic notes. what you’re describing is a multi-datacenter infrastructure configuration management engine, right? What makes that edge specific from your perspective. What’s the thing that makes it different from you know? What we have before to manage this type of environment.
SAM: Absolutely it’s you know, and that’s why you know will sell customers like okay, if you’re just doing for a small handful of facilities, if you want some infrastructure there, you might not want to come to us because we’re we really shine is in this is going to highlight the question you said what is the difference when you’re dealing with a highly geographically distributed, you know type of plane, so to speak you’re dealing with a lot of disparate hardware and resources available in each of those locations potentially also disparate workloads and a high level of multi-tenancy to you know, in a data center you’re going to have just the people managing the data center and then you’re going to have you know, whoever you give access to the IF and service, but with us you’re going to have like the SI that comes in and sets up the system and potentially does some deployment of it you’re going to have the End customer who’s going to have their own set of tenants potentially someone’s General as an administrator. Someone who’s just use you’re going to have you know, sometimes there are some use cases of ours where there are three or four large companies providing one like Enterprise service to you know, a big customer account to or there might be like a carrier and a computer systems maker and in other infrastructure or service provider and other vendors doing depend a little. So when that becomes extremely complicated to do at the edge, whereas in a data center is seen as one facility in one set of infrastructure.
ROB: And a data center people tend to assume you’re going to solve that with cloud and virtualization and give people individual virtual resources, right? If you’re doing containers only then you’re going to have things much more mixed sounds like that perspective.
SAM: Yeah, and you know, this is going to be different expectations from all the different tenants since.
ROB: So is there a rationale I mean you’re doing containers? It sounds I mean usually when I think of containers it’s very application focused. And so somebody’s delivering an application you um multi-tenant which says that each tenant might be delivering an application to the shared infrastructure. That’s a different sort of a different management Paradigm. How do you make sure that? People are getting put in the right places and the resources are where they think and then how does somebody who’s managing an app across a thousand systems troubleshoot it and figure out what the go. I know that’s a huge question set but.
SAM: Yeah, absolutely and that’s kind of related to kind of our go-to Market in general, you know, we leverage Partners a lot. Google Cloud was one of our first Partner’s in early days, and I don’t know if you guys have. Seem like the kind of the main diagram that I like to show a lot of different speaking sessions and stuff. But show how to like up in Google Cloud, there are some services and we match all the APIs on-premise in all of some of our service containers and you know for kind of in the middle in between transitioning from our Samsung days to where we are today. We did brand ourselves for a while as like the Google Cloud version of the azure stack because we had started. Is the Google Cloud thirst and they became a close partner of ours and helped us get a lot of our early Revenue a lot of our early customer accounts and we’ve kind of gone with that approach that strategy that ecosystem play right just like a VMware or Microsoft or one of these kinds of plastic cases of you know, software ecosystems. We leverage our partners and that includes not just, you know, finding new opportunities, but also helping us support these engagements right Our standard Customer is a kind of profile of a customer someone who’s going to be deploying a large scale like many many systems across many facilities. So we don’t have the boots on the ground to send everyone else.
ROB: right and you’re saying something else that I found really interesting based on edge conversations in general which is cloud services, right people are used to these cloud services and you’re actually saying that you’ve replicated the cloud services for the on-premises, so. Does that make applications portable from One Cloud infrastructure into the permit on-premises locations?
SAM: Like if you ask any of our customers this kind of two main value propositions, they get first is they don’t need to have an IT team on-premises. We give all the Remote Management ability to kind of diagnose everything. We’ve heard some customers tell us it costs as much as $2,500 to send an IT Pros. To a facility like with the fast-food chain example, and then the second value proposition to get out of us is they don’t have to re-architect the application as it moves between not just one Cloud to Edge but many in a multi-cloud paradigm or many many different edges. We can replicate not just the services of Google but also Microsoft, Amazon, you know, although sometimes private clouds too like with Samsung we replicated some of their private cloud services. So the idea is your Not just you don’t have to just avoid investing in an IT team. But you also avoid having to waste resources on your engineering team re-architecting again, the same workload. They just want to run on the edge.
ROB: Right, that makes a lot of sense. So, you know I can see there’s an element to hey, I have a platform and what you’re describing we talked about a bit and it’s really exciting to see. A company sort of embracing this as a component which is adjacent the necessary adjacencies when you’re bringing something up you don’t just get to say oh, yeah just bring up something in the edge. Sorry, you’re missing, you know, a load balancer and a storage infrastructure and a database component. Is there a set of services that you think are a minimum set or is it been more broadly distributed?
