Explore how the Spectro Cloud Palette platform secures Kubernetes & edge management for Government IT teams. Access the podcast series to accelerate ATO today.
[Anthony Jimenez]
Welcome back to Carahcast, the podcast from Carahsoft, the trusted government IT solutions provider. Subscribe to get the latest technology updates in the public sector. I'm Anthony Jimenez, your host from the Carahsoft production team.
On behalf of Spectro Cloud and Carahsoft, we would like to welcome you to today's podcast, focused around securing Kubernetes at the edge with Spectro Cloud Palette. Mark Perry, Director of Public Sector Growth, and Colton Shaw, Principal Architect, will discuss how government agencies operating in disconnected, rugged environments must deploy and manage modern cloud-native workloads while maintaining regulatory compliance.
[Mark Perry]
So I want to thank everybody for joining us today and tuning into this, whether you're seeing it on any of the platforms out there. But we're just really glad that you are tuning in and listening to us today. So we are going to be touching on, I think, a very prominent thing within the technology industry, specifically for the public sector, which would be Tactical Edge.
So my name is Mark Perry. I'm the Director of Growth for Spectro Cloud Government, and I'm joined with esteemed Colton. Yeah.
Colton, Principal Architect, Spectro Cloud. Perfect. Thanks, Colton.
So let's really unpack that, right? We talk about Tactical Edge, or we hear it. And we assume that it means one thing over another.
Even with this podcast, I was kind of reading through it, and I saw this large title, and it's talking about contested environments. It's talking about Edge. It's talking about equipment and hardware.
So when we take that, and again, if we unpack Tactical Edge to you, and especially on the engineering side and working with these programs, what does it mean to you? Yeah.
[Colton Shaw]
I think that's an interesting question. A lot of people think Tactical Edge is just getting internet out to the edge of the warfighter, right? But there's a lot more that goes into just general Tactical Edge.
It's how can I push compute downrange? How can I better understand the sensor data that I have downrange? How can I make decisions in a disconnected environment?
And how can I operate all of that seamlessly in the middle of war, really? And so we often might think that it's just the data, or it's just the hardware, or it's just the people. But it's really combining all of that together into one picture.
[Mark Perry]
Right. And I think you kind of pointed out a couple things. When we talk tactical, to really dive into it, right?
So if I kind of piece it apart, right? We think tactical, and again, I was an Army infantry guy, so when I think tactical, I'm thinking weapons. I'm thinking this is going to drive, we're going to drop something on someone, or something like that.
But tactical can mean different scenarios, right? Right. Okay.
So I think you touched on sensor, and I guess people talk about real-time telemetry. And there's a lot of buzzwords, and I think there's a lot of things driving this. In your mind, when we think of tactical edge now, and how prominent it is, what do you think are some of the things that are really driving it?
Why now? Yeah.
[Colton Shaw]
Oh, man. I think AI is a big thing. Just across the board, AI driving the need for all of the data that we have.
We finally got AI to a point to where we can deploy it in a small enough box to where we can push it down to the edge, and we know that it can take a heavy cognitive load off of anyone out there needing to process all of this data, and allow stuff just to happen immediately. So AI is probably the one thing, and the fact that we have hardware that can support it now, and that we know that really, enemies in contested environments can kill our communication, right? We can't rely on one thing.
So all of that combined together is forcing us to have to make decisions faster at the edge, and now we have the capability, and it's getting better every day, basically, to do any of this.
[Mark Perry]
Okay. So, and I really like that aspect. We talk about AI constantly now, right?
And I think that as we take it in, correct me if I'm wrong, we're kind of taking a lot of this stuff that we're so used to doing within a finite environment, right? In a data center, something that's completely locked down, are we finding a lot of challenges to do that? Because again, tactical edge, to me, is something that's happening very quickly at a pace, and it needs to be done, stood up fast, it needs to be, I guess, even innovated faster, and stuff like that.
So it poses a lot of risk when I think of it, taking it from these closed off type of environments, and then moving it out into, I guess, say the real world, right? Wherever we want. So I mean, besides AI, when you kind of think about it, what's some of the other things that we think of for edge, right?
And even then, you know, taking a step back, talking about AI, so are we kind of saying that now I can run like full fledged, you know, hey, I need to look at a bunch of different things, or there's some things that we're going to kind of just pick apart, right? Right.
[Colton Shaw]
So a lot of people think maybe AI at the edge means that you're going to take your chat GPTs, your clods, your large language models of the world, and I'm going to deploy them with these really big GPUs downrange, right? That's not happening, for the most part. If it is, it's in pretty experimental phases on how large can those models, those capabilities be downrange, really, what's happening is, can I take a model, train it aggressively on data that I understand, I control?
So my sensor data to identify a target that might look like this, right? I want to, and then I'm going to how small can I pack that down? So for some context, you have these large models, they 70 billion parameter models, right?
That's like considered a large local model. And that just means how large the training set is that we have for the data. Well, I can't do that at the edge, I need lots of big GPUs, all of that.
So how can I shrink that down small enough to where I have data with it, well, I have a reasonably sized model, but then I can push that to the edge onto a drone, run it without killing the battery on it, right? So it can have that constant loop of ingesting data, making decisions, and understanding it. So I'm not, to answer your question in a long way, I'm not shipping large models downrange.
I might be tweaking, updating models, but really what ends up happening is I might be swapping out the training data for the models downrange too, so lower bandwidth data coming in. So I'll push my multi-gig model once when I deploy the drone, deploy whatever my edge device is, and then I can keep it updated by swapping out the data, the knowledge database essentially behind it.
[Mark Perry]
Okay. You know, and I think in a lot of discussion today, I think about just the conversation you and I have had just from last night to this morning and everything else. And by the way, I saw that fashion towards me being an Army Infantry guy.
You were like, no, Mark, you can't take CHAT GPT with you into the battle. I mean, it would be great. But you know, I think about it now from not only the AI side and being able to do it.
So now if I kind of take that to a higher level. So what we're effectively telling people is with Tactical Edge and the innovation now, we don't have to worry so much about infrastructure, right? I feel now we could go back and pay attention to applications.
