Protect Your Application, Not Just the Network. Add Zero Trust Superpowers to Your Critical Applications and Systems

When (day):
Fri
At:
18:00 - 19:00
Project:



Session Video

About this session

A overview of zero trust, what it means to be application embedded zero trust and it will contain demo golang code for people to try out live if they desire.

“Applications are the new edge and zero trust is the security industry’s latest hot buzzword. The OpenZiti, open source project (https://openziti.github.io) is a one-stop shop for building truly zero trust solutions. Stop trying to protect your network. Securing the network isn’t enough. Adding zero trust directly to your app is the future and the best way to keep your users safe.

OpenZiti not only provides the required zero trust overlay network but also provides numerous SDKs for use in your favorite language. Can’t go fully app-embedded? OpenZiti also provides clients for all major desktop/mobile operating systems.

In this session you will: • learn some core tenants of zero trust and how it’s different from current network security • see what it means to embed zero trust into your app and why it’s the future for application security • discover the superpowers your app gains by simply incorporating an OpenZiti SDK in your app”

Transcript:

Clint Dovholuk - 00:00 You. Hi.

Dinis Cruz - 00:02 Welcome to the last session of the Open Security Summit in October 2023. And we have Clint, who’s going to talk us about the next evolution, or natural, I would say evolution of the whole Zero Trust, which is the application, which I completely agree that we need to do it there. And by the way, I love that. Isolate your app. Spot on. And we might get some language models too, here. So looking forward to it.

Clint Dovholuk - 00:30 Great. Well, thanks a lot, Denis. Yeah. Again. My name is Clint. Clint Obloblock. I work for a company that sponsors an open source project called Openzd. And as everybody knows, every good open source project must have a mascot. And so down in the lower right hand corner here, you’ll see Ziggy. And Ziggy is our mascot. He is a piece of ZD for zero trust. All right. So anyway, let’s get into it. Usually in the last few years, whenever I tell people that I work for an open source project, that is a Zero Trust overlay Network, this is usually the response I’ll get, right? A big deep eye roll. Because in the last few years, particularly, the term zero trust has kind of been turned into a marketing buzword. So hopefully you’ve seen this presentation. You said, is it real?

Clint Dovholuk - 01:26 Is it legit, or is it not? And by the end of this presentation, I’m going to turn you from a deep eye roll into Shaq here and that kitty cat with a little shimmy shake shake, right? Like, give me that. Zero trust. That’s my goal. If you’re watching this live, or if you’re watching it on a recording and you have Go, I invite you to participate in the demo. That’s going to happen towards the end of the presentation here. So if you’re watching it on recording, pause the video, go install Go, and you’ll get to enjoy the appetizer for our Zero Trust overlay Network. Here’s what we’re going to talk about, and you’ll forgive me if you have seen this sort of stuff before. Maybe you’re familiar with Openzd already. Maybe you’ve seen me give a similar presentation to this.

Clint Dovholuk - 02:14 But this is the current overview. I don’t want to assume anybody knows what Zero Trust is. So we’re going to start at the basics, and we’re going to go from there. So let’s first start by talking about what the current network security setup usually looks like, right? And here’s what it will usually look like. It’ll be a castle and a moat motif. So this is what you hear a lot of that castle and the moat kind of idea. And so that’s what most people think of when they think of current network security. So here we have just a regular basic network, right? And what is a regular basic network? Well, it’s just a bunch of castles and it’s a bunch of moats. And we have individual little networks and individual little locations.

Clint Dovholuk - 02:54 And our job is to allow network to allow devices to send traffic to one another, right? So once you get behind the walls, once you get over the moat, all these devices can send traffic to wherever they want. And that’s what normal network security is all about. Once you’re on the network, you’re considered trusted. Obviously that’s not a great idea. And this was such a bad idea way back in the day that we said, hey, http FTP, that’s maybe not the greatest idea. Let’s make our network a little bit more secure and let’s put TLS or transport layer security everywhere. And so that’s what we did. We went out and we got a little lock icon and we put it on our little individual devices, and now everybody has TLS. So instead of FTP, now we’re using SSH and SCP. Instead of Http.

Clint Dovholuk - 03:45 We’re using Https. Then people said, well, you know, that’s great, Https, but it’s kind of hard to get that certificate. So poof out. Popped this thing called https everywhere. HTPs Everywhere is a project by the Electronic Frontier Foundation, and it was so successful that they’re actually phasing it out. But what it would do is it would take your Http URL that you typed in Google.com and upgrade you automatically to Https. So it would check to see if your browser had access to an actual secure URL. If it did, it would automatically upgrade you. And that was great. Less encrypt was started because everybody was like, hey, this Https is everywhere now, right? How do I get some Https for me? Turns out back in the Dark Ages before let’s encrypt and SSL. Zero, I think is what it is.