SAM: Yeah. That’s a really good question. So, you know if you go to anyone. Cloud service providers website go have a hundred-plus give or take Services. You can see and leverage as a developer right now. We’re at 36 of those Services across the major clouds. I think of those hundred though, like half of them you can make a very compelling argument doesn’t make sense to run on the edge like some of those like DNS Services right like a. Yeah, dynamic service. We don’t see a use case at least today for that from our customer base. But you know, we’re kind of we’re nearing there we do View at least half of them. I think would have a use case on the premises for at least the customers were seeing. So we’re almost there. I’d say about 50 is what we want to get to.
ROB: Some of it would be demand-driven. So somebody shows up and says I depend on you know. Assuming something as big as lambdas is huge left, but something smaller, I mean, but if you had said ecosystem, is there a way that you can leverage ecosystem for these platforms instead of you know having to recreate them yourself.
SAM: Yeah, it’s part of our strategy to date.
We haven’t been able to live at least on that part of our strategy. We overall the partner strategy has been more focused on obtaining and supporting the existing customers and a little less on product development. But I think moving forward, you know, it’s not even just the specific cloud services. We’re looking to extend those partnerships we have some Partnerships in the works for services that never existed in the cloud that should exist on the edge like, you know, protocol translation Modbus, OPC, UA type of translation that is happening up in the cloud. You know, there are other ones too, like sometimes like Management Services and all that. They just weren’t using up in the cloud. But some of these providers some of our part soon to be partners. I don’t want to jump the gun and say their names before the announcement comes but some of these partners of ours have had a lot of success on the edge and you know, it just makes sense to have a partnership where we can orchestrate that.
ROB: Makes sense makes a lot of sense. I’m watching the clock a little bit and I have we have a ton of I still have a ton of questions. I want to you’ve been answering my questions really well. Well, is there something critical in how your system operates that we just haven’t touched on that? You know, I don’t want to look I spent all my time on questions. What else what’s it? What’s the big value proposition that we haven’t talked about?
SAM: I say there are two points. I just want to drive home one is you know, I think a lot of players in the space assume that they’re going to get Greenfield deployments and I think that’s tastes like you have to support the Roundfield. No one’s going to just take all their systems to throw them away and you know to buy all new stuff and just put it in there. Maybe there are but generally speaking. At least in our experience. We haven’t encountered that.
ROB: Yeah, that’s for strongly concur with that statement. I think it’s one of the hubris elements of a lot of people promoting new platforms. So I’m glad to see you saying no, we’re not telling you to your way is wrong. Your Hardware is broken. That’s good. What else
SAM: you know, and then the other thing also. It’s not necessarily knock on anyone, in particular, is just another kind of General Trend we see. I think a lot of players underestimate the value of hardening when you know, I think a lot of Enterprises today, they’ll you know, they’re open to doing POCs. They’re open to trying things out. But they’re that gap between moving from a POC to production deployment. And if I don’t think I mentioned in the earlier days are sorry earlier part of this conversation, but you know, we have over a million installs of our software to date and you know to get to that number you have to be hitting production deployments and we spent we’ve had to spend so long we’ve been doing this six years easily three or four of those years were just in hardening. forget developing the features and all that. It was just you know companies telling us. Oh, well are you know, do you have this level of secure some security standard? Like they might have a certain level of encryption. They want, you know where certain policies on you know, handling the data and removing the data. Overtime stuff like that. You know that alone is it’s not just you know building a feature set for the edge that meets the customer requirement is also all that extra nitty-gritty. That’s I mean, it’s not fun to do it’s not sexy to do but you need it if you’re going to get that traction.
ROB: SAM, you’re making my day. This is sustaining to me. This is the sustaining argument and I’m you know, I’m watching a lot we can Stephen. I talked to a lot of edge companies. And we’re watching a lot of these efforts, especially some of the open-source ones where they’re very eager to announce that they’ve completed a POC they have V1 out. And yeah, you don’t hit this stuff until you’re doing real production workloads and you’re hitting the nuances and I think the people of overlook how much sustaining engineering it takes to do an edge deployment because you don’t know those truck rolls are expensive. So that’s I think those are very important things to pick for people to understand, It makes me scratch my head a little bit on some of these, you know open source people thinking they’re going to get something out of the open and I mean even for us are we do a fair bit of Open Source stuff, but it’s the sustaining engineering this the production work. That you want to partner with. So it makes a lot of sense to me as if that’s interesting.