The ones that are being utilized, the ones that need to be utilized, right? And I think, you know, a word that you and I spoke about quite a bit lately is standardizing, right? Right.
And creating that type of environment that, yeah, AI is great. You might not be there now, but you still have, you know, what I call, you know, legacy applications that have to be supported, right? And these have to be done at the edge.
There's no, you know, ifs, ands, or buts about it, right? They have to be done. And so now as I'm doing that and I'm building it out, I have these subset of applications that have to be done.
So now I feel like, you know, this is probably where it comes into and a lot of the work you and I do at Spectral Cloud is, you know, the ability to do this at an agnostic level as best as you can, right? And then also knowing that you're probably not going to be the only one in that environment and being able to, I guess you say, collaborate, bring everybody together, I think is kind of essential, right? So we talk about software all the time, and this is leading somewhere in a formula at all from what you said is, you know, we're paying attention to AI quite a bit because it is prominent, but the problem with AI a lot of times is what?
I think you said it a few times, and I really want to dive more into that and kind of how you and I have been speaking a lot lately about connectivity, right? You know, everybody pays attention that we have these AI models and we use it, you know, we talk about cloud or chat GPT or, you know, large language models and, you know, generative AI, right? But when you start to unpack, you know, like a genetic AI and actual use cases for applications, how am I supposed to do that in an environment where it's so locked down, it's so legacy, and then even better yet, you know, how do I start to take that now to what we commonly refer to as like, you said it, you said yourself, like downrange and for the folks out there don't know downrange, which most people do, especially in our industry is that's anywhere away from base, right? So a lot of us were, when we went downrange, it was, hey, we're going out literally to the range or you're going out to the desert somewhere.
So when we say that and we're taking it, you know, how important is it, and again, kind of summarizing all that, and I feel like I'm kind of talking through my thoughts here, but again, when we take that and we kind of deliver it, how important would you say a disconnected, I'm going to kind of formally say it, disconnected standardized platform is when it comes to edge operations?
[Colton Shaw]
Right. So with standardizing around disconnected, I think is really what we've lacked the last 10, 15 years. We talk about it.
It's been a buzzword for a little while, but we've often designed our platforms to assume we have internet access. There was an army exercise back in 2024 that I was reading about that the issue wasn't signals jammed or anything like that. The issue was all their tools required bandwidth and they all got on the tools at the same time and caused a congested environment and took everything down.
So because it was designed around connectivity and bandwidth and everyone using it, when you need it, it's not going to work. So how I see this happening is we're pushing everything to the edge, but we're designing for that local disconnected environment. So one thing I would say is internet connectivity is more of an enhancement than anything.
So sure, I can design for internet, but that's my enhancement. So I make my decisions, I process my data, I pull out the anomalies all happening at the edge, and then when I have the chance to get back, sync up, plug it in, that's an enhancement. I can then send that data upstream, allow larger aggregation to happen there, but I'm not stopped because my signal's jammed, my Starlink's not working, something is bad.
I'm designing for that edge component here.
[Mark Perry]
Okay, yeah, see, and again, I like what you're saying here, and the fact is, if I had to go in somewhere and talk to somebody, so essentially leading off with, hey, let's build this disconnected first, is probably one of the most, I guess you could say, advantageous things I could do if I'm exploring that edge, tactical edge, far edge, near edge. There's so many, I wish we would have drinks. So every time we set a buzzword, you had to drink.
I don't know if we'd make it through this podcast.
[Colton Shaw]
I knew you wouldn't be able to make it through that at all.
[Mark Perry]
Yeah, I definitely don't think I'd make it. But there's a lot of edge, but I think, again, that aspect of it, and I think about it from the realm in the 90s, especially when I was in, and again, it was the infantry, I wasn't in the cyberspace or anything like that, but in my world, everything was disconnected. You know, there was no real connectivity.
It was shortwave or radio, and there wasn't a lot of things where you had to worry about it, but now we live in a day and age that even adversaries are ahead of us in some realms, right? And I think the old way, I guess you could say, of thinking and doing just doesn't really, I don't want to say it doesn't apply here because you still have things you have to support, but, again, thinking about it from the connectivity side, and honestly, I guess I've thought about it, but when you brought it up last night and this morning again, it started to resonate more with me that, yeah, you are right.
Like, you can't assume, and I think everybody knows what we say about assumptions or assuming, but, and we'll keep it PG, but that aspect is, let's touch on connectivity again. So we talked about AI, right, at the edge. Does edge include autonomous vehicles and UAVs, and do we classify that the same?
Because when we think about it, and, you know, I definitely want to start to talk about things like lightweight today and, you know, what the heavyweight factors could be within these environments. Do you consider, like, you know, you on that engineering side, and, you know, even you, some of your experience, like, at Mattermost, building out, you know, messaging, you know, within these environments and how people are doing things through all these different levels, and, again, taking that aspect, are we talking like edge is for UAVs and autonomous and, you know, drones? Like, I mean, is that considered edge?
[Colton Shaw]
Man, I think everything deployed out there is edge, and making it all work together in a perfect picture is, like, that's the goal one day, and what I mean by that, and also the local AI components. Like, imagine you have a drone flying overhead that's able to utilize LiDAR to understand the topology underneath it, right? We can identify enemies.
We can identify drain ditches, bushes, trees, and then you have autonomous vehicles that are able to go out and survey and push stuff forward as well. All of that is a closed-loop system, for the most part, because we have training on the drone that understands what things are, how to classify, identify things based on training that happens centralized, right? When we build the product, we can push out that training data for it.
It's running there. The autonomous vehicle is also running with its own local AI interface training data, all of that as well, and those all working together, but then they're pushing the next hop up, right? I'm locally deployed.
I'm able to ingest that data wherever my data center locally is, right? Back of a GE tent, doesn't really matter, but I'm able to ingest that data, and then I can start to make decisions right there based on that, but it also happens by itself, right? And then take that one step, I can take all of that, then I can start to aggregate the important stuff, because when we talk edge, I think 99% of what we see at the edge is junk data, as I would kind of call it, right?
