A new way to run Kubernetes: Meet CIQ Stratum Pro
Running Kubernetes usually means assembling about ten components, dreading every upgrade, and switching tools to see more than one cluster. CIQ Stratum Pro puts the whole lifecycle in one binary on the Rocky Linux you already run. Chris Wolford, CIQ's Director of Engineering, takes a CIQ Stratum Kubernetes Engine (SKE) cluster from a single download to a high availability install, joins it to the Stratum Pro fleet manager, puts an AI agent on it through the built-in Model Context Protocol (MCP) server, and upgrades it to the next Kubernetes version while it keeps running. Thirty minutes, one cluster, start to finish. Bring questions to the chat.
What you'll see
- Install a high availability cluster from one binary on Rocky Linux.
- Join that cluster to the Stratum Pro fleet manager and work it from the web console.
- Connect an AI agent through the MCP server, read-only until you say otherwise.
Speakers
![]()
Chris Wolford, Director of Engineering, CIQ
![]()
Brian Rieb, Sr. Technical Product Marketing Manager, CIQ
Transcript
Brian Rieb: Hi everyone, I'm Brian Reid from sick. Thanks for joining today. Today is the first public walkthrough of Secu Stratum Pro. It's built on our open source Kubernetes engine. Secu Stratum Kubernetes Engine or SQ and Pro is the commercial version on top of it. We'll be showing aspects of each of these today. If you've run Kubernetes yourself, you've probably assembled it from a pile of parts. Stress. Stress a little through a component upgrade, juggled the rest of the tools that you just have to put in. Just anything that you need to do to see just one one more cluster at a time. Stratum Pro is built to take that work off of your plate.
Before I hand it over, a little housekeeping and one ask. Don't try to save any questions for the end. Chris and I are both in the chat the whole time, and we'll answer any questions as they come in. If something on the screen makes you think, wait, how does that work? That's the moment to ask it in chat. Also, if you're watching the recording on demand later, leave your question in the comment sections and we promise we'll get back with you. The recording goes up on the same page you register from, and also on our YouTube channel at the same time. Now walking walking you through this all today is Chris Woolford, our Director of Engineering.
Chris leads the team that built stratum. So you're hearing it from the person who knows where every piece lives. Chris, thanks for being here. And let me hand this over to you right now.
Chris Wolford: Sounds good. Thank you sir. Hello everyone. So today I wanted to step you through a brief PowerPoint that goes over the architecture and how we built a Kubernetes engine, and then follow that up with a demo going through the command lines needed to actually build and do all the fun, exciting things. So without further ado, I'll get cracking. The beauty of stratum is that everything for Stratum Kubernetes Kubernetes Engine is internal to the binary. It is one single binary. To install your Kubernetes cluster with no registry access required. That is the purest forest simplistic method that we went in building stratum with the stratum Kubernetes engine and the single binary.
View full transcriptHide full transcript
As I mentioned, it's one binary. So there's one artifact for you to copy to your your notes that you're going to build. It's very conducive to being air gapped by default, because everything is in the binary that you need to spin up your cluster. Additionally, the components that make up stratum are all built from source. So I've seen Kubernetes at CD container D. All of those fundamental building blocks that make up stratum are built from source and incorporated into the single binary and inside of the binary. All those different versions of the microservices that make up the binary are pinned so that every time you spin up a stratum engine, you have the same consistent behavior across all of the nodes that you deploy to.
So there's no drift that occurs because of this. And also with the Pro version, there's a size profile that you can enable, and we'll get into more of that later on. But the important part that everybody is interested in most is the architecture. So the architecture of stratum itself is the single binary, but within it it's a Linux binary. And then it also has all the payloads for all the different components that you bring in and all the OCI images that you have. There are some opinions in here that are really grounded in the way that the folks at Secu have built clusters over and over again, as well as a lot of the industry standards that people have used time and time again in their deployment.
So we have incorporated those microservices into stratum itself, such as Etsy and the scheduler and the VIP in cilium for your network or core DNS for your DNS needs. All those things are within within the binary itself. Again, for simplicity, the goal here is to make it simple and easy to deploy and maintain with that as well. When we're talking about an Ha environment, we utilize ship to manage the, the cluster, the the control plane nodes themselves. And that's what communicates across through all the workers. Then we talk about Pro here towards the later of the demo. The things above stratum Kubernetes Engine in Stratum Pro are really two main things.