Clint Dovholuk - 04:39 There’s another one that’s similar to let’s encrypt now. You would have to pay for a certificate. So most people on their trusted network were like, it’s trusted, I don’t have to bother with this HTPs stuff. Of course, that’s obviously a bad idea. We should not trust our local network. And that’s a central tenet of the term zero trust, right? Trusting that network, you’re safe, you’re secure if you’re on that local network. So why is it a bad thing? Well, obviously if an attacker in this case, represented by our little fella down here in the lower left in red, because he’s dangerous, if they attach to your network even though you have TLS, well, what are they going to do?

Clint Dovholuk - 05:18 They’re going to start sending packets, they’re going to start probing your network, and then if they find an O day or a zero day, or some sort of vulnerability, they’re going to compromise your laptop and poof. Now you’re having a bad time. So horizontally moving across a network something that we really need to control. And if you didn’t know it, this is what VPNs are, right? When you attach to a VPN, you’re basically turning those three little castles into one giant castle that lets you get to anything, anywhere. Most of the time, now you might have an advanced VPN that does a little bit of segmentation, then fantastic, good for you, but lots of people don’t.

Clint Dovholuk - 05:54 And so if you are attaching to a VPN, oftentimes you’re able to get to all of those corporate resources without any worry other than having to be on the VPN. So again, I mentioned the word segmentation. We started to do this idea of segmenting things where maybe now we’ll take our little Ziggy and we’ll put him in a little policeman’s outfit and he’s going to police our traffic. So now when that nasty offender gets on our network and wants to send those packets, we have old Ziggy there taking care of business, stopping those. Cool, cool. Right. Everything’s good? Are we done? Call it a day? Of course not. We can’t call it a day. It’s never going to be good enough. So what happens?

Clint Dovholuk - 06:32 These attackers, they are ever vigilant, they are always out there and once they have a foothold, they’re not going to give up. So they’re going to keep trying. They’re going to compromise a laptop that does have access to the target that they’re looking for and then use that target as a stepping stone to the next network where they’re going to just compromise that target of opportunity. All right, so how do we combat some of this? Enter the idea of zero trust. Again, we’re back at our basic network and I’m going to get rid of the castles and the moats now and we’re going to just look at this diagram. But what if our network had this idea of a device identity? So this is one of the core pillars of Zero Trust, this device identity.

Clint Dovholuk - 07:15 And what it means is every device on your network has a strong identity that the network can use to verify the authenticity of the identity connecting to that network. So if the identity is permitted to connect to the network, then great, it’ll connect to the network and it can send traffic. And if it’s not, then it won’t be able to. And that’s called device identity. This is how Openzd bootstraps device identity. It’s a little complicated. If there’s any questions, hit us up on discourse, since you’ll be watching this in a replay, but hit us up in discourse and ask about how Openzd actually accomplishes secure enrollment. And this is just one flavor of getting your authentication to the overlay network and we’ll see more about what enrolling means in a bit. And then there’s device identity with our attacker.

Clint Dovholuk - 08:05 So if that firewall sorry, if that network has all these little firewalls, that understand device identity. Even if the attacker is to get on your network and be able to start trying to send packets, well, the network is smart enough to not permit that attacker to send these malicious packets, right? The network knows the device identity and doesn’t even let you on oftentimes. That’s called an overlay network. So what are they going to do? Well, they’re going to just go out and they’re going to find themselves an identity in a laptop that does have access. Because once they have a laptop that has access, well, guess what? Game’s over. They’re able to connect to that machine again and Zero trust been defeated, right? Call it a day. Of course not. There’s even more we can do. So what do we do next?

Clint Dovholuk - 08:51 Next we take that idea of segmentation and we lower it even more. We keep narrowing the segment. So instead of having a great big VPN, instead of having a little network, instead of having one device, we keep going further and further down to its logical end, which will be applications. And we’ll see that in a bit too. So this concept of least privilege says every identity on the overlay network is authorized to connect to other identities or other services on that overlay network. If it’s not authorized, you can’t connect. So here we have that same attacker. They’re trying to send data over to our target of opportunity, but ha, they can’t. Because of least privilege, the network knows what services this device can attach to.

Clint Dovholuk - 09:36 And that laptop in the lower right that it’s been trying to compromise all along is not able to connect to that laptop. So of course we’re done, right? Call it good. But as is always, if that identity is compromised and that identity has access to the target of opportunity and to the service in question, then you’re still going to be able to be attacked. But does that mean this is not a good approach? Of course not. We have made this already. Just imagine how many hoops, how many extra hoops this attacker has had to jump through just to find a machine that can even connect to the target of opportunity. It’s all about layers and layers of security. And the more layers you add, obviously the better it is. The smaller your attack surface, the better it is.

Clint Dovholuk - 10:22 So least privilege giving the identities only the ability to contact the things that they’re supposed to contact, narrowing that focus. Another core pillar of zero trust is continuous authorization or posture checking. So this would be things like am I connecting to the service from Windows? I can only get here from Windows or from Mac or from Linux or is my patch level of the operating system up to date? Things along those lines. So if a new patch comes out, you can update a posture check and then suddenly that laptop is no longer able to even attach to the network, no longer able to send traffic. All right, and that’s a little bit about zero trust. So now we’re going to get into application embedded zero trust. You heard mention the term overlay network prior. So what exactly is an overlay network?

