>>all right, We're five minutes after the hour. That's all aboard. Who's coming aboard? Welcome everyone to the tutorial track for our launchpad of them. So for the next couple of hours, we've got a SYRIZA videos and experts on hand to answer questions about our new product, Doctor Enterprise Container Cloud. Before we jump into the videos and the technology, I just want to introduce myself and my other emcee for the session. I'm Bill Milks. I run curriculum development for Mirant us on. And >>I'm Bruce Basil Matthews. I'm the Western regional Solutions architect for Moran Tissue esa and welcome to everyone to this lovely launchpad oven event. >>We're lucky to have you with us proof. At least somebody on the call knows something about your enterprise Computer club. Um, speaking of people that know about Dr Enterprise Container Cloud, make sure that you've got a window open to the chat for this session. We've got a number of our engineers available and on hand to answer your questions live as we go through these videos and disgusting problem. So that's us, I guess, for Dr Enterprise Container Cloud, this is Mirant asses brand new product for bootstrapping Doctor Enterprise Kubernetes clusters at scale Anything. The airport Abu's? >>No, just that I think that we're trying Thio. Uh, let's see. Hold on. I think that we're trying Teoh give you a foundation against which to give this stuff a go yourself. And that's really the key to this thing is to provide some, you know, many training and education in a very condensed period. So, >>yeah, that's exactly what you're going to see. The SYRIZA videos we have today. We're going to focus on your first steps with Dr Enterprise Container Cloud from installing it to bootstrapping your regional child clusters so that by the end of the tutorial content today, you're gonna be prepared to spin up your first documentary prize clusters using documented prize container class. So just a little bit of logistics for the session. We're going to run through these tutorials twice. We're gonna do one run through starting seven minutes ago up until I guess it will be ten fifteen Pacific time. Then we're gonna run through the whole thing again. So if you've got other colleagues that weren't able to join right at the top of the hour and would like to jump in from the beginning, ten. Fifteen Pacific time. We're gonna do the whole thing over again. So if you want to see the videos twice, you got public friends and colleagues that, you know you wanna pull in for a second chance to see this stuff, we're gonna do it all. All twice. Yeah, this session. Any any logistics I should add, Bruce that No, >>I think that's that's pretty much what we had to nail down here. But let's zoom dash into those, uh, feature films. >>Let's do Edmonds. And like I said, don't be shy. Feel free to ask questions in the chat or engineers and boosting myself are standing by to answer your questions. So let me just tee up the first video here and walk their cost. Yeah. Mhm. Yes. Sorry. And here we go. So our first video here is gonna be about installing the Doctor Enterprise Container Club Management cluster. So I like to think of the management cluster as like your mothership, right? This is what you're gonna use to deploy all those little child clusters that you're gonna use is like, Come on it as clusters downstream. So the management costs was always our first step. Let's jump in there >>now. We have to give this brief little pause >>with no good day video. Focus for this demo will be the initial bootstrap of the management cluster in the first regional clusters to support AWS deployments. The management cluster provides the core functionality, including identity management, authentication, infantry release version. The regional cluster provides the specific architecture provided in this case, eight of us and the Elsie um, components on the UCP Cluster Child cluster is the cluster or clusters being deployed and managed. The deployment is broken up into five phases. The first phase is preparing a big strap note on this dependencies on handling with download of the bridge struck tools. The second phase is obtaining America's license file. Third phase. Prepare the AWS credentials instead of the adduce environment. The fourth configuring the deployment, defining things like the machine types on the fifth phase. Run the bootstrap script and wait for the deployment to complete. Okay, so here we're sitting up the strap node, just checking that it's clean and clear and ready to go there. No credentials already set up on that particular note. Now we're just checking through AWS to make sure that the account we want to use we have the correct credentials on the correct roles set up and validating that there are no instances currently set up in easy to instance, not completely necessary, but just helps keep things clean and tidy when I am perspective. Right. So next step, we're just going to check that we can, from the bootstrap note, reach more antis, get to the repositories where the various components of the system are available. They're good. No areas here. Yeah, right now we're going to start sitting at the bootstrap note itself. So we're downloading the cars release, get get cars, script, and then next, we're going to run it. I'm in. Deploy it. Changing into that big struck folder. Just making see what's there. Right now we have no license file, so we're gonna get the license filed. Oh, okay. Get the license file through the more antis downloads site, signing up here, downloading that license file and putting it into the Carisbrook struck folder. Okay, Once we've done that, we can now go ahead with the rest of the deployment. See that the follow is there. Uh, huh? That's again checking that we can now reach E C two, which is extremely important for the deployment. Just validation steps as we move through the process. All right, The next big step is valid in all of our AWS credentials. So the first thing is, we need those route credentials which we're going to export on the command line. This is to create the necessary bootstrap user on AWS credentials for the completion off the deployment we're now running an AWS policy create. So it is part of that is creating our Food trucks script, creating the mystery policy files on top of AWS, Just generally preparing the environment using a cloud formation script you'll see in a second will give a new policy confirmations just waiting for it to complete. Yeah, and there is done. It's gonna have a look at the AWS console. You can see that we're creative completed. Now we can go and get the credentials that we created Today I am console. Go to that new user that's being created. We'll go to the section on security credentials and creating new keys. Download that information media Access key I D and the secret access key. We went, Yeah, usually then exported on the command line. Okay. Couple of things to Notre. Ensure that you're using the correct AWS region on ensure that in the conflict file you put the correct Am I in for that region? I'm sure you have it together in a second. Yes. Okay, that's the key. Secret X key. Right on. Let's kick it off. Yeah, So this process takes between thirty and forty five minutes. Handles all the AWS dependencies for you, and as we go through, the process will show you how you can track it. Andi will start to see things like the running instances being created on the west side. The first phase off this whole process happening in the background is the creation of a local kind based bootstrapped cluster on the bootstrap node that clusters then used to deploy and manage all the various instances and configurations within AWS. At the end of the process, that cluster is copied into the new cluster on AWS and then shut down that local cluster essentially moving itself over. Okay. Local clusters boat just waiting for the various objects to get ready. Standard communities objects here Okay, so we speed up this process a little bit just for demonstration purposes. Yeah. There we go. So first note is being built the best in host. Just jump box that will allow us access to the entire environment. Yeah, In a few seconds, we'll see those instances here in the US console on the right. Um, the failures that you're seeing around failed to get the I. P for Bastian is just the weight state while we wait for a W s to create the instance. Okay. Yes. Here, beauty there. Okay. Mhm. Okay. Yeah, yeah. Okay. On there. We got question. Host has been built on three instances for the management clusters have now been created. We're going through the process of preparing. Those nodes were now copying everything over. See that? The scaling up of controllers in the big Strap cluster? It's indicating that we're starting all of the controllers in the new question. Almost there. Yeah. Yeah, just waiting for key. Clark. Uh huh. Start to finish up. Yeah. No. What? Now we're shutting down control this on the local bootstrap node on preparing our I. D. C. Configuration. Fourth indication, soon as this is completed. Last phase will be to deploy stack light into the new cluster the last time Monitoring tool set way Go stack like to plan It has started. Mhm coming to the end of the deployment Mountain. Yeah, America. Final phase of the deployment. Onda, We are done. Okay, You'll see. At the end they're providing us the details of you. I log in so there's a keeper clogging. You can modify that initial default password is part of the configuration set up with one documentation way. Go Councils up way can log in. Yeah, yeah, thank you very much for watching. >>Excellent. So in that video are