The fleet manager, which allows you to manage multiple Kubernetes clusters at once, and then a dashboard single sign on experience that's that's vital to enterprises that use the environment. A part of the core things that are baked into the binary, into stratum are things like rolling, upgrades, rollbacks, all using the single binary itself. Additionally, snapshots and restoring those snapshots of CD lifecycle management of your certificates for the control plane itself. And then, of course, as I mentioned, this is for the pro environment. All Cooper Day two operational thing is that as an administrator myself, I find incredibly valuable when I start trying to manage a multitude of clusters.
And as time has progressed, what are you managing? How does that work? So having these all baked in into the binary is quite nice in that regard. And also, one of the biggest selling points for stratum itself is the web UI. So a part of the binary is the web UI. So the moment you deploy a cluster and do stratum up, as I mentioned on the first slide, a part of that is you get the web UI, which I'll showcase here in just a little bit. The overall layout is here, and everything within the web UI you can do also a lot from the CLI, a lot of it's API baked, all of it is Kubernetes API.
So a lot of familiar things. If you've used other experiences, you'll feel very at home in a lot of ways. But the user experience here inside the UI, there's several different features that I particularly love, such as the host shell. So you can jump into any one of your nodes, or look at the logs or jump into the exact experience of a pod are two of my favorite things that exist within the environment, but a multitude of different features within them, and happy to to dive into those more just looking over. So we have a couple of different screenshots of the user experience for the Stratum Kubernetes Engine, web UI, everything from the overview, health to nodes, outputs, and different workloads that exist on the environment.
And then for Pro, we also have an app catalog that features a different variety of home charts in its current state that you can go ahead and put a single click of a button install. So next up, before I jump into the demo side, I just wanted to delay the found one work for what I have today and what I'm going to show. So the current environment that I'm going to showcase is made up of eight different VMs that exist on my network here. So it's it's on prem, the demo East cluster is going to be what we're going to be building out. It's the H, a cluster with the worker node that we're going to do all the commands needed to actually facilitate that and build it out.
The fleet manager is Stratum Pro. So I also have a series of notes that run Stratum Pro as the Ha environment of it as well. And we're going to join the Demo East cluster that we built out to stratum Pros and Environment. Again, these are all VMs that run Rocky Linux 9.8 SELinux is enabled. The VMs themselves have four CPU cores. It gives a memory and a little bit of disk base to go along the hold. The the binary and some other fun stuff. Also, we have a mic with stratum, so I'll step through that a little bit and all the fun things that you can do with the MCP server that we have on the system.
Now that we've been able to go through stratum engine architecture, let's go through how to install and set up an A cluster on my system here I have a couple of nodes that I outlined previously. The CP one through three are going to be our Ha nodes, while the eight the W one nodes going to be our worker on the CP one nodes. Just to show that stratum is on the box. Here's the versioning information for the system. The versioning info starts with the Kubernetes version plus stratum release version, and then the architecture on the system. And then underneath it you can see all the different pin version of services that make up Stratum Kubernetes Engine.
So with that in mind, I'll clear the screen and we'll begin setting up the cluster. The first step is going to be to get on this node, taking that single binary and making it run as system D. So I've got this series of commands here that do just that. They take this single binary and they set it up to run as a system D, and then wait for the, the node to come online. So what happens? There is a series of different things. Chiefly among them is really just spinning it up to get to the point where it's ready. So the single binary does everything. In this case it loads cilium and all the components at Cty.
All the good stuff that I outlined in the architecture note. And it gets to the point when cilium is fully loaded to become to the ready state. So as soon as we get to that point, we'll see the the cube cuddle git nodes show up. Roughly speaking, this takes on average about 40s, give or take for it to fully complete and get online. And then this is just the first node that we'll do, and then join all the other nodes from here. Also, while we wait just to recap the, the lab environment here. So each one of the VMAs is a four core box with eight gigs of memory and about 40 gigs of disk space that can be used.
Which just be about their. And there you have it, the node is ready and we can continue on with adder adding the other two nodes. So the next step in the process is to go ahead and create a token for the control plane, clear my screen and issue the command to create the token. Now it's simple. It's from the single binary stratum token crate with the control plane flag and the cube VIP address that's outlined as the server, and how long you want this token to exist for. So now that we have that, I'm going to go ahead and copy the token. And where's my other window.