Clint Dovholuk - 11:17 You might have already kind of figured it out, but here we’re going to take the Internet, right? Generally speaking, it’s really convenient if the Internet can just be your overlay network. So can we turn our Internet, the Internet, into our overlay network with Opencd? Yeah, you can. So what does that look like? We’ll deploy a controller somewhere out in the Internet, and this controller’s job in life is to manage authorization and authentication. So when a device comes online, it talks to the controller. The controller says, yes, you are who you say you are, and you are now authorized to connect to the network. Sorry, authenticated to connect to the network. And you can actually connect. Authorization is the green light, if you will, giving the identity the ability to actually connect to the servicing question. So we have a controller now.

Clint Dovholuk - 12:11 We have these things we call routers, and there are different kinds of routers, some of which can service links between other routers and some service edge connections. These things come online and they form connections to everything else. So Openzd is an overlay network that’s also a mesh network, which does differentiate it from other overlay technologies, like normal VPNs, where you have one concentrator and one input, one output kind of place. With Opencd overlay, you can have multiple inputs and multiple outputs, and those things all go together and they form secure Mutual TLS connections. And that’s important. And this diagram gets really busy when I add all these locks on here. But that’s important because every single edge router to edge router connection or router to controller, that’s all Mutual TLS, meaning the device attaching, must provide a certificate to the server it’s attaching to.

Clint Dovholuk - 13:04 And the authenticity of the certificate that is being presented is verified by both the client and the server that uses public key infrastructure. If you haven’t looked that up, fun topic, deep topic, but that’s what Mutual TLS is all about. I’m going to take those lock icons away because it makes it busy. And then the edge routers also are there, as I said before, to service connections between edge devices or edge SDKs, and the edge router. And that’s all done for the express purpose of one thing, and that’s sending data back and forth. Obviously, the whole idea of a network is to send and share data. Ideally, it’s with identities which are only authenticated and only authorized to send that data, like on an opencd overlay network.

Clint Dovholuk - 13:52 Here’s the three basic forms of zero trust as we define zero trust, and we’ll go into each one here in a moment. Zero trust, network access. This is what most vendors will tell you is zero trust. You trust your local network. Wink. Not a good idea, right? You trust your local network, you trust your remote network, and then everything in between, we will make zero trust. And that’s pretty darn good. It’s actually not bad, right? It’s pretty good. That’s not terrible, but we can do better. So what if we take an agent and put the agent on their computer. Like for example, on my Windows computer, I’m running the Windows desktop edge for Windows, which looks like this. And so I have a little agent on my computer.

Clint Dovholuk - 14:38 And that little agent’s job in life is to make sure that traffic is only trusted on the device. And that’s pretty darn good. Openzd works like this. Other cool technologies like WireGuard works like this as well. Or your VPN, even if you want to consider the VPN right, there’s a little agent that runs and its job is to intercept all of this traffic and then shuttle it safely over to the other side. If it’s a server, if it’s a client, wherever it’s going to. But everything in between is safe. So even if you’re using Http and Insecure protocol in your device, you can successfully and safely tunnel Http to the server. And we’ll see that happen here in a minute. And then finally, obviously, the end all be all application access.

Clint Dovholuk - 15:22 Keep reducing the attack surface all the way down as far as you can into the application itself. No longer do I trust my host’s OS stack. Let’s go back a moment. If we look at the network access, not application. Come on, where’s my build? Right there. That’s host access. If we look at this, anything on my local machine that sends Http is able to go into this tunnel. So if I curl there, or if I have malware, or if I have the actual app that I’m trying to use, all of which can go into this tunnel and safely tunnel, that is what application embedded zero trust can solve. You no longer have a listening port here, intercepting traffic. You don’t have anything that does that sort of interception. The application simply writes into the zero trust stream by itself. That’s a huge benefit.

Clint Dovholuk - 16:14 And you can even go one step further. With Openzd, you can use a hardware route of trust, which is really cool because then you don’t even have to trust that the file is safe on your operating system. Always there’s some root of trust somewhere here. It would be a little green dongle that you plug into your computer. All right, so all that’s what our application embedded zero trust is all about. Let’s focus now and we’ll just take a look at the attacker and the client that it’s trying to attach to. So what are we going to do? We’re going to take an SDK, we’re going to stuff it into our application, and then magically lock icon shows up and we have a secure application. And that’s basically what end to end embedded zero trust is all about.

Clint Dovholuk - 16:56 Application embedded zero trust cool is no firewalls are needed. We have application embedded zero trust already. We’re good to go. What does that look like when the attacker gets there? Well, that attacker compromises the laptop. You have an application embedded zero trust SCP program, SQL program, and FTP program, right? Secure protocol. SQL. You never know if it’s secure or not. Depends on how you’ve done it. FTP? Definitely not. So if the SCP application wants to send traffic to SCP, it’s permitted to. If the SQL Server wants to, it’s permitted to. If the FTP wants to, it’s permitted to. Obviously. So now what happens if the SCP application tries to send traffic to that FTP application? The SCP application has no idea how to send traffic to FTP. It doesn’t even know how to get there.