Great point. And that's not the data that is valuable to anyone, really. We know that's a tree.
Cool. But the ability to say, oh, that's an enemy. It's got a Z painted on it.
It looks like this kind of profile that I've been trained on. Let's push it up so we can build up that training data set. So all of that working together, when we talk tactical edge, I think it's all of that, and it's really no one vendor, no one company can do everything.
It comes down to how you all work together. So at SpectraCloud, right, we're able to deploy and standardize what's at the edge. You can push out profiles, packs that have controlled data, controlled applications on top of it, but we don't do the networking.
We don't build. You might be able to push networking configuration, but it's all the ability to have this kind of harmonious ecosystem of companies working together at that edge.
[Mark Perry]
Perfect. And so, man, okay, this is why I love our conversations, why it's always fun talking to you and getting your aspect on a lot of things. So hinting at a couple of things right there, I think, again, you talked about standardization, and then I do want to talk about, because I noticed you're not talking the infrastructure piece a lot, like I even brought up like, hey, drones and stuff like that, but I think this is where, and I don't want to speak for our audience out there, but I think where a lot of our heads at is, without saying it, which you did, you mentioned applications and data multiple, multiple times. And I think one of the issues in our industry is we focus so much on infrastructure, and when I speak of infrastructure, I think that tangible, what kind of servers are there?
What kind of units are we talking about? Flyaway kits, field kits, whatever. Do you think sometimes that people are focusing too much on the infrastructure and not enough on, again, I think you're naming a lot of the points, data, applications, connectivity.
How about mission outcome? What are we trying to solve? And I think that conversation comes up almost after the fact, right?
We talk about, hey, I know you're going to be doing this. Hey, I have this great piece of software, you could do it, and not so much, by the way, do you have connectivity? So if you had to, and I guess let's take a step back and then again, put that into formulating.
What I really want to ask you here is, as we build that out in my frame of mind, and stack rank them the way you want, I think disconnected first, I'm going to go with that. And then I go paying attention to the applications, hash mission outcomes, and then how is that data going to be accessed, secured, moved? Your thoughts kind of on that when we start to take even AI into perspective, because that's huge data loads, right?
It doesn't matter, right? We can pull in things, and in our realm, it's that whole hub and spoke. I heard it, and I like to refer to it now as micro cloud, right?
Being able to kind of take something, maybe a large, more data center-esque kind of environment, but then pushing them out to multiple devices. So again, as we do that, should I automatically, in my mind, very first would be disconnected first. Because this seems to me very sensitive data.
And then instead of so much thinking about infrastructure, and again, labeling itself for you, and kind of proposing it this way, right? Is I don't think I have to worry so much about infrastructure. We have just, I mean, we talked about earlier, right?
You have devices, I can get a petabyte literally in this area now. So why do you think we're still so worried about infrastructure instead of not focusing on the applications, and missions, and things like that?
[Colton Shaw]
That's a big question. I think infrastructure, the DOW knows really well right now. So it's easy to talk to.
It's easy to understand how to deploy infrastructure, because that has clear ins and outs, right? We were talking like routers and stuff like that. If you want to debug a router, right?
I know what I plug in. I know what comes out. I don't really know what happens in the middle.
Infrastructure is easy, right? Easy to an extent. Obviously, you have to understand what you need, all of that, but those ins and outs are easier.
It's just a unit. But then once you get the software overlaid on top of it, that's where it starts to get complicated, because you're running different networks inside of that software. You're running different applications that rely on different parts of software, right?
I might want one stack, or one app that does secure collaboration, but in order for me to do that, I need authentication, I need a data store, and then maybe I want to interact with other data, sensor data coming in and push that into that. Now I need four or five things just for that one, and then I got to get it ATOed. I have to understand the security risk, all of that around my one application.
So it makes it hard to deploy that rapidly, because that changes. Software changes every month nowadays. So I think that's where components like standardized deployments, right?
So if you think ATO in a box is a kind of concept that everyone's been throwing around lately, but that's the ability to pull out components and say, well, here's Keycloak. That's already been STIG. That's a FIPS deployed version of it that has all of the certification passing, so I can deploy that without an issue.
So the ability to pull all of those nuggets together and then deploy it, that's where the issue becomes difficult. And one of the things that I was reading as well, we don't want people deployed at the edge to be Kubernetes experts. That's not what they signed up for, how to debug Kubernetes there.
And it's difficult, right? So if you're deploying, again, infrastructure stuff, I can unplug it, I can swap it around, clear stuff. But the software side of it can take a really long time to get correct.
So you don't want them to have to, in the middle of a fight, have to go back and troubleshoot what's going on. Oh no, why is this application not booting up, right? What's happened and how can I get it working again?
All of that becomes risk if you think about it that way. So I don't know clear answer to your question, but it just becomes quite difficult to deploy all of that.
[Mark Perry]
There you go. So man, and again, there's so many things, I feel like this could turn into two, three hours easily here. And I think this is hard to summarize all of this and why I think more people should talk about, again, going back to what the tactical edge is and what it means.
I mean, there's so many factors. I mean, you nailed quite a few again that I think every company should pay attention to. And again, there's a lot of things that I've seen in Spectral Cloud that I like that we're doing and that I think we should always continue to do to be able to support, especially the government and even outside of that.
And that's something we can quickly pick apart too, is we talk about the government, we use DOW or DOD or something like that. But there's very much a fundamental risk and need for security in all regulated industries. We see a flood of edge devices, let's just call them that, that are out on the market from oil and gas to energy and healthcare.
These telemetry devices and carts. Heck, I don't know if you ever seen it. I don't know how much you spend in a VA hospital, but it's, being a veteran, I've worked in the VA hospital, there's robots running around and delivering things within the hospital and stuff like that, that you have adversaries and you have people that are actively trying to break through and work within these.
So again, there is a question here and not so much like a question. I think this is a great topic for us to get into. You mentioned AI, you mentioned security, we mentioned disconnected in quite a few things.