So I have it for future copies. Okay. Now I got that copied. We'll go ahead and copy over the needed key files to each one of the servers that we'll be using. So the two VMs, CP two and CP three are the ones that I'm referencing. So from the control plane or the key exists, I'll go ahead and copy it to keep two. And then I'll do CP three while I'm at it. And inside tress ends engine 13 instead of 12. So now we have both of those copied over. And we can continue on with the process. So jumping over now to the node I want to add.
So the commands itself are rather straightforward. Again we want to make this run as system D instead of just having it run in the shell. So the series of commands are grabbing the token and hash, injecting that in and then starting it up, and you can see the configuration information and the join information that's used in this line here. Then we'll go ahead and kick that off. There you go. Symlink is up. And just like the first node it takes a good 30 40s for the second node to join the first instance and get fully online. Again, it's doing all the same things and talking with control D to make sure it starts to begin quorum.
Now that we've issued the symlink and all the commands, we should be able to jump over to CPU ones and see the system. Come on. And there you go. The second node is on the line, so we'll do the same exact thing this time on three. So CP three ssh into that box and run the same exact commands to join this this VM to the high cluster as well. And again just like the CP two, this will take several seconds for it to boot itself up and get itself started. Get cilium running and then perform the join steps that are necessary to add it to the cluster.
So if we jump back over to keep one to see the status. Same commands as before, just kubectl get nodes and in a few minutes we should see it start to come online. And there you go. Not ready. We'll give it a just a few more seconds and wait for it to come up. And there you have it. Third nodes on. So the next step will be to add a worker node to the cluster. To do that, we need to create a new token for just the work out. Let me go ahead and clear my screen. And unlike the control plane this one's just the worker node.
So in the previous command that was similar it would have a dash dash control plane. In this case since it's just a worker node it's just dash dash issue that and just like the other command we'll get the output of the command needed. So I'll go ahead and grab the token and the session or excuse me and the hatch. Now we want to jump over to worker one the worker node and go ahead and copy all the steps required. So again just like the other commands slightly different because of the different variable there. We'll go ahead and run this as well as system D. And then go back to our control plane one and run the cube cuddle command again.
Again. This should take just a few seconds to go ahead and shoot. All right. Check again for the node B online. And there you go. No it is ready. So we have a three node a cluster and one worker node set and ready to go. Now that we have a Ha cluster full fully online and going, let's take a look at the web UI. That's a part of stratum engine that comes as a part of the single binary. So to go ahead and get to that, I'm just going to look at it from CP one. So to get the token and be able to verify that in fact the dashboard is up and running, I have a series of two different commands.
The first one is just to check the the dashboard is up and running as part of the deployment, and the second one will output the needed token that I can use for the web UI. Okay. So as you can see, this is the token that I'll use. And then the node port that's expose. All right. Jumping over to the web UI we have the Stratham login when you first log in. So we're going to go ahead and take the token that we got from the previous command in the terminal window and log in. And when you first log in you get to the overview dashboard page.
That shows you a highlight of everything that's going on in the cluster, from the node utilization that you have for the nodes that exist. A couple of events of all the recent things that have happened, and then their readiness state of the nodes and the overall utilization of both memory and CPU usage and a simple node count. Scrolling down a little bit, you see some more information about the control plane, as well as the cluster itself and its versioning information that is running on the system. Up at the top you have the namespaces. So this is a global filter. Here are all the namespaces that exist.
So anything that you change here is going to filter the menu options and the results on the left hand side. So I'm going to go ahead and leave it as all namespaces for now. But to look at it you see all the pods that exist, the overview of the workflows, the nodes themselves, the different details. When you click on any one of these items, I'm going to click on note here one. For example, I can see the information about the the node, specifically the system information as well as its readiness state, any any pods that are running on that specific node. I think in time, I can stretch this out and make this larger to see if I wanted to as well.
And I can jump to a different node and updates the details pane accordingly so you can see what's going on there. Additionally, I can cordon or encoding a node directly from the web UI. I can grab a host shell session of it as well, and interact with the node itself. And all very, very useful things from an administrator standpoint. And jumping back to the pods, that same sort of behavior I can do at a pod level. So for this specific one, let's say I want to look at the logs for it. I can make the logs window smaller or larger by clicking and dragging on it.