Clint Dovholuk - 17:45 As they say, you can’t get there from here, so it’s denied. Same would be true for SQL traffic going to FTP or for the SQL Server program to try to send SCP. It literally cannot be done because it must go over the overlay, and the overlay won’t even permit it to. So why that’s? Super powerful. Now you have effective immunity to malware, showing up on your local computer, compromising, one of those bad applications, super cool stuff. And that’s where Openzd comes into the picture. Now, you might be saying to yourself, this sounds like a lot of work and it is a lot of work, and it was a lot of work. And this slide is one of my favorites. It represents the duration or the longest path a human can walk across the earth, right? And this is basically the path to Zero Trust.

Clint Dovholuk - 18:37 We realize everybody’s not going to go straight from Brownfield application to application embedded Zero Trust Greenfield application. If you can, then great. You get to start in Siberia or maybe in Africa, depending on where your start and end is. But otherwise you’re going to have to go the whole entire path. And so we realize that it is a journey. Open Zero Trust is a journey, and people aren’t going to jump right into Zero Trust and application embedded Zero Trust immediately. So maybe you’ll start somewhere in the middle of Asia, maybe you’ll start somewhere in the middle of Africa, and your journey might be different. But this is the longest path a human can walk and it represents the fact that it is quite the journey.

Clint Dovholuk - 19:20 But Openzd has you covered because we have those things I remember I referred to before as those tunneler apps. So if you have Linux or if you have iOS or Android or Windows like me, or whatever, your particular operating system target is of choice, you can use one of those tunnelers to bridge the gap between application. Embedded Zero Trust and brownfield deployments tunnelers do have a really cool property that is worth noting because if you have an Open ZD overlay network, you will have these tunnelers available to you. And so suddenly you can start doing cool things with all your brownfield apps too. So for example, one of the superpowers from Openzd that tunnelers provide is true private DNS. You can create fictitious DNS names that your users can then connect to.

Clint Dovholuk - 20:12 Like my ZD which is not a valid top level domain or Linux Foundation one summit. Like I made this slide for Bodibic boat face, right? Fluffernutter. These are all DNS entries that you can create and would be private to your bespoke overlay network. So that’s pretty cool. But not only is it private DNS super cool, but it’s authenticated DNS. It’s truly private. Right? If you deny the user and the identity, I should say access to your overlay network poof that DNS entry goes away. And so those are some really powerful superpowers of a tunneler, and we could get into tunnelers someday. That sounds interesting, but what are we here for? We’re here to talk about the superpowers of application embedded zero trust. So let’s blow it up and let’s get going. So what do we have here? We have two pictures.

Clint Dovholuk - 21:06 Perhaps you’re familiar with the picture on the top. That is called the Beast. The Beast is a limousine that the United States presidents drive around in. It is bulletproof. It is blast proof. Right? It’s got all this security built into it, not bolted on top like the Mad Max approach. Right? With the Mad Max approach, you’re like, oh, there’s a chink in the armor right there. Oh, I can shoot through here. Oh, it has regular tires, right? The Beast. Sleek, sexy. Security built inside. Secure from day one, secure by design. Mad Max. We’ll do our best we can. Day two, security. It happens, right? We’ll just keep bolting stuff on until we make it secure because we’ll get there, I’m sure, right? So that’s a key difference of application embedded zero trust is it’s built secure by default.

Clint Dovholuk - 21:59 You have zero trust security built into it right out of the gate. That’s a superpower. Another one. From a developer’s point of view, I don’t know if you’ve ever had a service that’s behind a load balancer, but all too often when you’re behind a load balancer, you as a developer, you don’t even know where this thing is going. So you’re told to connect to my application server, and then that gets turned into an IP address. And then if you’re lucky, the load balancer will forward source IP. And so you as a developer could have a shot at understanding who is connecting to your service before they actually connect to your service. But that’s not the case with an overlay network. Like Openzd. With Openzd, you know, Clint is trying to connect to Prometheus. There’s no question about it. Right?

Clint Dovholuk - 22:46 You know the exact identity connecting, you know the exact identity that’s being connected to that is also a superpower. On top of that superpower is it’s not just for the clients. It’s also for the server side. Right. Usually we think about zero trust. We think about security. You think about the cloud. The cloud is safe, right? Nobody’s going to get into my cloud. My VPC is secure. I’ve only got 85 holes open to my VPC. That’s only 85 holes. Not a worry. Well, with a zero trust overlay like Openzed, you can have zero open holes to your firewall. So your VPC can be truly firewalled off from the world. No open listening ports at all on top of that, so that’s the firewall. On top of that, the server has no listening ports either.