So now what happens when we have these foes, adversaries that have the same things, right? Because I could use AI technically, I could go in and use a chat GPT, figure out code or some way to break into something. So when we think of security, if I had to go back and revamp my top three now, what if I was to say disconnected first, security, and then, hey, let's maybe pay attention to apps and data.
So again, how important is it when we think about now things running out in the real world? Security. And I really want to dive in here and get your answer because we didn't actually, I don't think the last day we really went that deep on the security piece of it, but now you really got my brain just working overtime because I think about the need and you said it, software changes so rapidly.
Our adversaries learn so rapidly. So how important is it to have built-in security features within these platforms that allow you to move hopefully at the same speed, if not faster, but vulnerabilities automatically tracked and applied, updates, versions. And I think that actively takes a lot of people to do it.
But then to your point, how are people going to do this at the edge and keep it secure?
[Colton Shaw]
Right. And I mean, it's always interesting because you can see, I mean, all the IoT devices we have today, too, same concept. Obviously, lives aren't necessarily, well, some might be, but your smart fridge, all of your stuff like that, right?
Vulnerabilities out. Someone's pushing data or stealing data from your network. It's the same concept repeated over again.
It's the hardest thing to control in the fastest moving environment, I would say. And especially with at the edge, because now, like you said, we have to control security when we may not ever have access or not ever, but we may not have reliable access to internet, right? Or back home to pull updates or whatever that looks like.
So I think it comes down to really standardizing how we would deploy that. So going back to my ATO in a box example, like at the edge, I would deploy an application with the security controls all around it that essentially lives isolated, right? You can't fix everything, but we have that known working, good working state of this is the risk.
This is how it's operating today. And then someone central can patch those updates and push them down. So when you do have access, you're able to download rapid updates to versions, stuff like that, pass that over and then deploy and maintain that.
Because I think the security issue becomes even worse when you say operators at the edge have to manage the security vulnerabilities. Because now they're not just managing themselves, equipment, the lives of those around them. Now they're managing security vulnerabilities, all of that versus if you say, okay, well, we can have a secure when you have access to connectivity, we have a secure easy, you don't even have to touch it.
You plug it in, it parses updates, runs it, boom, you're done. If it fails, you have that AB processing, right? We're going to load up the new operating system, all the stuff.
If that fails to boot, you still have your known good working state. So all of that together, I think is how we maintain. It's not perfect, but it's how we maintain those security vulnerabilities.
Again, in that closed loop system though, too, I don't have to have internet access. I don't have to rely on signal jamming being an issue. I'm operating on the device with its known how it's going to operate, right?
I'm operating on that device. It's running everything locally for me. So outside interference, not necessarily an issue.
Vulnerabilities, it's not connected to the internet. So someone would have to physically grab the device or something along those lines to then cause an issue with it. So you've kind of isolated those risks the best you can, moving as fast as we are moving nowadays.
[Mark Perry]
Okay. So, man, you know, I start to get going when you get going, man. And I'm like, so I love, I love this again, because so I take it back and I think of 20 year old, you know, Mark running around, you know, in the infantry and, you know, I always thought I was smarter than I was, but definitely a knucklehead when I think about it.
And, you know, what you are referring to, there's been again, gosh, I wish we had a drink for every time we have these, but we've said Kubernetes and, you know, edge and all these different things. So in my mind at 20 years old, I would have been like, you want me to do what? Right.
So I think what we're talking about here and where I think we need to make sure, you know, ourselves in Spectral Cloud and how, you know, we're attacking this front is making sure this is not just built out from a secure state, not, you know, robust enough or even lightweight enough in some instances to be able to run on multiple devices, you know, multiple clouds, whether you have connectivity, whether you don't, all these aspects, but we need to build it for 20 year old Mark.
Right.
[Colton Shaw]
Plug and play, USB drive, plug it in. It does its thing. Yeah.
You don't have to break it. You can't. You're good.
[Mark Perry]
Yes.
[Colton Shaw]
Yeah.
[Mark Perry]
Because I can break some stuff, especially when I was 20 years old, definitely. And I think that's one of the things I definitely think about when I go in there, you know, and sometimes I get kind of nervous because I'm like, you're going to be doing it? Okay.
Let's make this as easy as possible, right? And I'm trying to remember exact things you said, but, you know, I'd like your aspect of, you know, we use some terminology, you know, within it, when we think about taking all these layers and profiles, and you said packs, right. And all these different things, you know, I think of words like immutable.
Right. You know, declarative, you know, it's another hardened, you know. That's three shots.
Yeah. So can you, like, I think you did a really good job. You know, maybe it was some of the stuff I was reading, but, or when we were talking about, you know, if you could pick, let's just talk about immutable and declarative.
You know, what aspect of why are those, or I guess kind of breaking down the definition of those two, and then why they're so important for the edge. Right. No.
[Colton Shaw]
So immutable, essentially, when I've deployed that operating system, that is the operating system, right? It's a signed version of the operating system that we can't make changes on. Aside from potentially pulling something from upstream, that is the operating system.
So I think that's where it comes down to, you can't break it. An enemy can't deploy bad software onto it or something, right? It is, that is what we've designed it to be.
Declarative being like this, and this works, like I've declared that this is the version. This is what it is. These are the components of it.
And then that gets baked into the software. So that way something happens, it needs to reboot, go back to a state, something like that. We have all that information built onto it, the steps of how to recover.
Everything lives right on that device.
[Mark Perry]
Okay, perfect. So one, don't threaten me with a good time, I could probably try to break it. Two, okay, so I see how this is important.
If I kind of formulate this in my brain. So you're telling me that you could give me this type of environment, no matter where I need to take this. And almost in the aspect, you build these out kind of prior to the mission you're trying to do.
[Colton Shaw]
Right.
[Mark Perry]
And I can have multiple of these. And I think in our realm, let's just call it what it is, right? A lot of these are containerized, cloud native type of environments where I could build out multiple of these, secure them, push all these policies, everything that's attributed to that environment, hand it off.
They could click a couple buttons, deploy it, and then not mess with it.