I can filter it for any, any sort of context that's in the log that I'm interested in seeing. I can also jump to an exact window for it as well. So just like the node shell, you can see what's going on there and be able to play and interact with the the pod itself. Really, really useful when you're trying to troubleshoot and look at different things going on inside of your cluster. Perhaps one of my favorite things about the entire user experience. And with that as well, consistency is always a helpful thing. So the layout of it, the details, pain is always going to be on the right hand side that you can drag in and look out, and that's across the different things that we have in all the different pages that exist.
And also all of these things that are on the right hand are excluding the left hand side, are all your common elements that you would find in managing a cluster. So extremely useful, extremely powerful in managing the cluster and seeing what all I can do. And also for those that love the real estate, when you're starting to look at the details of the pods and want to see the logs, you can make the menu a lot smaller than what is coming out. So with that, that's the highlight and overview of the open source web UI. Right now that we've been able to go through the web UI of the open source strategy engine, let's go ahead and add that cluster to the fleet manager, which is a part of Stratum Pro.
So here before me is the main Stratum Pro login. So I'm going to go ahead and use the SSL authentication method to log into the system. There you have it logged in to Stratum Pro. Much like the open source version of the system. You will first log in, you get to the dashboard, and then all the left hand menu options are completely identical with the addition of a few slide things. Mainly above them is the clusters option. So here is where you'll be able to add additional clusters to the system, as well as remove clusters as needed. So I'm going to go ahead and work on adding our Stratum Kubernetes engine cluster that we went ahead and build out just a few minutes ago to this cluster now.
So you do so by clicking Register Existing. And then go ahead and give your cluster a name. I'm going to give this one demo east and how long I want the token to exist for. In this case I'm just going to let it run for an hour and you'll get an output of the curl command that you'll need to run on your, s t cluster. So I've copied that. So I'm going to jump back over to my terminal session. And clear what I had on there before, and then paste the needed join steps. So in this case of the on this specific node, cube control is not installed on it.
It's a part of stratum. So I'm going to load it in that way. And it should expand. And it will go ahead and do all the steps needed to join this cluster to the control plane of the other system. And switching back to the Pro UI. You can see that now we have the Demo East cluster as a cluster option inside of the user interface. And to simply switch the the UI, the context, all you have to do is click on the use, and it completely switches the entire context of the application to that specific cluster. So now you can see that, for example, we have the worker one node.
And we can do all the same things that we did in the S key environment including clicking on the host shell, looking and seeing what's going on with a specific host, just like we did in the previous versions of the run that we showed, including seeing all the pods and all the glorious things that exist inside of the UI. So you just have the clusters that you can switch context to and manage your multiple different clusters from one user interface with the Pro you just, which is quite nice. The other thing that's different about the Pro UI is we have the application catalog. So with the application catalog you can go in and really install several different components that are in the application catalog.
These are the standard ones that we have listed. Now this list will continue to grow. You can also add your own application catalog but just using a different JSON file. But to show the installation process of how it goes and how this works on the back end is utilizing come charts to go ahead and do the deployment. So in this example we're going to go ahead and install pod info on replica set and kick off the install. And you'll kill a notification that's going. You can see it's installing it's it's processed there. And once it's done you'll get an indication that it's installed and good to go.
And then you could go and look at the pods and everything else that's running. That's a part of this specific deployment. And they get really complex and they can get really simple. This is just a nice, quick, simple one that will go ahead and install everything as needed. So that is the application catalog. All right. Now the next thing I'd like to showcase is the MCP server that we have. So let's go ahead and switch back to our terminal prompt and looking at the first server. So I'll go ahead and and connect to my session that I have here. So this is just my local cloud code session that I have on my system.
And as you can see I have a mic server set up for my demo cluster that currently has 18 tools. So by default the MCP server just has those basic read only tools that that really lets you interact with the cluster. So let's take a look at some of those. For example, I want to see what's the status of the notes in the cluster. And you can see there that it's used two different tools from the Stratham demo. And then there's the output of the nodes in the cluster. Similarly, if we want to check the pods running in the cluster, we can do the same thing.