Clint Dovholuk - 23:33 We already talked a little bit before about side channel attacks being impervious to side channel attacks and the local computer, that goes for the server too. So if that FTP server is listening out in VPC land, or virtual network land, or Cloud land, wherever Kubernetes land, it’ll have no listening ports, it is not attackable by IP and port, literally not attackable. And that’s what we’re going to see in our demo in a bit. Oh yeah, I guess this is just more no inbound Firewall polls. Same exact point. I should have looked ahead of my slides, but you get the point, right? No listening ports, no inbound firewall rules. That is just super cool stuff. I don’t know if you are familiar, but a couple of years ago there was a CVE, a critical vulnerability and exploit around Java and Spring called Spring Boot.

Clint Dovholuk - 24:26 And it was kind of a big deal, right? If you got a hole in that firewall, then anybody on the open Internet can do what? Oh, they can just attach right through that hole in the firewall, hit that Spring Boot server, and compromise whatever they want to compromise, because that vulnerability was of astronomical proportions. That’s a good zero day. Guess what you can’t do when you have a zero trust network like Open ZD. You can’t even connect to that, what’s the word? Attackable target. I can’t come up with the right word, so you can’t even connect to it. So the only people who could connect to it are people who have that strong identity. Now of course, if one of them gets compromised, then your bets are off.

Clint Dovholuk - 25:08 But you have reduced your attack vector from something that’s on the Open Internet to, if you’re lucky, whitelisting of IP addresses. But you don’t even have to worry about that because Openzd you can just bypass all that and secure your application directly, have no listening ports, that attacker has no chance to get to that Spring Boot application. And so that was also another issue that the Spring framework had, and I don’t remember exactly what CVE oh, what’s this CVE? Also a 9.8. So a 9.8. If you haven’t seen the CVSS scoring, I do recommend you go check it out. The critical Vulnerability scoring system, it’s pretty neat. It has like six to eight parameters, I don’t know exactly, but the two that I focus one of which is attack vector is Network. Permissions required none.

Clint Dovholuk - 25:58 If you have an attack vector of network with a permissions required of none, you’re already in a bad place, right? That means anybody on the network is able to potentially attack that vulnerable target. So if it’s on the Open Internet, I don’t even know how many billion devices there are out on the Open Internet nowadays that are available to attack that target. With Openzd, you can close all that down. And again, before I told you, another superpower is contacting the client from the server. Usually, if you’re lucky, you know what a server side event is, or you know what a WebSocket is, and you can open a WebSocket to your server and you can get events that way, or you have some sort of IoT type of implementation that allows you to do this sort of stuff too.

Clint Dovholuk - 26:44 Openzd’s overlay network already allows you to do that. So if you were to embed Zero Trust into your server, into your client, it’s just another client on the network, your server is just another client and your client is just another client. And it’s possible for your client to declare that it binds or accepts connections from other clients on the network. So let’s think of like SSH would be one of those ones where you would define an identity as able to SSH or RDP for example. But with Zero Trust enabled by Openzd, you can use application embedded Zero Trust and talk from your server straight to your client. So if you want to notify them all, not a problem. You can just do that. You can just connect to them. But you must be authorized.

Clint Dovholuk - 27:29 The server must be authorized to send a message or dial, and the client must be authorized to accept a message. And then soon Openzd is going to allow the clients to turn off that ability of listening, even so that the device you with your tunneler could turn off the ability to listen. So you are in the control there too. That’s super cool. And we use that with Amazon Lambda. So because it’s application embedded, it doesn’t matter where you deploy it, right? This is not Kubernetes only, this is not VPCs and Amazon only. It is literally anywhere it’s application embedded. You can run this on your local laptop, and I’m going to run my client on my local laptop here in a minute and when I stop blathering and get to the actual demo, right?

Clint Dovholuk - 28:15 So another killer feature, Openzd, gives you end to end encryption out of the gate. So by adopting Zero Trust into your application and your server, you have true end to end encryption. If you remember, before I talked about Mutual TLS, I talked about the router and the client connecting and forming a Mutual TLS connection. And that’s great between links, but that means when you’re on the router or at the endpoint, you can see whatever the traffic is. With end to end encryption, even if you were sending traffic through that router has no ability to be able to read that traffic. On top of that, even if you had. If you use a secure protocol like SSH on your other side, on both sides, the SSH application protocol is also another layer of encryption.

Clint Dovholuk - 29:08 Being able to obtain secrets and defeat the security of that connection is going to be incredibly difficult. Attackers are going to have such a hard time. It’s definitely not going to be worth it. Because of course, if it’s worth it, maybe people will find a way to get through it. But it’ll take nation state actors in order to go through the amount of effort it’ll take to bypass those three layers of encryption. So don’t forget application layer end to end encryption. Mutual TLS makes it basically unhackable, and I’ll say basically because everybody who is sufficiently motivated, perhaps they’ll find a way. But as we know, of course, there’s not right now. And that uses Libsodium, which is built for small devices. So you’ll see, Openzd oftentimes is targeting IoT type devices because we have a Csdk and the Csdk is tiny.