wonderful field CTO Shauna Vera bootstrapped up management costume for Dr Enterprise Container Cloud Bruce, where exactly does that leave us? So now we've got this management costume installed like what's next? >>So primarily the foundation for being able to deploy either regional clusters that will then allow you to support child clusters. Uh, comes into play the next piece of what we're going to show, I think with Sean O'Mara doing this is the child cluster capability, which allows you to then deploy your application services on the local cluster. That's being managed by the ah ah management cluster that we just created with the bootstrap. >>Right? So this cluster isn't yet for workloads. This is just for bootstrapping up the downstream clusters. Those or what we're gonna use for workings. >>Exactly. Yeah. And I just wanted to point out, since Sean O'Mara isn't around, toe, actually answer questions. I could listen to that guy. Read the phone book, and it would be interesting, but anyway, you can tell him I said that >>he's watching right now, Crusoe. Good. Um, cool. So and just to make sure I understood what Sean was describing their that bootstrap er knows that you, like, ran document fresh pretender Cloud from to begin with. That's actually creating a kind kubernetes deployment kubernetes and Docker deployment locally. That then hits the AWS a p i in this example that make those e c two instances, and it makes like a three manager kubernetes cluster there, and then it, like, copies itself over toe those communities managers. >>Yeah, and and that's sort of where the transition happens. You can actually see it. The output that when it says I'm pivoting, I'm pivoting from my local kind deployment of cluster AP, I toothy, uh, cluster, that's that's being created inside of AWS or, quite frankly, inside of open stack or inside of bare metal or inside of it. The targeting is, uh, abstracted. Yeah, but >>those air three environments that we're looking at right now, right? Us bare metal in open staff environments. So does that kind cluster on the bootstrap er go away afterwards. You don't need that afterwards. Yeah, that is just temporary. To get things bootstrapped, then you manage things from management cluster on aws in this example? >>Yeah. Yeah. The seed, uh, cloud that post the bootstrap is not required anymore. And there's no, uh, interplay between them after that. So that there's no dependencies on any of the clouds that get created thereafter. >>Yeah, that actually reminds me of how we bootstrapped doctor enterprise back in the day, be a temporary container that would bootstrap all the other containers. Go away. It's, uh, so sort of a similar, similar temporary transient bootstrapping model. Cool. Excellent. What will convict there? It looked like there wasn't a ton, right? It looked like you had to, like, set up some AWS parameters like credentials and region and stuff like that. But other than that, that looked like heavily script herbal like there wasn't a ton of point and click there. >>Yeah, very much so. It's pretty straightforward from a bootstrapping standpoint, The config file that that's generated the template is fairly straightforward and targeted towards of a small medium or large, um, deployment. And by editing that single file and then gathering license file and all of the things that Sean went through, um, that that it makes it fairly easy to script >>this. And if I understood correctly as well that three manager footprint for your management cluster, that's the minimum, right. We always insist on high availability for this management cluster because boy do not wanna see oh, >>right, right. And you know, there's all kinds of persistent data that needs to be available, regardless of whether one of the notes goes down or not. So we're taking care of all of that for you behind the scenes without you having toe worry about it as a developer. >>No, I think there's that's a theme that I think will come back to throughout the rest of this tutorial session today is there's a lot of there's a lot of expertise baked him to Dr Enterprise Container Cloud in terms of implementing best practices for you like the defaulter, just the best practices of how you should be managing these clusters, Miss Seymour. Examples of that is the day goes on. Any interesting questions you want to call out from the chap who's >>well, there was. Yeah, yeah, there was one that we had responded to earlier about the fact that it's a management cluster that then conduce oh, either the the regional cluster or a local child molester. The child clusters, in each case host the application services, >>right? So at this point, we've got, in some sense, like the simplest architectures for our documentary prize Container Cloud. We've got the management cluster, and we're gonna go straight with child cluster. In the next video, there's a more sophisticated architecture, which will also proper today that inserts another layer between those two regional clusters. If you need to manage regions like across a BS, reads across with these documents anything, >>yeah, that that local support for the child cluster makes it a lot easier for you to manage the individual clusters themselves and to take advantage of our observation. I'll support systems a stack light and things like that for each one of clusters locally, as opposed to having to centralize thumb >>eso. It's a couple of good questions. In the chat here, someone was asking for the instructions to do this themselves. I strongly encourage you to do so. That should be in the docks, which I think Dale helpfully thank you. Dale provided links for that's all publicly available right now. So just head on in, head on into the docks like the Dale provided here. You can follow this example yourself. All you need is a Mirante license for this and your AWS credentials. There was a question from many a hear about deploying this toe azure. Not at G. Not at this time. >>Yeah, although that is coming. That's going to be in a very near term release. >>I didn't wanna make promises for product, but I'm not too surprised that she's gonna be targeted. Very bracing. Cool. Okay. Any other thoughts on this one does. >>No, just that the fact that we're running through these individual pieces of the steps Well, I'm sure help you folks. If you go to the link that, uh, the gentleman had put into the chat, um, giving you the step by staff. Um, it makes it fairly straightforward to try this yourselves. >>E strongly encourage that, right? That's when you really start to internalize this stuff. OK, but before we move on to the next video, let's just make sure everyone has a clear picture in your mind of, like, where we are in the life cycle here creating this management cluster. Just stop me if I'm wrong. Who's creating this management cluster is like, you do that once, right? That's when your first setting up your doctor enterprise container cloud environment of system. What we're going to start seeing next is creating child clusters and this is what you're gonna be doing over and over and over again. When you need to create a cluster for this Deb team or, you know, this other team river it is that needs commodity. Doctor Enterprise clusters create these easy on half will. So this was once to set up Dr Enterprise Container Cloud Child clusters, which we're going to see next. We're gonna do over and over and over again. So let's go to that video and see just how straightforward it is to spin up a doctor enterprise cluster for work clothes as a child cluster. Undocumented brands contain >>Hello. In this demo, we will cover the deployment experience of creating a new child cluster, the scaling of the cluster and how to update the cluster. When a new version is available, we begin the process by logging onto the you I as a normal user called Mary. Let's go through the navigation of the U I so you can switch. Project Mary only has access to development. Get a list of the available projects that you have access to. What clusters have been deployed at the moment there. Nan Yes, this H Keys Associate ID for Mary into her team on the cloud credentials that allow you to create access the various clouds that you can deploy clusters to finally different releases that are available to us. We can switch from dark mode to light mode, depending on your preferences, Right? Let's now set up semester search keys for Mary so she can access the notes and machines again. Very simply, had Mississippi key give it a name, we copy and paste our public key into the upload key block. Or we can upload the key if we have the file available on our local machine. A simple process. So to create a new cluster, we define the cluster ad management nodes and add worker nodes to the cluster. Yeah, again, very simply, you go to the clusters tab. We hit the create cluster button. Give the cluster name. Yeah, Andi, select the provider. We only have access to AWS in this particular deployment, so we'll stick to AWS. What's like the region in this case? US West one release version five point seven is the current release Onda Attach. Mary's Key is necessary Key. We can then check the rest of the settings, confirming the provider Any kubernetes c r D r I p address information. We can change this. Should we wish to? We'll leave it default for now on. Then what components? A stack light I would like to deploy into my Custer. For this. I'm enabling stack light on logging on Aiken. Sit up the retention sizes Attention times on. Even at this stage, at any customer alerts for the watchdogs. E consider email alerting which I will need my smart host details and