[Colton Shaw]
Right. I mean, imagine being able to essentially ship a flash drive with everything on it. And like, I can say, I'm going to create a bootable ISO, right, back home with my applications, my policies, my layers on top of it, plug that in, boot from that on your device, whatever it is, and I've now deployed exactly what I intended to back home.
And then you give it internet connectivity. And now it can pull up stream updates based on whatever is available at that time. Right.
We're pulling updates. Now we're syncing with into a forward known good state as well. And then it, you know, simple that way.
But then you can also do that same concept with drones, with UAVs, with other equipment that can be deployed, containerized, even just the OS though itself. We have that bootable disk that we can plug and play and boot from. And it's dummy proof to maybe use a bad example there.
[Mark Perry]
I don't know. He really stared at me a long time before he said that. But okay.
So rounding that back up, you know, again, and this isn't so much the environment I came from, because again, you know, I think tactical edge definitely meant something. I don't think I ever used the word edge when I was in the military, because that just didn't respond to me. But if I had to use an analogy or kind of round this back up in how important I think, not only from the edge, and I do think that's a very prominent thing, but how we are tackling, you know, everything from adversaries to, you know, and let's kind of touch on this.
And I think we talked about again is not everything is a war time conflict. Not everything is, everything has to be built for war. It's everything is also for training.
There'd be training or, you know, learning and, you know, things like that. But again, if I had to choose an analogy, you know, my rifle or a weapon that I think essentially what you're saying is, what if I had a rifle that acted like every other rifle that I picked up in the world? Not only that, but I never had to worry about what type of round it's firing.
It could fire all of them. It actually, based upon my mission, it automatically has it already loaded. I go out there and it fires exactly the same way every time.
It's standardized that even anybody in my unit could pick it up and use it. But that to me sounds, one, pretty crazy. I'd probably call some BS.
But in my mind, that's extremely valuable though, right? A constant standardized, even better yet when I think about it, rounds are different sizes. Rounds have different abilities.
And knowing that I don't have to worry about it not firing or misfiring or not happening. Because to me, that's what we're talking about here, right? You're talking about building a platform that, yeah, there's some customization that has to happen, but you built it for those mission outcomes and created these templates based upon where they're going, what they have access to.
And I can have multiple templates, say within something that I can push out. And I could do that in a couple of clicks. So kind of, we are talking about the same thing, right?
[Colton Shaw]
Yeah, I think to use a different word for your analogy too, predictability. That's what we're looking for in its entirety is the ability to know I do these things, and this is my outcome. Again, looping back to that black box router, right?
Predictable. I plug in my ethernet here. I plug in one here.
I send a packet over it and it goes in and out. It's predictable, right? I might not, this is a whole black box to me.
I might not know the router what's happening. But I know I set it up this way, just like your guns being within calibration and understanding how to work them, right? It's predictable.
I pick up that gun, shoots a bullet that way. And that's the same for software. That's the goal, really.
I pick up this drone. I use this device. I use my radio.
And I know when I go to use it, it does this. So that way when we're not building for war, but people often, you have to think under pressure, right, in those situations. And that's, I think, what we want to build for in the long run as well, is the ability to not have to add cognitive load under pressure, because that's where fail points can happen much easier.
If you say, the operator can pick up this device, click this button or do this thing, they know it's going to do it. I mean, that's your immutable, defined operating system and applications all on top. I know I'm going to pick up this device.
I'm going to deploy it. And it's going to do what I was trained and told it's going to do, because it's been designed that way. And we're pushing policies and constantly maintaining that.
[Mark Perry]
See, I love that. See, and I love the aspect you made. That sounds so much better.
But yeah, that's, I think you're touching on what that true edge, tactical edge, and you mentioned it again, that build, right? You know, humans were fallible. You know, and I think you're nailing it again.
And I don't think it's a buzzword that you're saying. It's a necessity, predictability. We can always predict, one, that some of the major causes could be human error.
But if we could discount those and create a platform that's predictable, that is repeatable, and it's not, it doesn't change its looks or how it reacts or acts, you know, in different environments, but looks agnostic, allows you to run within all these different infrastructure environments and everything else. I think that's how we kind of win that tactical edge story, right?
[Colton Shaw]
Oh, yeah.
[Mark Perry]
And, you know, not even taking into account that we talk about adversaries, but we have friends in the world, right? You know, I think a great example is, and man, now you got me thinking about like this morning, you talked about Australia, right? What about, what happens when things break?
[Colton Shaw]
Oh, man, that's always fun because, so that story for everyone. So in my past company, we had a customer deploy doing a training exercise, multi-nation training exercise deployed out in Australia. And they were trying to build up essentially secure communication, pulling in all these different data sets from one.
And they had a known working Kubernetes deployment, but it was still in like that testing phase. And they had called me up and they're like, well, it broke, right? We need to fix it.
It broke. And we don't ship Kubernetes experts out here. So we don't really know how to bring it back up.
It's like, okay, well, what can you show me? Well, I can't show you anything. Well, what can you share with me?
I can't share with you anything. And in those situations, again, imagine this was training. So they were very lucky.
And obviously that's where these bugs are essentially designed to be kind of pulled out in, but so we had to get on a call and they were able to call and you close your eyes and you imagine what's happening. And because you have to bring them up as someone sitting in the U.S. who knows how they're deployed. I have to help them be brought back up because the deployment is being tested.
It's not been standardized. And so we're showing those flaws on it. But stuff like that can happen in the real world.
Probably happens all the time. So the ability to really control everything that you can control, right? Like they made a change and it broke.
There wasn't a bad change, but they made, okay, well, what if you could just roll back to a known working state, right? Didn't come up. Let's go back.
Oh, we're still working. We haven't stopped the mission because I did an update last night, a security update, and now everything's broken today. No, I tried it.
Didn't apply. We're going to go back over here. We're going to keep running.
All my software's still working good.
[Mark Perry]
Bam. So I don't think, well, I know I could have said it better, but that's why I bring it up because now bringing this back into, you know, even us at SpectraCloud and what we do from day to day in how important it is, and you touched on it from not only the people, the security side of it, you know, the standardization, you know, we talked about declarative and immutable. So everything you're saying not only leads to the fact is I can have a unified platform to deliver all my applications, mission outcomes, and in easily digestible, easily to deploy, but I think where you just plugged in another piece that people don't think about, what happens though when something goes wrong?