Clint Dovholuk - 29:58 And Chacha 20 poly 13 Five was built for tiny devices. So even if you have a Raspberry pi or something teeny tiny, zero trust is not out of reach. Take an SDK, stuff it in your application. Now you have zero trust available to you. Another just amazing superpower, in my opinion, port inference. If you go out to any network and you scan that network, you’re going to find out, oh, Clint’s using MySQL clinch using VNC, right? It doesn’t matter. When you send all of your traffic through the secure zero trust overlay every port becomes 443. So good luck figuring out what services somebody is using just by scanning ports. You’re never going to do it. It’ll just be all 443 or whatever you choose. 443 is just a common one.

Clint Dovholuk - 30:50 So all your traffic pardon me 1 second, all your traffic gets synthesized over a single connection through the Openzd overlay, which doesn’t impact performance. So don’t think of it like that. It’s a big pipe. It’s as big as your network speed is. So all of your protocols travel over that pipe. And so they all look like port 443, whether it’s FTP, SQL or SSH or whatever I mentioned before. Continuous authorization. You get that for free with Openzd as well. Domain checking, Mac address checking, process checking. Also, ZD supports two factor authentication out of the box. So that’s all cool stuff you get. You get a self healing mesh network, which is also really important. So this way, if your device is sending traffic and one of those routers goes down, no problem. Openzd will just route the traffic from one place to the other.

Clint Dovholuk - 31:39 That’s super cool. And then this is what it looks like. This is an example of taking an actual Java application which creates a client and then sends traffic over that client from the before to the after. Key things to pay attention here, let me make it a little bit easier for you to see it. Let me make it even easier for you to see it, right? So you have no need to know the IP address anymore. You don’t need to know its local host or whatever the IP is. You don’t have any need as the developer to know what port it is. You just know you have some super secure service, and your identity is authorized to, in this case, bind to bind that identity. So this is acting as a server. Bind means I accept traffic. Dial means I’ll send traffic.

Clint Dovholuk - 32:29 And so as a developer, you don’t need to know where things are defined or where they’re hosted, et cetera. One of my favorites, maybe my favorite embedded superpower is this thing I refer to as Implicit multifactor authentication. So if you have a strong identity, you can’t even get onto the network. You can’t even send traffic to your API. Now, presumably your API will have some sort of authorization that’s baked into it. So this SDK on the left will talk to this API on the right, and the API will verify the traffic in some way. The network also verifies that identity by having that implicit identity baked into it. So, realistically, you have two factors of authentication already. By just using a zero trust overlay network, you have your strong identity from openzd, and you have your authentication from your API service.

Clint Dovholuk - 33:25 And so those are the selling points about application embedded zero trust. And I’ve blathered long enough. Maybe you just skipped right ahead to the demo. Maybe they put nice little chapter marks in here for you, and you just skipped into the good stuff. So that’s cool. So here’s where we actually see all of this in action. We’re going to put our money where our mouth is. Let’s take a look at what our demo is going to look like. So I’ve already deployed. In fact, I won’t say I did, because I didn’t deploy this network. The Net Foundry console deployed this network on my behalf. We call it the Appetizer Network to get you to want some openzd. And what we have is we have a network deployed, we have a controller, we have some routers in the mix.

Clint Dovholuk - 34:03 And on the left, you’ll see the Reflect client. That’s my client running on my local computer. And then on the right, you’ll see Reflect server, HTP server. And that’s deployed in Amazon, Fargate. So I made an application, I bundled it into a docker container, and then my good friend Mike Guthrie deployed it out onto Fargate. And now we have the appetizer out there running. This is all open source. I don’t think I even mentioned this. All of this is free and open source, right? You can go get all of this for free. Host your own openzd network today. Go get it. And if you want to look at the demo, you’ll go to the Openzd test kitchen. ZD zero trust. ZT. ZD. Italian pasta. So our little logo here, this little piece of ZD. So here’s Ziggy back again.

Clint Dovholuk - 34:52 He’s got a chef hat on this time and he’s in our Openzd test kitchen where we basically test things out before we put them into the main repository, which is just Openzd GitHub.com slash Openzd. All right, out there is an appetizer repository and aptly named. We have an appetizer Openzd IO pause right now because you can go there right now. And presumably we haven’t taken it down whenever you watch this video, who knows? But if you’re doing it right now, you’ll probably have this available to you and you can see I’ve gone there appetizer Openc IO. If you enter your email address, which is what I’d ask you to do because that’s a good, nice, unique name, then fantastic. We’ll get your email.

Clint Dovholuk - 35:34 Hopefully we won’t solicit you, but maybe we but you can add yourself to my Openzd overlay network and if you don’t want to be bothered with it, you can just click the don’t bother with me, don’t bother me right now. But if you want to let us know that you care about Openzd, you can go ahead and put an email in there. Once you’ve done that and click the button, you’re going to see, it brings you to a page that looks like this. And then you’ll see run some of the sample programs. You’ll see the git clone, you’ll see where the repository is and then it’ll give you the little kind of overview that I just showed you and tell you some samples to run. Fantastic. That’s what we’re going to see because I’m actually going to just go and do it.