authentication details. Andi Slack Alerts. Now I'm defining the cluster. All that's happened is the cluster's been defined. I now need to add machines to that cluster. I'll begin by clicking the create machine button within the cluster definition. Oh, select manager, Select the number of machines. Three is the minimum. Select the instant size that I'd like to use from AWS and very importantly, ensure correct. Use the correct Am I for the region. I commend side on the route device size. There we go, my three machines obviously creating. I now need to add some workers to this custom. So I go through the same process this time once again, just selecting worker. I'll just add to once again, the AM is extremely important. Will fail if we don't pick the right, Am I for a boon to machine in this case and the deployment has started. We can go and check on the bold status are going back to the clusters screen on clicking on the little three dots on the right. We get the cluster info and the events, so the basic cluster info you'll see pending their listen cluster is still in the process of being built. We kick on, the events will get a list of actions that have been completed This part of the set up of the cluster. So you can see here we've created the VPC. We've created the sub nets on We've created the Internet gateway. It's unnecessary made of us and we have no warnings of the stage. Yeah, this will then run for a while. We have one minute past waken click through. We can check the status of the machine bulls as individuals so we can check the machine info, details of the machines that we've assigned, right? Mhm Onda. See any events pertaining to the machine areas like this one on normal? Yeah. Just watch asked. The community's components are waiting for the machines to start. Go back to Custer's. Okay, right. Because we're moving ahead now. We can see we have it in progress. Five minutes in new Matt Gateway on the stage. The machines have been built on assigned. I pick up the U. S. Thank you. Yeah. There we go. Machine has been created. See the event detail and the AWS. I'd for that machine. Mhm. No speeding things up a little bit. This whole process and to end takes about fifteen minutes. Run the clock forward, you'll notice is the machines continue to bold the in progress. We'll go from in progress to ready. A soon as we got ready on all three machines, the managers on both workers way could go on and we could see that now we reached the point where the cluster itself is being configured. Mhm, mhm. And then we go. Cluster has been deployed. So once the classes deployed, we can now never get around our environment. Okay, Are cooking into configure cluster We could modify their cluster. We could get the end points for alert alert manager on See here The griffon occupying and Prometheus are still building in the background but the cluster is available on you would be able to put workloads on it the stretch to download the cube conflict so that I can put workloads on it. It's again three little dots in the right for that particular cluster. If the download cube conflict give it my password, I now have the Q conflict file necessary so that I can access that cluster Mhm all right Now that the build is fully completed, we can check out cluster info on. We can see that Allow the satellite components have been built. All the storage is there, and we have access to the CPU. I So if we click into the cluster, we can access the UCP dashboard, right? Shit. Click the signing with Detroit button to use the SSO on. We give Mary's possible to use the name once again. Thing is, an unlicensed cluster way could license at this point. Or just skip it on. There. We have the UCP dashboard. You can see that has been up for a little while. We have some data on the dashboard going back to the console. We can now go to the griffon, a data just being automatically pre configured for us. We can switch and utilized a number of different dashboards that have already been instrumented within the cluster. So, for example, communities cluster information, the name spaces, deployments, nodes. Mhm. So we look at nodes. If we could get a view of the resource is utilization of Mrs Custer is very little running in it. Yeah. General dashboard of Cuba navies cluster one of this is configurable. You can modify these for your own needs, or add your own dashboards on de scoped to the cluster. So it is available to all users who have access to this specific cluster, all right to scale the cluster on to add a notice. A simple is the process of adding a mode to the cluster, assuming we've done that in the first place. So we go to the cluster, go into the details for the cluster we select, create machine. Once again, we need to be ensure that we put the correct am I in and any other functions we like. You can create different sized machines so it could be a larger node. Could be bigger disks and you'll see that worker has been added from the provisioning state on shortly. We will see the detail off that worker as a complete to remove a note from a cluster. Once again, we're going to the cluster. We select the node would like to remove. Okay, I just hit delete On that note. Worker nodes will be removed from the cluster using according and drawing method to ensure that your workouts are not affected. Updating a cluster. When an update is available in the menu for that particular cluster, the update button will become available. And it's a simple as clicking the button, validating which release you would like to update to. In this case, the next available releases five point seven point one. Here I'm kicking the update by in the background We will coordinate. Drain each node slowly go through the process of updating it. Andi update will complete depending on what the update is as quickly as possible. Girl, we go. The notes being rebuilt in this case impacted the manager node. So one of the manager nodes is in the process of being rebuilt. In fact, to in this case, one has completed already on In a few minutes we'll see that there are great has been completed. There we go. Great. Done. Yeah. If you work loads of both using proper cloud native community standards, there will be no impact. >>Excellent. So at this point, we've now got a cluster ready to start taking our communities of workloads. He started playing or APs to that costume. So watching that video, the thing that jumped out to me at first Waas like the inputs that go into defining this workload cost of it. All right, so we have to make sure we were using on appropriate am I for that kind of defines the substrate about what we're gonna be deploying our cluster on top of. But there's very little requirements. A so far as I could tell on top of that, am I? Because Docker enterprise Container Cloud is gonna bootstrap all the components that you need. That s all we have is kind of kind of really simple bunch box that we were deploying these things on top of so one thing that didn't get dug into too much in the video. But it's just sort of implied. Bruce, maybe you can comment on this is that release that Shawn had to choose for his, uh, for his cluster in creating it. And that release was also the thing we had to touch. Wanted to upgrade part cluster. So you have really sharp eyes. You could see at the end there that when you're doing the release upgrade enlisted out a stack of components docker, engine, kubernetes, calico, aled, different bits and pieces that go into, uh, go into one of these commodity clusters that deploy. And so, as far as I can tell in that case, that's what we mean by a release. In this sense, right? It's the validated stack off container ization and orchestration components that you know we've tested out and make sure it works well, introduction environments. >>Yeah, and and And that's really the focus of our effort is to ensure that any CVS in any of the stack are taken care of that there is a fixes air documented and up streamed to the open stack community source community, um, and and that, you know, then we test for the scaling ability and the reliability in high availability configuration for the clusters themselves. The hosts of your containers. Right. And I think one of the key, uh, you know, benefits that we provide is that ability to let you know, online, high. We've got an update for you, and it's fixes something that maybe you had asked us to fix. Uh, that all comes to you online as your managing your clusters, so you don't have to think about it. It just comes as part of the product. >>You just have to click on Yes. Please give me that update. Uh, not just the individual components, but again. It's that it's that validated stack, right? Not just, you know, component X, y and Z work. But they all work together effectively Scalable security, reliably cool. Um, yeah. So at that point, once we started creating that workload child cluster, of course, we bootstrapped good old universal control plane. Doctor Enterprise. On top of that, Sean had the classic comment there, you know? Yeah. Yeah. You'll see a little warnings and errors or whatever. When you're setting up, UCP don't handle, right, Just let it do its job, and it will converge all its components, you know, after just just a minute or two. But we saw in that video, we sped things up a little bit there just we didn't wait for, you know, progress fighters to complete. But really, in real life, that whole process is that anything so spend up one of those one of those fosters so quite quite quick. >>Yeah, and and I think the the thoroughness with which it goes through its process and re tries and re tries, uh, as you know, and