What happens when it doesn't? And it's going to go wrong, most likely. And we all think about, we all know that.
And how come, you know, even me, it took me, you know, I don't know how far we are in this podcast right now. I might need help on time checks, but it's, you know, it took me a second to go back and go, you know what? You brought up a great point.
What about the support? So that underlying thing, not only from the infrastructure side, because infrastructure fails, we know it back from the old time. There's a meantime before failure, right?
A component, there's only a certain time that that will live. It will operate everything else. What happens now?
What happens to these software stacks that are on them? What happens when you are in that difficult situation? You can't call in.
So now I think about Tactical Edge being that disconnected has a whole nother meaning. That not only are you disconnected from the internet, you're disconnected from support. So having that.
You don't have Google. Exactly. So I'm not sitting there asking chat GPT, you know, how do I change my breaks?
I'm not saying I've done that, never. Um, but, you know, it's, I now think of declarative, mutable, hardened, you know, and then I, now let's ask more buzzwords. Why not?
Yeah, I think we'd probably be done with that whiskey bottle by now. But, you know, um, drift. Yeah.
Stuff happens, right? Right. I'm in a different time zone now.
What happens when I, I used to have issues with, you know, um, time protocols, you know, and like those things, you know, there's, there's so much that goes through my mind now that I'm like, what happens? So now, you know, again, having that easy, repeatable, standardized, you know, platform is even more valuable because now I've kind of limited that support. Right.
And it's kind of a question. It's more of like a tub light. You know, I don't think we talk about enough, right?
That it's easy. We talk about, you know, maybe small, you know, um, frontline teams and everything else that require certain units that are disposable. They could just throw it away.
You know, something happens, right? Well, how do I mimic that again then? You know, and how do I make that easy to mimic?
So again, I think I'm rounding up everything you're talking about. And it kind of blows my mind that we don't think of the support side so much of it because they need it. You know, I guess, right.
[Colton Shaw]
Yeah.
[Mark Perry]
Thinking about this way, you know, you talked about it before, you know, some, some of your time and even mine was spent in the commercial world. Right. And when you think about it from a commercial world, it's just, Hey, these guys have people that are sitting there.
They're not going into, you know, uh, war theaters or things like that, but they do have dispersed, um, headquarters or, you know, things like that. And it's almost the same thing. What happens when something goes down?
Right. So when we think about that from the tactical edge right now, and we haven't talked about this, how does cloud play into this? Because there's still mandates within, you know, DOW, DOD, and I wouldn't say mandates, but there's still cloud spend that's happening.
Right. Yeah. Can't get away from that.
No. So is, is this something that, I don't know, just magically thinking, let's think I have, you know, Viasat or something, and I'm out there and then all of a sudden I do have a support issue. Um, how important it is, and I'm hinting here, um, to have something that's agnostic enough that, Hey, by the way, uh, corporal Perry, that dumb, dumb 20 year old lost the USB key.
Yeah. Um, what if I could get connectivity? Can I, do I have the ability and how important is that ability to pull down that same stack again?
[Colton Shaw]
I, I mean, I think that is the only, like, it's the most important thing to be able to, you have an issue at the edge and you need to redeploy for whatever various reason, you need to redeploy exactly what you just had in pretty much every way. You have to be able to declare what that is, pull that down and redeploy it without that. I mean, you said config drift and all these other things can happen that will essentially, um, cause your data to not be correct.
So I guess an analogy, um, a gun analogy, right? You have 10 guns and you want them all in calibration to make sure that they're all shooting the same way. Like you said before, shooting the same way.
When I point that way, it shoots that way and it does what I expect it to do. That's the same for your applications, your data, all of that. If we have drift in our config or we don't have that standard pack, that standard control of what I'm deploying.
Well, now my data is probably not reliable because I don't know what changes have happened downrange. I don't know what changes have happened to that device, that data, any of that. So when I do go ingest it right at the big clouds, I don't know if I can trust it because there could be changes.
So when we've standardized in a repeatable controlled way to deploy all of that, now I can trust my data because I know what's going in, it's coming out. I know all of that process in there. So I think, and going back to your cloud, um, like we can't get away from clouds and stuff.
I think, I think they all work really well in kind of my perfect world picture. I think a lot of people's perfect world picture is you have your kind of micro clouds as you call them, deployed locally, and they are essentially self-sufficient. They're able to deploy, understand what's happening, ingest the data, all in its own closed loop system, and then train and learn in that system.
But then on top of that, when it gets connectivity, it can then be pushed up to the large enterprise cloud. And then you can have, we've detected this anomaly, this anomaly, this anomaly. We've already decided on them, right?
It's past data for us, but it's valuable training data for us. So then push that back up. So now when we go to redeploy again, gen two, gen three, whatever, we can have a better data set, a better understanding of what's happening in the commercial side.
But then they can also understand these, right? And this goes back to standardization. They have, they can understand on the commercial side that this group with these devices did this, and this is what happened.
This is what was detected. This is the reaction. This is how it impacted the overall effort.
Now that becomes data. Push back up, and now we can aggregate that at a larger scale in the commercial cloud to then understand our big picture, understand how we need to move, what we need to do, and then start to push that down as well. So this is happening, and then we can standardize all of that.
So that's kind of that big picture when everything works really well together. You can kind of think about it like a smart home, maybe with a bad analogy, right? Like I walk in the door.
I want my certain lights to turn on. I want some music to turn on. I want the temperature to be set right when I'm driving home.
It's the same concept as that too, but again, you're deploying your, it's people, it's lives, all of that, but you want it to happen. And think about how angry you are. I know I'm going on a tangent, but my Siri sometimes doesn't turn on the right lights, doesn't do what you ask it to do.
And you're like, come on, I got to go get up and do that. But it's the same level of automation, the same level of reliability. You become so reliant on the data coming from the cloud, the data coming from my teams down range, and then I become so reliant on how to action on that when something breaks.