Clint Dovholuk - 36:11 So let’s just go and do it, shall we? First thing I need to do is I need to bring up Microsoft Edge, a browser which I literally never use and with a bunch of junk on it. So let’s go to Appetizer Openzd IO and you’ll see I get the same thing that I showed you before. I’m going to enter my I’m not gonna put my email address, but it’s Clint@openzed.org. If you want to email me, you can. But I’m not going to type it in here because who knows who’s going to look at it. So we’ll just use Clint and let’s do Open SEC Summit. Let’s do that. That sounds great. That’s probably pretty unique. And add me to openzd. All right, so now once I’ve added myself to no, I can’t click on that.

Clint Dovholuk - 36:58 Apparently once I’ve added myself to Openzd, then I can go and I can clone the repository. Now I’ve already cloned the repository, but you can just go ahead and copy that if you want to. And then I need to download my token. So remember I talked about enrolling an identity. The identity enrollment starts with a signed document, which in this case is a JWT, and so it says, where would I like to save it? I’m going to save this thing into my appetizer. GitHub openzdest kitchen. Where is it? It’s in here somewhere. Work GitHub openzd. Okay, let’s go back up and find it. Where is my test kitchen? That oh, it’s because it’s on Linux. That’s right. I am running this inside of Linux, which is why I don’t remember where it is.

Clint Dovholuk - 37:58 And let’s see, my username is going to be in home and CD and Git, where is Git? In here, and here’s GitHub. And here’s opencd test kitchen. And here’s the appetizer. So I’m going to save this there. Now when I saved it there, I can then just go run the application and I’ve already CD into that location. So if I bring up my terminal and I PWD, you’ll see I’m in the appetizer repository, I can go ahead and just run that command that I copied and pasted. Now when I do this, what’s going to happen? What’s going to happen is the application is going to enroll that token, which you can see it happened and that’s going to connect to a server. I’m going to bring up the diagram. Actually, I want to bring up the prettier diagram if I can find it.

Clint Dovholuk - 38:50 Let me find it real quick, this one. So it’s going to contact a service that has no listening ports that I stood up that will do one thing and it will reflect back to me or echo whatever I type. So if I say hi, then it’ll say hi and it’ll say you sent me hi. Now, what also happens that you’re not seeing this is normal stuff. I challenge you to find this service on the open Internet. You won’t find it at appetizer openZ IO because it’s all application embedded. So this has no listening ports anywhere. But you can try to find it, you won’t. But what happened here is I’ve sent a message up into the cloud, in this case an Amazon Fargate where this thing is running.

Clint Dovholuk - 39:41 And I have a server that’s running up there accepting connections on the overlay network, not on the IP based underlay network. And it’s returned back to me, this text. And so that’s what it’s done. Also what it’s done is it’s looked at my input and decided if the input was profanity. So I added a Go library to check for profanity. And then we did something even cooler. We have another service that can deploy it out there. In fact, I don’t think I was sharing my screen, so I’m sorry about that. We have another service that Ken deployed and that service is doing language model work. It’s a text classifier. So AI being all the rage, right? Everybody needs an AI service.

Clint Dovholuk - 40:27 We have a service that looks at my input and classifies it as to whether it’s offensive, I don’t know where Ken deployed his service. Think about microservices, right? Thinking about the what you do all day long. You got Team A that does one thing. Team B, that does another thing. You need team A to send traffic to team B. Team A has to say well, how do I whitelist my IP? Here’s my IP. Blah. Right. None of that with openzd. Ken stood up a service, told me the name of the service and told me go dial it and that’s all I had to do. And so that language model is running out there somewhere and I don’t know where Ken put it.

Clint Dovholuk - 41:02 So if I do something like say I hate babies, then you’re going to see, hey, your message seems to be offensive and we are not going to relay it. Also, if you will notice in our oh, Clint is a dingleberry. Somebody is being a funny dude right now. That’s really interesting. Who hates babies? I wouldn’t have expected that to kit through our language classifier. But you can see somebody else is actually using this right now. I don’t know who. This honestly wasn’t the plant. Hopefully it’s somebody on the chat, I don’t know. So if I then say I like babies, you can see what I didn’t show you before is that the service also reflects the message back into your screen right here. And if you wanted to have somebody else play along, you could just send them to messages. Messages. M-E-S-S-A-G-E-S.

Clint Dovholuk - 42:00 HTML spell it wrong. Too many s’s, maybe too many S’s, I think. And they can also get the messages too. Like example. So that is application embedded zero trust working. We have sent message from my computer here in western New York up into the cloud to a server that has no listening ports. That server sent traffic to another server that was doing language model identification using Python. Oh, I should mention, I wrote it in go. Ken wrote it in Python. You can find that also on the test kitchen GitHub account too. And then it returned a response and I printed it out here and you can see the latency that incurs. Here is an example. If I hit enter, you can see basically instantaneous pops up back on my screen. So common question is what kind of latency can you expect?