it was evident when we went through the initial ah video of the bootstrapping as well that the processes themselves are self healing, as they are going through. So they will try and retry and wait for the event to complete properly on. And once it's completed properly, then it will go to the next step. >>Absolutely. And the worst thing you could do is panic at the first warning and start tearing things that don't don't do that. Just don't let it let it heal. Let take care of itself. And that's the beauty of these manage solutions is that they bake in a lot of subject matter expertise, right? The decisions that are getting made by those containers is they're bootstrapping themselves, reflect the expertise of the Mirant ISS crew that has been developing this content in these two is free for years and years now, over recognizing humanities. One cool thing there that I really appreciate it actually that it adds on top of Dr Enterprise is that automatic griffon a deployment as well. So, Dr Enterprises, I think everyone knows has had, like, some very high level of statistics baked into its dashboard for years and years now. But you know our customers always wanted a double click on that right to be able to go a little bit deeper. And Griffon are really addresses that it's built in dashboards. That's what's really nice to see. >>Yeah, uh, and all of the alerts and, uh, data are actually captured in a Prometheus database underlying that you have access to so that you are allowed to add new alerts that then go out to touch slack and say hi, You need to watch your disk space on this machine or those kinds of things. Um, and and this is especially helpful for folks who you know, want to manage the application service layer but don't necessarily want to manage the operations side of the house. So it gives them a tool set that they can easily say here, Can you watch these for us? And Miran tas can actually help do that with you, So >>yeah, yeah, I mean, that's just another example of baking in that expert knowledge, right? So you can leverage that without tons and tons of a long ah, long runway of learning about how to do that sort of thing. Just get out of the box right away. There was the other thing, actually, that you could sleep by really quickly if you weren't paying close attention. But Sean mentioned it on the video. And that was how When you use dark enterprise container cloud to scale your cluster, particularly pulling a worker out, it doesn't just like Territo worker down and forget about it. Right? Is using good communities best practices to cordon and drain the No. So you aren't gonna disrupt your workloads? You're going to just have a bunch of containers instantly. Excellent crash. You could really carefully manage the migration of workloads off that cluster has baked right in tow. How? How? Document? The brass container cloud is his handling cluster scale. >>Right? And And the kubernetes, uh, scaling methodology is is he adhered to with all of the proper techniques that ensure that it will tell you. Wait, you've got a container that actually needs three, uh, three, uh, instances of itself. And you don't want to take that out, because that node, it means you'll only be able to have to. And we can't do that. We can't allow that. >>Okay, Very cool. Further thoughts on this video. So should we go to the questions. >>Let's let's go to the questions >>that people have. Uh, there's one good one here, down near the bottom regarding whether an a p I is available to do this. So in all these demos were clicking through this web. You I Yes, this is all a p. I driven. You could do all of this. You know, automate all this away is part of the CSC change. Absolutely. Um, that's kind of the point, right? We want you to be ableto spin up. Come on. I keep calling them commodity clusters. What I mean by that is clusters that you can create and throw away. You know, easily and automatically. So everything you see in these demos eyes exposed to FBI? >>Yeah. In addition, through the standard Cube cuddle, Uh, cli as well. So if you're not a programmer, but you still want to do some scripting Thio, you know, set up things and deploy your applications and things. You can use this standard tool sets that are available to accomplish that. >>There is a good question on scale here. So, like, just how many clusters and what sort of scale of deployments come this kind of support our engineers report back here that we've done in practice up to a Zeman ia's like two hundred clusters. We've deployed on this with two hundred fifty nodes in a cluster. So were, you know, like like I said, hundreds, hundreds of notes, hundreds of clusters managed by documented press container fall and then those downstream clusters, of course, subject to the usual constraints for kubernetes, right? Like default constraints with something like one hundred pods for no or something like that. There's a few different limitations of how many pods you can run on a given cluster that comes to us not from Dr Enterprise Container Cloud, but just from the underlying kubernetes distribution. >>Yeah, E. I mean, I don't think that we constrain any of the capabilities that are available in the, uh, infrastructure deliveries, uh, service within the goober Netease framework. So were, you know, But we are, uh, adhering to the standards that we would want to set to make sure that we're not overloading a node or those kinds of things, >>right. Absolutely cool. Alright. So at this point, we've got kind of a two layered our protection when we are management cluster, but we deployed in the first video. Then we use that to deploy one child clustering work, classroom, uh, for more sophisticated deployments where we might want to manage child clusters across multiple regions. We're gonna add another layer into our architectural we're gonna add in regional cluster management. So this idea you're gonna have the single management cluster that we started within the first video. On the next video, we're gonna learn how to spin up a regional clusters, each one of which would manage, for example, a different AWS uh, US region. So let me just pull out the video for that bill. We'll check it out for me. Mhm. >>Hello. In this demo, we will cover the deployment of additional regional management. Cluster will include a brief architectures of you how to set up the management environment, prepare for the deployment deployment overview and then just to prove it, to play a regional child cluster. So, looking at the overall architecture, the management cluster provides all the core functionality, including identity management, authentication, inventory and release version. ING Regional Cluster provides the specific architecture provider in this case AWS on the LCN components on the D you speak Cluster for child cluster is the cluster or clusters being deployed and managed? Okay, so why do you need a regional cluster? Different platform architectures, for example aws who have been stack even bare metal to simplify connectivity across multiple regions handle complexities like VPNs or one way connectivity through firewalls, but also help clarify availability zones. Yeah. Here we have a view of the regional cluster and how it connects to the management cluster on their components, including items like the LCN cluster Manager we also Machine Manager were held. Mandel are managed as well as the actual provider logic. Mhm. Okay, we'll begin by logging on Is the default administrative user writer. Okay, once we're in there, we'll have a look at the available clusters making sure we switch to the default project which contains the administration clusters. Here we can see the cars management cluster, which is the master controller. And you see, it only has three nodes, three managers, no workers. Okay, if we look at another regional cluster similar to what we're going to deploy now, also only has three managers once again, no workers. But as a comparison, here's a child cluster This one has three managers, but also has additional workers associate it to the cluster. All right, we need to connect. Tell bootstrap note. Preferably the same note that used to create the original management plaster. It's just on AWS, but I still want to machine. All right. A few things we have to do to make sure the environment is ready. First thing we're going to see go into route. We'll go into our releases folder where we have the kozberg struck on. This was the original bootstrap used to build the original management cluster. Yeah, we're going to double check to make sure our cube con figures there once again, the one created after the original customers created just double check. That cute conflict is the correct one. Does point to the management cluster. We're just checking to make sure that we can reach the images that everything is working. A condom. No damages waken access to a swell. Yeah. Next we're gonna edit the machine definitions. What we're doing here is ensuring that for this cluster we have the right machine definitions, including items like the am I. So that's found under the templates AWS directory. We don't need to edit anything else here. But we could change items like the size of the machines attempts. We want to use that The key items to ensure where you changed the am I reference for the junta image is the one for the region in this case AWS region for utilizing this was no construct deployment. We have to make sure we're pointing in the correct open stack images. Yeah, okay. Set the correct and my save file. Now we need to get up credentials again. When we originally created the bootstrap cluster, we got