And then we'll go out and use them c p server to talk to stratum in that way. And again this is just read only access. So you'll be able to do the basic things that you would normally do with a cluster. Nothing destructive. So you can't delete a pod or upgrade a cluster or take out a node or a node, any of those sorts of things. For that activity, you would need a higher level of permissions. So with that, it's a dash dash destructive at that and allows you to do that. So if I look at the CP server list that I have, you're going to go ahead and disabled the read only one that I have and enabled the read write one I have.
So this will allow me to go ahead and do a little bit more intricate things, a lot more exciting things depending on which way you look at it. So in this case I want to try upgrading the cluster. So. Here is the prompt that I'm going to use to to upgrade our 1.36 cluster to 1.37. And for this, as you can see, I've staged already the upgraded binaries to this location and the upgrade pass of every note. So I'll go ahead and let cloud work with MCP server and upgrade the entire cluster. So as it's doing this thing, it will take obviously a few minutes to go ahead and perform that upgrade.
One of the things that it does is grab a CD snapshot of all the nodes before it proceeds with going ahead and doing the rolling upgrade of all the clusters. So I'll give it some time and we'll let it do is think, and I'll be back in just a few minutes. All right. Now the cluster is finished upgrading. And you can see that we now officially have 137.1 on all of our nodes in the cluster. So if we jump over to the web UI, we can take a look at that for the cluster as well, and see that the corresponding notes show the same exact same there.
All right. Now that we've been able to upgrade the cluster, I also wanted to showcase being able to roll back a specific node. I'll was still using the single binary. So let me go ahead and switch back to my terminal prompt. And I'm going to downgrade the worker one. And it's again a relatively straightforward process of just running the rollback command option from the stratum single binary. You can also, like the other commands, do a dry run and see exactly what it's going to go ahead and do as a part of the rollback process. So here's all the different stages that it will go ahead and proceed with.
I'm going to go ahead and clear the screen this time. And we will go ahead and run the actual rollback command on just the specific node. All right. There you have it. It's finished doing all of its steps that it needed to. So let's jump over to keep one and take a look at what stratum has to say about the nodes now. And as you can see it is officially downgraded. So as simple as that. As far as downloading downgrading the cluster nodes. All right. Now that it's rolled back I'm going to go ahead and restore it to 137. So from the worker one node I'm going to jump back to it and go ahead and run the single binary of the 1.37 version, targeting the 1.37 version of Cooper, and go ahead and upgrade the cluster node specifically, specifically the worker node.
Excuse me. So we'll go ahead and do its thing. All right. Now that the upgrade is finished for worker node one, let's go ahead and take a look at it with AB. Let's go ahead and look at it with the cube control. Get notes. And as you can see it's also running 1.37.1 as the rest of the cluster. All right. For the final thing, we've talked a lot about the different actions you can do with the single binary. So let me just go through the overview real quick. And again this is all the main commands that you have with stratum itself. And several of these we've actually run today, everything from upgrading the cluster itself with the MCP server and looking at the dashboard and getting into the getting into the token.
And there's also a couple other fun and interesting things you can do, such as looking at the snapshots that exist on a specific note. So looking at CP one, we can see the snapshot from earlier. And the another example that you can do is we've looked at a lot of cube control commands, but not a lot of Etsi de communes. So just to show that that actually is is running inside of the single binary as well. Let's look at the the health. So there you can see it returns proper health of the active. And then another fun one is just the the token list. And as you can see there's no join tokens that exist on on the node.
So this is just a brief overview of all the different commands that exists. Feel free to download and take a look at all the different options. And with that I'll turn it back over to Brian. Thank you everyone and I hope you all have enjoyed this demo walkthrough.
Brian Rieb: Thanks, Chris for taking the time today and showing us Secu stratum. We're going to hang around in chat for a couple of minutes live, but if you want any more information or if you want to see it sick stratum running in your own environment, please reach out to us and schedule an evaluation with our engineers by visiting our website at six. Also, if you're headed to KubeCon in Salt Lake City soon, come find us at the Secu booth. We'd love to meet you and talk with you further. With that said, thank you everyone for attending and thank you for your time.
Built for scale. Chosen by the world’s best.
2.75M+
Rocky Linux instances
Being used world wide
90%
Of fortune 100 companies
Use CIQ supported technologies
250k
Avg. monthly downloads
Rocky Linux
9
Enterprise products
Spanning the kernel to the orchestrator
Have questions about your infrastructure?
Talk to a CIQ engineer about Rocky Linux, HPC, and AI infrastructure.