I get really mad. I yell at my Siri, what's going on? Turn that light on.
I got to get up and do it.
[Mark Perry]
I swear I shouldn't be talking back to her, but I really do. I think they remember. Yeah, probably.
I know. I'm probably on a watch list somewhere for multiple reasons. But I know that's a lot to unpack.
I kind of know a lot there. You know, and I think this is constantly what we do too, is I start to kind of think of how I put this together. And you ever played laser tag?
Oh God, yes. Okay, so you got me thinking that it really fundamentally, and I'm going to have to use this later if this hits, but thinking about it like this, right? You get hit and there was this form of laser tag that when you got hit, you had to go back to the base.
Or like tag in. Tag in, and then it charged you back up. And then there was power-ups that were added.
And you've got to go to the base and not each power-up was the same for each one. And then those bases could not be accessed by the other team. That was your base, right?
So I think that's what we're kind of talking about here, right? Is it essentially, if I have something, boom, I get hit, it goes down. I could go back to that base.
Right. I could, boom, come back up to that active state I was in. Bring up my last known good state.
Last known good state. And then even better yet, I want to add something to it. Now I'm at the edge and I'm like, cool, I'm doing my normal, I don't know, we're uploading surveillance, photos, videos, documents.
I want to add AI aspect to it. I go back to that base station. I have a power-up now.
Bam. I can add that into my environment now, specifically for me. That's my power-up.
Nobody else can access it. And that's kind of what we're talking about here, right? And no deviation.
I don't have to worry about that power-up's going to be there or that baseline. Oh, crap, I didn't charge back up. You know, it's going to happen, right?
And I think those are the things that we could do a better job on because I think what we're talking about today that, you know, if we haven't alluded to it enough, data centers are leaving. They're moving out. People want more and more now at the edge and being able to do it locally.
[Colton Shaw]
Yeah, especially with hardware changing so much. We're able to have such small, really highly powerful hardware down at the edge now.
[Mark Perry]
Yeah. And so, you know, as I do that, again, I don't think we could probably hammer it home enough today is the need for all of us to create a platform that is as agnostic as possible because not everybody is going to be using the exact same thing because that is something when we talk about laser tech, everybody has the same pack. Every single one of those packs are the exact same.
You know, paintball was different. Paintball, anybody could have, somebody could have the better gun. Get your upgrades in paintball, yeah.
Yeah. So now in those realms, how are we competing? How are we making sure that everything works and is compatible with all those upgrades that are out there?
How do we make it available not only when you are connected, but I think the real, you know, proponent is when you're disconnected.
[Colton Shaw]
Right.
[Mark Perry]
You know, how about the fact that laser tech, what if it's, you know, and I think we moved into that where all of a sudden I didn't have to go back to a base. Bam, I get hit, three second delay, recharge, you know, something like that, right? So now we move to that disconnected state that everybody needs to make sure they pay attention to, which, you know, I'll be brutally honest within the industry and, you know, feel like I'm calling you out, probably am, but there's some that don't, right?
They don't build it like that. And I think, you know, I mentioned earlier, and I don't know if you agreed with me. I want the smart guy to agree with me, but, you know, I said that we should live in a disconnected first world, right?
If we build to that, I feel like that's easier than to move to that connected world, right? And then again, making it easy, secure, not something that, you know, old soldier Perry could break and then not have it come back up again, you know? And lastly, we talked, you know, kind of the pace of things and how fast things are happening.
You know, and I think about Spectral Cloud and I think about how flexible we need to be, you know, what Spectral Cloud is about six years old, right? You know, we've had what we call our vertex, you know, I guess you could say our pallet vertex, our software edition that's for regulated environments. Right.
Yeah. That was word for word marketing. You better love me for that.
But when I talk about the reason I'm bringing it up, it's about a three, four-year-old product and the way it's matured, the way, you know, it's moved, I feel like it's light years. You look at some of the other ones out there and they're still acting like legacy things, you know, and so again, it's not really a question coming to you for it is more of you to kind of jump in on that. I think lastly, to kind of bring this all together is I think we need to, as an industry, embrace tactical edges and go anywhere.
This is the new realm, you know, we are doing it as, you know, society as consumers. You mentioned smart homes, you mentioned the fact that we're taking all this stuff on the go and we have been for a while, you know, iPhone, if we call it out, came out in 2007, right? But I think you had it in your papers, you might have mentioned it, we still fundamentally can use an iPhone when it's completely disconnected still.
Right. Now there's some things we can't, there's still things we can access and use.
[Colton Shaw]
You still have all your data on there, your apps, and that's essentially what the tactical edge is, is your iPhone and airplane mode, right? You're flying and you might be able to play games, I can see data, I can understand it, I can connect to people local, but I have what I have today. And then when I connect back, I sync up, I get my text messages, all that again.
It's the same thing. And I think to help wrap up a little kind of what you're saying for that standardization, all of that. So an example, I've been working with Air Mobility Command AMC out of the 618th and Scott for the better part of four or five years.
They've built this very large public program called Project Nightmare, not project, just nightmare. They've renamed it to MICE, I'll butcher that. But essentially what it is, is it is ingesting all of this data from all these different sources that usually you'd have to log in at a computer or terminal, log into that, see that, okay, well, that says that.
Okay, now let me log out. Oh, I want to go check over here. And so all this data exists in all these other places.
[Colton Shaw]
Sounds like a nightmare.
[Colton Shaw]
And so it's bringing all that data into a standardized place. So the UDL, Unified Data Library, and then allowing an overlay of AI to then make decisions off of it. So I know this data for this weather, this base comes in.
I know I have mission data coming in. Oh, wow, that's the same base. There's a weather alert and a mission going there.
I'm now going to push that out to the relative people. So they built this incredible system for doing that, that has really operationalized their effectiveness. And so now it's become so good that they want to rinse and repeat to other AOCs.
But it's been built in for them, right? And it's been designed and moving super fast. And it's an incredible product.
But it's been designed in that silo. So now, in order to pick it up, lift and shift it to use it somewhere else, what do we got to do? We've got to essentially rebuild everything again.