credentials from eight of the U. S. If we hadn't done this, we would need to go through the u A. W s set up. So we're just exporting the AWS access key and I d. What's important is CAAs aws enabled equals. True. Now we're sitting the region for the new regional cluster. In this case, it's Frankfurt on exporting our cube conflict that we want to use for the management cluster. When we looked at earlier Yeah, now we're exporting that. Want to call the cluster region Is Frank Foods Socrates Frankfurt yet trying to use something descriptive It's easy to identify. Yeah, and then after this, we'll just run the bootstrap script, which will complete the deployment for us. Bootstrap of the regional cluster is quite a bit quicker than the initial management clusters. There are fewer components to be deployed. Um, but to make it watchable, we've spent it up. So we're preparing our bootstrap cluster on the local bootstrap node. Almost ready on. We started preparing the instances at W s and waiting for that bastard and no to get started. Please. The best you nerd Onda. We're also starting to build the actual management machines they're now provisioning on. We've reached the point where they're actually starting to deploy. Dr. Enterprise, this is probably the longest face. Yeah, seeing the second that all the nerds will go from the player deployed. Prepare, prepare. Yeah, You'll see their status changes updates. He was the first night ready. Second, just applying second already. Both my time. No waiting from home control. Let's become ready. Removing cluster the management cluster from the bootstrap instance into the new cluster running the date of the U. S. All my stay. Ah, now we're playing Stockland. Switch over is done on. Done. Now I will build a child cluster in the new region very, very quickly to find the cluster will pick. Our new credential has shown up. We'll just call it Frankfurt for simplicity a key and customs to find. That's the machine. That cluster stop with three managers. Set the correct Am I for the region? Yeah, Do the same to add workers. There we go test the building. Yeah. Total bill of time Should be about fifteen minutes. Concedes in progress. It's going to expect this up a little bit. Check the events. We've created all the dependencies, machine instances, machines, a boat shortly. We should have a working cluster in Frankfurt region. Now almost a one note is ready from management. Two in progress. Yeah, on we're done. Clusters up and running. Yeah. >>Excellent. So at this point, we've now got that three tier structure that we talked about before the video. We got that management cluster that we do strapped in the first video. Now we have in this example to different regional clustering one in Frankfurt, one of one management was two different aws regions. And sitting on that you can do Strap up all those Doctor enterprise costumes that we want for our work clothes. >>Yeah, that's the key to this is to be able to have co resident with your actual application service enabled clusters the management co resident with it so that you can, you know, quickly access that he observation Elson Surfboard services like the graph, Ana and that sort of thing for your particular region. A supposed to having to lug back into the home. What did you call it when we started >>the mothership? >>The mothership. Right. So we don't have to go back to the mother ship. We could get >>it locally. Yeah, when, like to that point of aggregating things under a single pane of glass? That's one thing that again kind of sailed by in the demo really quickly. But you'll notice all your different clusters were on that same cluster. Your pain on your doctor Enterprise Container Cloud management. Uh, court. Right. So both your child clusters for running workload and your regional clusters for bootstrapping. Those child clusters were all listed in the same place there. So it's just one pane of glass to go look for, for all of your clusters, >>right? And, uh, this is kind of an important point. I was, I was realizing, as we were going through this. All of the mechanics are actually identical between the bootstrapped cluster of the original services and the bootstrapped cluster of the regional services. It's the management layer of everything so that you only have managers, you don't have workers and that at the child cluster layer below the regional or the management cluster itself, that's where you have the worker nodes. And those are the ones that host the application services in that three tiered architecture that we've now defined >>and another, you know, detail for those that have sharp eyes. In that video, you'll notice when deploying a child clusters. There's not on Lee. A minimum of three managers for high availability management cluster. You must have at least two workers that's just required for workload failure. It's one of those down get out of work. They could potentially step in there, so your minimum foot point one of these child clusters is fine. Violence and scalable, obviously, from a >>That's right. >>Let's take a quick peek of the questions here, see if there's anything we want to call out, then we move on to our last want to my last video. There's another question here about, like where these clusters can live. So again, I know these examples are very aws heavy. Honestly, it's just easy to set up down on the other us. We could do things on bare metal and, uh, open stack departments on Prem. That's what all of this still works in exactly the same way. >>Yeah, the, uh, key to this, especially for the the, uh, child clusters, is the provision hers? Right? See you establish on AWS provision or you establish a bare metal provision or you establish a open stack provision. Or and eventually that list will include all of the other major players in the cloud arena. But you, by selecting the provision or within your management interface, that's where you decide where it's going to be hosted, where the child cluster is to be hosted. >>Speaking off all through a child clusters. Let's jump into our last video in the Siri's, where we'll see how to spin up a child cluster on bare metal. >>Hello. This demo will cover the process of defining bare metal hosts and then review the steps of defining and deploying a bare metal based doctor enterprise cluster. So why bare metal? Firstly, it eliminates hyper visor overhead with performance boost of up to thirty percent. Provides direct access to GP use, prioritize for high performance wear clothes like machine learning and AI, and supports high performance workloads like network functions, virtualization. It also provides a focus on on Prem workloads, simplifying and ensuring we don't need to create the complexity of adding another opera visor. Lay it between so continue on the theme Why Communities and bare metal again Hyper visor overhead. Well, no virtualization overhead. Direct access to hardware items like F p G A s G p us. We can be much more specific about resource is required on the nodes. No need to cater for additional overhead. Uh, we can handle utilization in the scheduling. Better Onda we increase the performances and simplicity of the entire environment as we don't need another virtualization layer. Yeah, In this section will define the BM hosts will create a new project will add the bare metal hosts, including the host name. I put my credentials I pay my address the Mac address on then provide a machine type label to determine what type of machine it is for later use. Okay, let's get started. So well again. Was the operator thing. We'll go and we'll create a project for our machines to be a member off helps with scoping for later on for security. I begin the process of adding machines to that project. Yeah. So the first thing we had to be in post, Yeah, many of the machine A name. Anything you want, que experimental zero one. Provide the IAP my user name type my password. Okay. On the Mac address for the common interface with the boot interface and then the i p m I i p address These machines will be at the time storage worker manager. He's a manager. Yeah, we're gonna add a number of other machines on will. Speed this up just so you could see what the process looks like in the future. Better discovery will be added to the product. Okay. Okay. Getting back there we have it are Six machines have been added, are busy being inspected, being added to the system. Let's have a look at the details of a single note. Yeah, you can see information on the set up of the node. Its capabilities? Yeah. As well as the inventory information about that particular machine. I see. Okay, let's go and create the cluster. Yeah, So we're going to deploy a bare metal child cluster. The process we're going to go through is pretty much the same as any other child cluster. So we'll credit custom. We'll give it a name, but if it were selecting bare metal on the region, we're going to select the version we want to apply. No way. We're going to add this search keys. If we hope we're going to give the load. Balancer host I p that we'd like to use out of dress range on update the address range that we want to use for the cluster. Check that the sea ideal blocks for the Cuban ladies and tunnels are what we want them to be. Enable disabled stack light. Yeah, and soothe stack light settings to find the cluster. And then, as for any other machine, we need to add machines to the cluster. Here. We're focused on building communities clusters, so we're gonna put the count of machines. You want managers? We're gonna pick the label type manager and create three machines is the manager for the Cuban eighties. Casting