Stephen: Rob. I would say, you know, it’s interesting and Scott talked about it. I’m sorry. That’s got Sam’s excuse me. He talked about real customers and real deployments for an edge company. This kind of seems to be the first I’ve seen that much further along than other companies.
ROB: I think you’re right from that perspective to Sam. One of the things I would ask you right we’ve seen CDN. Being pretty, you know the sort of a slam dunk from an edge perspective. It sounds like you’re moving beyond, you know, helping people with CDN type actions are helping people run infrastructure in these locations. Is that a fair characterization?
ROB: Is there some type of application that fits really well for this type of management Paradigm or is this just fixing a management infrastructure problem?
SAM: To some extent is both but I yeah, I’d say we tend to see a bit more skewed inquiries around like high data were close like Thing video analytics, you know, you don’t want to be sending video data up to the cloud. Of course. I mean that I know probably preaching to the choir, but even then, you know when you want to run down multiple analytics on one stream that also can get complicated too. Look at it too and you know, so we see a lot of requests for that. We there say their name, but they did you go to you go into any building. You see those like shielded cameras everywhere made by usually that one company and they have a bunch of different vendors that they want to be able to incorporate and deploy to like, you know, for I think I mentioned earlier like, you know, you can see if there’s a fire on. Nowadays, right instead of waiting for the smoke detector to work. It’s a smoke detector. And so that becomes complicated. Especially when you want to run on the same infrastructure on the same single camera stream and not have every different vendor provide their own separate, you know NVR box and you know the kind of pass it to each one. That’s a really common one just because of the flexibility of cameras. I think the retails a pretty good one for us that we see a lot. We’ve gotten most of our Traction in like retail industrial energy Telecom. And security surveillance would be kind of the main verticals and use cases. I think that we generally see.
ROB: So along those lines, I mean you’re describing something that I think is important. In that you know, this is the multi-tenancy people sharing data so that this you know as if you can put a bow on this might be the right thing to close out the podcast on. Great, you’re talking about data that is you know sources of data that are valid that are valuable to multiple Downstream components, right is your platform enabling that is there some super challenge your secret sauce. That’s part of enabling that type of many to many mapping.
SAM: It’s part of our offering I wouldn’t say it’s necessary. Like something we’ve submitted as patents or anything like that, but it is a core piece of the overall solution write the overall operating and you know, I think you know that there’s a quote that still sticks with me from one of our bigger customers. They had this head architect in one of the early meetings we did and he took a look at this and he said, you know, this is we need this we should be working very closely with this company for our next rollout. And I remember there was I think someone from marketing on the team or someone from the business side who you know after we have done a deep dive was still wrapping their head around it a little bit and the Architects summed it up so nicely said look The Edge is dirty. It’s not easy to deploy that infrastructure figure out what environment you’re in what that networking looks like what Hardware you’re running on and then figure out how to you know, manage all the different applications and tenants on it and share those resources accordingly. You know, I thought he summed it up nicely. That’s just a quote that’s always stuck with me and just dirty and we make it, you know, Simple and Clean and easy.
Stephen: So Sam I, it’s my time to come and I think I told you before I could keep it around 30 minutes and keep track. I love the edge is dirty that is going to be the title of this podcast in some way. But that’s a new quote like that. So before we end up was like to you know, people Sam want to know more about your company or anything like that, you know, what websites and social media. They should follow are there any events coming up in the next few months this Summer that they might want to look for you.
SAM: Yeah, I’m actually speaking. I have a speaking engagement at sensors Expo next week. But you know in general I’m open to anyone sending me an email. I love we love collaborating with people. I’m a Hands-On CEO, you know, we have a hundred people but almost all of our people our Engineers were a very product-focused company. So I get involved a lot in a lot of our accounts and collaboration. So I mean anyone can send me an email anytime just at firstname.lastname@example.org.
Stephen: Awesome Sam seeing I appreciate you coming on. I know you’ve given us some insights. Both of us are super impressed because of your tangible answers. And so I hope people will take a look at what you’re doing get involved with you and give us all some feedback.
Stephen: Thank you
SAM: Thank you appreciate you guys for inviting me for this.