Clint Dovholuk - 42:54 Humans never notice it. Now, I also did one more thing using application embedded zero trust. I actually created a little bot in my application and every time a message is sent to our demo application here, it actually pops up here in mattermost and it tells me I hate babies. If were to go look at this one, if we look at them both at the same time and see which one gets the message and which one doesn’t get the message, you’ll see that the message that’s reflected publicly doesn’t come through. But the message that might be offensive is put in our chat message here. And so I can then decide is this actually offensive? Is it not offensive?

Clint Dovholuk - 43:39 And maybe we could doesn’t do this right yet because I just made this demo literally today, but maybe someday that yes button will train our language model so it can learn that this is offensive or that’s not offensive, and that’s all entirely secure. I don’t know where Ken deployed it, so just think about multi cloud solutions now, right? If he deployed it in Azure, I would have no clue that I deployed it in Azure while mine runs in Amazon. And I think that’s kind of mind blowing. So hopefully you think that’s mind blowing, too. Now, that’s my demo. I hope you enjoyed it. You can go out to the test kitchen. Like I said, let’s skip past all of this. That was my demo. Opencd has all kinds of other Zdefied, as we call it, apps.

Clint Dovholuk - 44:20 We have ZSS, which is neat because you can SSH here to a place without having a port open, right? You can deny all the firewall rules like Ken did when Ken deployed that service. Ken used Openzd to basically be the bastion, if you would, the entry point to that server. He can only access his server via Openzd. We use SCP as part of the ZSS package. Some examples we have mattermost. You saw me doing mattermost. That mattermost server that I just it’s a chat app, which is equivalent to Slack or Discord or something like that. But self hostable. That Mattermost server that you just saw me accessing that I was messaging, can only be messaged via Openzd because it’s protected by Openzd.

Clint Dovholuk - 45:11 But it’s a good example of the tunneler based approach and an Embedding not Embedding joining an application embedded zero trust approach with a Brownfield existing app. We send messages from GitHub all the time to my Mattermost. So when I get a notification oh, there’s Ken right there. When I get a notification that a commit has happened or that Clint approved a message, I’ll get a notification from GitHub in my Mattermost. But the only way to hit that Mattermost service is via Openzd. So there’s a Zdefied webhook for that. Zdbc. Actually, Marcos has talked at a security summit in the past about me. If you haven’t seen that one, go check it out. Really good one. Great. Marcos, great job presenting works with all your databases out there because it’s not bespoke per database. So it works with Postgres, Oracle, you name it, right?

Clint Dovholuk - 46:10 It’s just about poking the JDBC driver in the right way. Have good blog posts about it. Cubezetle helm. I’m going to just skip all through these because you can go out there and check them out on your own. Prometheus actually one of my favorites here. I have a complicated Prometheus example where we have a Prometheus server that’s scraping Kubernetes over the Openzd overlay, where the Kubernetes control plane is entirely off the overlay before it had to be on the Open Internet. And you have a Kubernetes ingress after you have kubernetes API. Totally hidden, totally private kubernetes. And yet Prometheus can still scrape it from anywhere. Cool blog post. If you’re interested, let us know. Zdefi is also neat. Very important. Every country, I believe, not just the United States, is now issuing executive orders that secure infrastructure.

Clint Dovholuk - 47:03 Infrastructure must be secured by a Zero Trust mechanism of some flavor, obviously why it’s relevant in today’s world. It’s mentioned eleven times in that document. Here are the Pillars of Zero. Trust. And you’re thinking to yourself, clint, what a great job. What a great demo. Blew my mind. Put this in a cart. Let me bring it home. How do I go get it? Well, like I said before, we are out on GitHub. You can go to oh my goodness. Shame on me. Openzd GitHub. IO is no longer the URL. It is currently openzd. IO is the URL. So I’ll have to go and fix that. That’s where you’ll find our documentation and GitHub. We’re on GitHub at Slash Openzd so you can find us on all our socials here. You can follow me and Ken weekly. We’ll do a ZDTV out on YouTube.

Clint Dovholuk - 47:53 We have a Twitter handle while Twitter is still around. And look at that. Slide is so old. Still got the old bird. Who wants to go to X, right? Nobody wants to go to X. We have a discourse group where you can hit us up in discourse, where you can ask questions and whatnot. And if I can ask one thing of you, it’s to go to our GitHub repository openzd, and give us that star. That star does mean a lot to us. It helps other people know that ZD is a cool project and people should check it out. If you look at our star progression here, you can see we’re picking up some steam. So hopefully you all who watch this can bump this up way higher. Let’s see another big spike. And yeah, give us that star. And that’s the whole thing.

Clint Dovholuk - 48:35 That’s application embedded. Zero Trust. I hope you’ve enjoyed it. Hope the presentation was exciting. And if you have any questions, hit us up on those socials. All right? Cheers.