Okay thing. We're having workers to the same. It's a process. Just making sure that the worker label host level are I'm sorry. On when Wait for the machines to deploy. Let's go through the process of putting the operating system on the notes validating and operating system deploying doctor identifies Make sure that the cluster is up and running and ready to go. Okay, let's review the bold events waken See the machine info now populated with more information about the specifics of things like storage and of course, details of a cluster etcetera. Yeah, yeah, well, now watch the machines go through the various stages from prepared to deploy on what's the cluster build? And that brings us to the end of this particular demo. You can see the process is identical to that of building a normal child cluster we got our complaint is complete. >>All right, so there we have it, deploying a cluster to bare metal. Much the same is how we did for AWS. I guess maybe the biggest different stepwise there is there is that registration face first, right? So rather than just using AWS financials toe magically create PM's in the cloud. You got a point out all your bare metal servers to Dr Enterprise between the cloud and they really come in, I guess three profiles, right? You got your manager profile with a profile storage profile which has been labeled as allocate. Um, crossword cluster has appropriate, >>right? And And I think that the you know, the key differentiator here is that you have more physical control over what, uh, attributes that love your cat, by the way, uh, where you have the different attributes of a server of physical server. So you can, uh, ensure that the SSD configuration on the storage nodes is gonna be taken advantage of in the best way the GP use on the worker nodes and and that the management layer is going to have sufficient horsepower to, um, spin up to to scale up the the environments, as required. One of the things I wanted to mention, though, um, if I could get this out without the choking much better. Um, is that Ah, hey, mentioned the load balancer and I wanted to make sure in defining the load balancer and the load balancer ranges. Um, that is for the top of the the cluster itself. That's the operations of the management, uh, layer integrating with your systems internally to be able to access the the Cube Can figs. I I p address the, uh, in a centralized way. It's not the load balancer that's working within the kubernetes cluster that you are deploying. That's still cube proxy or service mesh, or however you're intending to do it. So, um, it's kind of an interesting step that your initial step in building this, um and we typically use things like metal L B or in gen X or that kind of thing is to establish that before we deploy this bear mental cluster so that it can ride on top of that for the tips and things. >>Very cool. So any other thoughts on what we've seen so far today? Bruce, we've gone through all the different layers. Doctor enterprise container clouds in these videos from our management are regional to our clusters on aws hand bear amount, Of course, with his dad is still available. Closing thoughts before we take just a very short break and run through these demos again. >>You know, I've been very exciting. Ah, doing the presentation with you. I'm really looking forward to doing it the second time, so that we because we've got a good rhythm going about this kind of thing. So I'm looking forward to doing that. But I think that the key elements of what we're trying to convey to the folks out there in the audience that I hope you've gotten out of it is that will that this is an easy enough process that if you follow the step by steps going through the documentation that's been put out in the chat, um, that you'll be able to give this a go yourself, Um, and you don't have to limit yourself toe having physical hardware on prim to try it. You could do it in a ws as we've shown you today. And if you've got some fancy use cases like, uh, you you need a Hadoop And and, uh, you know, cloud oriented ai stuff that providing a bare metal service helps you to get there very fast. So right. Thank you. It's been a pleasure. >>Yeah, thanks everyone for coming out. So, like I said we're going to take a very short, like, three minute break here. Uh, take the opportunity to let your colleagues know if they were in another session or they didn't quite make it to the beginning of this session. Or if you just want to see these demos again, we're going to kick off this demo. Siri's again in just three minutes at ten. Twenty five a. M. Pacific time where we will see all this great stuff again. Let's take a three minute break. I'll see you all back here in just two minutes now, you know. Okay, folks, that's the end of our extremely short break. We'll give people just maybe, like one more minute to trickle in if folks are interested in coming on in and jumping into our demo. Siri's again. Eso For those of you that are just joining us now I'm Bill Mills. I head up curriculum development for the training team here. Moran Tous on Joining me for this session of demos is Bruce. Don't you go ahead and introduce yourself doors, who is still on break? That's cool. We'll give Bruce a minute or two to get back while everyone else trickles back in. There he is. Hello, Bruce. >>How'd that go for you? Okay, >>Very well. So let's kick off our second session here. I e just interest will feel for you. Thio. Let it run over here. >>Alright. Hi. Bruce Matthews here. I'm the Western Regional Solutions architect for Marantz. Use A I'm the one with the gray hair and the glasses. Uh, the handsome one is Bill. So, uh, Bill, take it away. >>Excellent. So over the next hour or so, we've got a Siris of demos that's gonna walk you through your first steps with Dr Enterprise Container Cloud Doctor Enterprise Container Cloud is, of course, Miranda's brand new offering from bootstrapping kubernetes clusters in AWS bare metal open stack. And for the providers in the very near future. So we we've got, you know, just just over an hour left together on this session, uh, if you joined us at the top of the hour back at nine. A. M. Pacific, we went through these demos once already. Let's do them again for everyone else that was only able to jump in right now. Let's go. Our first video where we're gonna install Dr Enterprise container cloud for the very first time and use it to bootstrap management. Cluster Management Cluster, as I like to describe it, is our mother ship that's going to spin up all the other kubernetes clusters, Doctor Enterprise clusters that we're gonna run our workloads on. So I'm gonna do >>I'm so excited. I can hardly wait. >>Let's do it all right to share my video out here. Yeah, let's do it. >>Good day. The focus for this demo will be the initial bootstrap of the management cluster on the first regional clusters. To support AWS deployments, the management cluster provides the core functionality, including identity management, authentication, infantry release version. The regional cluster provides the specific architecture provided in this case AWS and the Elsom components on the UCP cluster Child cluster is the cluster or clusters being deployed and managed. The deployment is broken up into five phases. The first phase is preparing a bootstrap note on its dependencies on handling the download of the bridge struck tools. The second phase is obtaining America's license file. Third phase. Prepare the AWS credentials instead of the ideas environment, the fourth configuring the deployment, defining things like the machine types on the fifth phase, Run the bootstrap script and wait for the deployment to complete. Okay, so here we're sitting up the strap node. Just checking that it's clean and clear and ready to go there. No credentials already set up on that particular note. Now, we're just checking through aws to make sure that the account we want to use we have the correct credentials on the correct roles set up on validating that there are no instances currently set up in easy to instance, not completely necessary, but just helps keep things clean and tidy when I am perspective. Right. So next step, we're just gonna check that we can from the bootstrap note, reach more antis, get to the repositories where the various components of the system are available. They're good. No areas here. Yeah, right now we're going to start sitting at the bootstrap note itself. So we're downloading the cars release, get get cars, script, and then next we're going to run it. Yeah, I've been deployed changing into that big struck folder, just making see what's there right now we have no license file, so we're gonna get the license filed. Okay? Get the license file through more antis downloads site signing up here, downloading that license file and putting it into the Carisbrook struck folder. Okay, since we've done that, we can now go ahead with the rest of the deployment. Yeah, see what the follow is there? Uh huh. Once again, checking that we can now reach E C two, which is extremely important for the deployment. Just validation steps as we move through the process. Alright. Next big step is violating all of our AWS credentials. So the first thing is, we need those route credentials which we're going to export on the command line. This is to create the necessary bootstrap user on AWS credentials for the completion off the deployment we're now running in AWS policy create. So it is part of that is creating our food trucks script. Creating this through policy files onto the AWS, just generally preparing the environment using a cloud formation