And then, oh, well, you want someone else. Probably have to rebuild it again. So the ability to have started from the very beginning, and obviously hindsight's always 20-20, but the ability to now say, well, we want to rebuild it.
Let's rebuild it in a standardized, repeatable way that we can have declarative config, declarative applications, all of that baked on top of it. So now I have mission outcomes defined in a box. And I have provable real-world experience from those mission outcomes.
So now I can start to rinse and repeat that and drop that in every single AOC around the world. You can modify that around. Different data coming in, different outcomes, but the same defined concept.
So I think that's kind of the ultimate goal where I see this industry, the tactical edge, all of this going towards is the ability to say, here is your platform. All the data, all the AIs that you want in the world, all that's pulling in from your local sensors, your local data, whatever might be available. And it's pushing that out and creating a closed loop that can deliver real-world mission outcomes, no matter what that is, what that looks like.
[Mark Perry]
All right. So, I mean, I'm sold. But, yeah, done.
And now, I mean, not to make light of it, not joking, that is, that's a great way to summarize it. Great way to, you know, talk about the need of, you know, I mean, everything. I could even bring this back to when you mentioned ATO, right?
You know, it's not something that's like an afterthought. It's a necessity. It has to be ATO.
It has to have that authorization to be able to run in these environments. And the way you just explained it, understanding these environments, and if I was someone out there, I'd be like, wait a minute. So, you're telling me I can authorize and get that ATO for that mission outcome, and then I can make it immutable, I can harden it, and then I can ship that off to different AOCs or different environments.
And then, well, I just, at worst, maybe have to do an ATO amendment instead of reauthorization. Right.
[Colton Shaw]
Two small changes.
[Mark Perry]
See, that I start to think about and how fast the tactical edge is coming about and how fast things are happening. You know, I think anybody out there that deals with ATOs know how slow they can be. Oh, yeah.
But- 12 to 18 months minimum. Yeah. To me now, you're talking weeks.
Right. That, to me, is the speed we need to move at for a tactical edge. Because, again, when I think of tactical from my side of it, I don't think slow.
I don't think, you know, difficult, non-secure. I don't think all of that is, I think, fast, secure, easy, you know? Right.
Maybe not easy, but it might not be easy to objective, but I've done enough beforehand that I'm prepared.
[Colton Shaw]
Yep, and it's just swapping in and out parts, right? If you were to define an ATO approved version of software, version of the mission outcome, whatever's underneath it, you're swapping in and out parts. Instead of waiting 18 months, you're able to say that, that, that, and that all together.
Oh, they've already been all ATOed. Great. We're going to ship this out now.
[Mark Perry]
Perfect. I mean, I could probably stay here for a lot longer, but, you know, I think lastly, you know, in fact, I'll go last, but if you, if you had to add one more thing or you wanted people to know, or, I mean, kind of your thoughts on tactical edge and what people should take away, not only from this podcast, but what they should really start to have those conversations and how they should be kind of, I guess, formulating their thoughts around tactical edge and where it's going.
[Colton Shaw]
Yeah, I would say if we were to, one or two kind of key things is understanding deploying and relying on a disconnected environment is number one, right? So when we design new tools, software applications, whatever, design it in a way that works air gap, if you want to use another buzzword. So we don't need to rely on what's pulling from the internet, right?
I can deploy it and move it however I want, but then also really understanding what that data is that we want and how to make decisions from it. So that's going to be an ongoing battle, I think, to understand all of the sensor data. There's just, I mean, terabytes of data being generated.
So then instead of solving the problem of pushing all the data upstream, what, how do I define downstream what is important for me and how to action on it so then I can take that nugget and move it up so then I can add to my training data from that? So I think kind of combining both of those together is what I would say we should start to really focus on. What would you say your top ones are?
[Mark Perry]
Yeah, my factor is, you know, kind of the commonality. I think you said the predictability, right? I know that we have separate military services for a reason.
There's different parts of government that does different things, but a standardized platform that no matter a shift that happens with my job or the mission or anything like that, it's still familiar to me. Like, that is a big piece that, you know, I can think about it from, you know, my time I was a mortar guy and there was, you know, 81 millimeter and 60 millimeter and 120 millimeter. But again, they all kind of operated the exact same way.
They basically did, right? That I could go to any of them and still be able to get on target the best I could. And I think about what we're doing and how important it is to the tactical edge to find that commonality, that predictability, that, you know, not only that, but the sustainability.
That as I move from environments to missions to, you know, even next generation things, I need something that can support me five years down the road as I add in things. Or I decide now that it is an air gap, you know, hey, a lot of this data that we've collected and everything else is no longer, you know, classified information. I want to move it.
I want to be able to do it. So having the ability to do that, and especially around the tactical edge, I think is just not only essential, it's crucial. So, I mean, that's kind of my take on it.
I would love everything just to be, you make it look like an old Nintendo controller. I would love that, just ABA, you know. So that's what I really like about it.
And why I think this talk, again, we could probably do 10 episodes. We probably really should continue on different topics for all of this. Yeah.
Because I think, again, you know, and nobody can probably see it here, but we do have notes and everything and kind of, you know, our thought process. And I think you found out just in the last day is there's a lot out there. There's almost in the fact that sometimes we think about it a little wrong.
But again, to round up what we talked about today and a lot of key aspects that you touched on, was that predictability, commonality, security, and, you know, that disconnected mentality and first. And that kind of goes back into that security. So, yeah, I would love to do more with you.
Absolutely. So we'll have to see if that happens. Appreciate everybody for tuning in, checking out the podcast.
Hopefully we'll be back. MC, I don't know. I don't know what we call it.
We're going to have to come up with a name if we keep doing this. So appreciate it. And thank you.
Awesome. Thank you.
[Anthony Jimenez]
Thanks for listening. And thank you to our guests, Mark Perry and Colton Shaw. Don't forget to like, comment and subscribe to Carahcast.
Be sure to listen to our other discussions. If you'd like more information on how Spectro Cloud can assist your organization, please visit www.carahsoft.com or email us at SpectroCloud@carahsoft.com. Thanks again for listening and have a great day.
Bye.