script, you'll see in a second, I'll give a new policy confirmations just waiting for it to complete. And there is done. It's gonna have a look at the AWS console. You can see that we're creative completed. Now we can go and get the credentials that we created. Good day. I am console. Go to the new user that's being created. We'll go to the section on security credentials and creating new keys. Download that information media access Key I. D and the secret access key, but usually then exported on the command line. Okay, Couple of things to Notre. Ensure that you're using the correct AWS region on ensure that in the conflict file you put the correct Am I in for that region? I'm sure you have it together in a second. Okay, thanks. Is key. So you could X key Right on. Let's kick it off. So this process takes between thirty and forty five minutes. Handles all the AWS dependencies for you. Um, as we go through, the process will show you how you can track it. Andi will start to see things like the running instances being created on the AWS side. The first phase off this whole process happening in the background is the creation of a local kind based bootstrapped cluster on the bootstrap node that clusters then used to deploy and manage all the various instances and configurations within AWS at the end of the process. That cluster is copied into the new cluster on AWS and then shut down that local cluster essentially moving itself over. Yeah, okay. Local clusters boat. Just waiting for the various objects to get ready. Standard communities objects here. Yeah, you mentioned Yeah. So we've speed up this process a little bit just for demonstration purposes. Okay, there we go. So first note is being built the bastion host just jump box that will allow us access to the entire environment. Yeah, In a few seconds, we'll see those instances here in the US console on the right. Um, the failures that you're seeing around failed to get the I. P for Bastian is just the weight state while we wait for AWS to create the instance. Okay. Yeah. Beauty there. Movies. Okay, sketch. Hello? Yeah, Okay. Okay. On. There we go. Question host has been built on three instances for the management clusters have now been created. Okay, We're going through the process of preparing. Those nodes were now copying everything over. See that scaling up of controllers in the big strapped cluster? It's indicating that we're starting all of the controllers in the new question. Almost there. Right? Okay. Just waiting for key. Clark. Uh huh. So finish up. Yeah. No. Now we're shutting down. Control this on the local bootstrap node on preparing our I. D. C configuration, fourth indication. So once this is completed, the last phase will be to deploy stack light into the new cluster, that glass on monitoring tool set, Then we go stack like deployment has started. Mhm. Coming to the end of the deployment mountain. Yeah, they were cut final phase of the deployment. And we are done. Yeah, you'll see. At the end, they're providing us the details of you. I log in. So there's a key Clark log in. Uh, you can modify that initial default possible is part of the configuration set up where they were in the documentation way. Go Councils up way can log in. Yeah. Yeah. Thank you very much for watching. >>All right, so at this point, what we have we got our management cluster spun up, ready to start creating work clusters. So just a couple of points to clarify there to make sure everyone caught that, uh, as advertised. That's darker. Enterprise container cloud management cluster. That's not rework loans. are gonna go right? That is the tool and you're gonna use to start spinning up downstream commodity documentary prize clusters for bootstrapping record too. >>And the seed host that were, uh, talking about the kind cluster dingy actually doesn't have to exist after the bootstrap succeeds eso It's sort of like, uh, copies head from the seed host Toothy targets in AWS spins it up it then boots the the actual clusters and then it goes away too, because it's no longer necessary >>so that bootstrapping know that there's not really any requirements, Hardly on that, right. It just has to be able to reach aws hit that Hit that a p I to spin up those easy to instances because, as you just said, it's just a kubernetes in docker cluster on that piece. Drop note is just gonna get torn down after the set up finishes on. You no longer need that. Everything you're gonna do, you're gonna drive from the single pane of glass provided to you by your management cluster Doctor enterprise Continue cloud. Another thing that I think is sort of interesting their eyes that the convict is fairly minimal. Really? You just need to provide it like aws regions. Um, am I? And that's what is going to spin up that spending that matter faster. >>Right? There is a mammal file in the bootstrap directory itself, and all of the necessary parameters that you would fill in have default set. But you have the option then of going in and defining a different Am I different for a different region, for example? Oh, are different. Size of instance from AWS. >>One thing that people often ask about is the cluster footprint. And so that example you saw they were spitting up a three manager, um, managing cluster as mandatory, right? No single manager set up at all. We want high availability for doctrine Enterprise Container Cloud management. Like so again, just to make sure everyone sort of on board with the life cycle stage that we're at right now. That's the very first thing you're going to do to set up Dr Enterprise Container Cloud. You're going to do it. Hopefully exactly once. Right now, you've got your management cluster running, and they're gonna use that to spend up all your other work clusters Day today has has needed How do we just have a quick look at the questions and then lets take a look at spinning up some of those child clusters. >>Okay, e think they've actually been answered? >>Yeah, for the most part. One thing I'll point out that came up again in the Dail, helpfully pointed out earlier in surgery, pointed out again, is that if you want to try any of the stuff yourself, it's all of the dogs. And so have a look at the chat. There's a links to instructions, so step by step instructions to do each and every thing we're doing here today yourself. I really encourage you to do that. Taking this out for a drive on your own really helps internalizing communicate these ideas after the after launch pad today, Please give this stuff try on your machines. Okay, So at this point, like I said, we've got our management cluster. We're not gonna run workloads there that we're going to start creating child clusters. That's where all of our work and we're gonna go. That's what we're gonna learn how to do in our next video. Cue that up for us. >>I so love Shawn's voice. >>Wasn't that all day? >>Yeah, I watched him read the phone book. >>All right, here we go. Let's now that we have our management cluster set up, let's create a first child work cluster. >>Hello. In this demo, we will cover the deployment experience of creating a new child cluster the scaling of the cluster on how to update the cluster. When a new version is available, we begin the process by logging onto the you I as a normal user called Mary. Let's go through the navigation of the u I. So you can switch Project Mary only has access to development. Uh huh. Get a list of the available projects that you have access to. What clusters have been deployed at the moment there. Man. Yes, this H keys, Associate ID for Mary into her team on the cloud credentials that allow you to create or access the various clouds that you can deploy clusters to finally different releases that are available to us. We can switch from dark mode to light mode, depending on your preferences. Right. Let's now set up some ssh keys for Mary so she can access the notes and machines again. Very simply, had Mississippi key give it a name. We copy and paste our public key into the upload key block. Or we can upload the key if we have the file available on our machine. A very simple process. So to create a new cluster, we define the cluster ad management nodes and add worker nodes to the cluster. Yeah, again, very simply, we got the clusters tab we had to create cluster button. Give the cluster name. Yeah, Andi, select the provider. We only have access to AWS in this particular deployment, so we'll stick to AWS. What's like the region in this case? US West one released version five point seven is the current release Onda Attach. Mary's Key is necessary key. We can then check the rest of the settings, confirming the provider any kubernetes c r D a r i p address information. We can change this. Should we wish to? We'll leave it default for now and then what components of stack light? I would like to deploy into my custom for this. I'm enabling stack light on logging, and I consider the retention sizes attention times on. Even at this stage, add any custom alerts for the watchdogs. Consider email alerting which I will need my smart host. Details and authentication details. Andi Slack Alerts. Now I'm defining the cluster. All that's happened is the cluster's been defined. I now need to add machines to that cluster. I'll begin by clicking the create machine button within the cluster definition. Oh, select manager, Select the number of machines. Three is the minimum. Select the instant size that I'd like to use from AWS and very importantly, ensure correct. Use the correct Am I for the region. I convinced side on the route. Device size. There we go. My three machines are busy creating. I now need to add some workers to this cluster. So I go through the same process this time once again, just selecting worker. I'll just add to once again the am I is extremely important. Will fail if we don't pick the right. Am I for a Clinton machine? In this case and the deployment has started, we can go and check on the bold status are going back to the clusters screen on clicking on the little three dots on the right. We get the cluster info and the events, so the basic cluster info you'll see pending their listen. Cluster is still in the process of being built. We kick on, the events will get a list of actions that have been completed This part of the set up of the cluster. So you can see here. We've created the VPC. We've created the sub nets on. We've created the Internet Gateway. It's unnecessary made of us. And we have no warnings of the stage. Okay, this will then run for a while. We have one minute past. We can click through. We can check the status of the machine balls as individuals so we can check the machine info, details of the machines that we've assigned mhm and see any events pertaining to the machine areas like this one on normal. Yeah. Just last. The community's components are waiting for the machines to start. Go back to customers. Okay, right. Because we're moving ahead now. We can see we have it in progress. Five minutes in new Matt Gateway. And at this stage, the machines have been built on assigned. I pick up the U S. Yeah, yeah, yeah. There we go. Machine has been created. See the event detail and the AWS. I'd for that machine. No speeding things up a little bit this whole process and to end takes about fifteen minutes. Run the clock forward, you'll notice is the machines continue to bold the in progress. We'll go from in progress to ready. A soon as we got ready on all three machines, the managers on both workers way could go on and we could see that now we reached the point where the cluster itself is being configured mhm and then we go. Cluster has been deployed. So once the classes deployed, we can now never get around. Our environment are looking into configure cluster. We could modify their cluster. We could get the end points for alert Alert Manager See here the griffon occupying and Prometheus are still building in the background but the cluster is available on You would be able to put workloads on it at this stage to download the cube conflict so that I can put workloads on it. It's again the three little dots in the right for that particular cluster. If the download cube conflict give it my password, I now have the Q conflict file necessary so that I can access that cluster. All right, Now that the build is fully completed, we can check out cluster info on. We can see that all the satellite components have been built. All the storage is there, and we have access to the CPU. I. So if we click into the cluster, we can access the UCP dashboard, click the signing with the clock button to use the SSO. We give Mary's possible to use the name once again. Thing is an unlicensed cluster way could license at this point. Or just skip it on. Do we have the UCP dashboard? You could see that has been up for a little while. We have some data on the dashboard going back to the console. We can now go to the griffon. A data just been automatically pre configured for us. We can switch and utilized a number of different dashboards that have already been instrumented within the cluster. So, for example, communities cluster information, the name spaces, deployments, nodes. Um, so we look at nodes. If we could get a view of the resource is utilization of Mrs Custer is very little running in it. Yeah, a general dashboard of Cuba Navies cluster. What If this is configurable, you can modify these for your own needs, or add your own dashboards on de scoped to the cluster. So it is available to all users who have access to this specific cluster. All right to scale the cluster on to add a No. This is simple. Is the process of adding a mode to the cluster, assuming we've done that in the first place. So we go to the cluster, go into the details for the cluster we select, create machine. Once again, we need to be ensure that we put the correct am I in and any other functions we like. You can create different sized machines so it could be a larger node. Could be bigger group disks and you'll see that worker has been added in the provisioning state. On shortly, we will see the detail off that worker as a complete to remove a note from a cluster. Once again, we're going to the cluster. We select the node we would like to remove. Okay, I just hit delete On that note. Worker nodes will be removed from the cluster using according and drawing method to ensure that your workloads are not affected. Updating a cluster. When an update is available in the menu for that particular cluster, the update button will become available. And it's a simple as clicking the button validating which release you would like to update to this case. This available releases five point seven point one give you I'm kicking the update back in the background. We will coordinate. Drain each node slowly, go through the process of updating it. Andi update will complete depending on what the update is as quickly as possible. Who we go. The notes being rebuilt in this case impacted the manager node. So one of the manager nodes is in the process of being rebuilt. In fact, to in this case, one has completed already. Yeah, and in a few minutes, we'll see that the upgrade has been completed. There we go. Great. Done. If you work loads of both using proper cloud native community standards, there will be no impact. >>All right, there. We haven't. We got our first workload cluster spun up and managed by Dr Enterprise Container Cloud. So I I loved Shawn's classic warning there. When you're spinning up an actual doctor enterprise deployment, you see little errors and warnings popping up. Just don't touch it. Just leave it alone and let Dr Enterprises self healing properties take care of all those very transient temporary glitches, resolve themselves and leave you with a functioning workload cluster within victims. >>And now, if you think about it that that video was not very long at all. And that's how long it would take you if someone came into you and said, Hey, can you spend up a kubernetes cluster for development development A. Over here, um, it literally would take you a few minutes to thio Accomplish that. And that was with a W s. Obviously, which is sort of, ah, transient resource in the cloud. But you could do exactly the same thing with resource is on Prem or resource is, um physical resource is and will be going through that later in the process. >>Yeah, absolutely one thing that is present in that demo, but that I like to highlight a little bit more because it just kind of glides by Is this notion of, ah, cluster release? So when Sean was creating that cluster, and also when when he was upgrading that cluster, he had to choose a release. What does that didn't really explain? What does that mean? Well, in Dr Enterprise Container Cloud, we have released numbers that capture the entire staff of container ization tools that will be deploying to that workload costume. So that's your version of kubernetes sed cor DNs calico. Doctor Engineer. All the different bits and pieces that not only work independently but are validated toe work together as a staff appropriate for production, humanities, adopted enterprise environments. >>Yep. From the bottom of the stack to the top, we actually test it for scale. Test it for CVS, test it for all of the various things that would, you know, result in issues with you running the application services. And I've got to tell you from having, you know, managed kubernetes deployments and things like that that if you're the one doing it yourself, it can get rather messy. Eso This makes it easy. >>Bruce, you were staying a second ago. They I'll take you at least fifteen minutes to install your release. Custer. Well, sure, but what would all the other bits and pieces you need toe? Not just It's not just about pressing the button to install it, right? It's making the right decision. About what components work? Well, our best tested toe be successful working together has a staff? Absolutely. We this release mechanism and Dr Enterprise Container Cloud. Let's just kind of package up that expert knowledge and make it available in a really straightforward, fashionable species. Uh, pre Confederate release numbers and Bruce is you're pointing out earlier. He's got delivered to us is updates kind of transparent period. When when? When Sean wanted toe update that cluster, he created little update. Custer Button appeared when an update was available. All you gotta do is click. It tells you what Here's your new stack of communities components. It goes ahead. And the straps those components for you? >>Yeah, it actually even displays at the top of the screen. Ah, little header That says you've got an update available. Do you want me to apply? It s o >>Absolutely. Another couple of cool things. I think that are easy to miss in that demo was I really like the on board Bafana that comes along with this stack. So we've been Prometheus Metrics and Dr Enterprise for years and years now. They're very high level. Maybe in in previous versions of Dr Enterprise having those detailed dashboards that Ravana provides, I think that's a great value out there. People always wanted to be ableto zoom in a little bit on that, uh, on those cluster metrics, you're gonna provides them out of the box for us. Yeah, >>that was Ah, really, uh, you know, the joining of the Miranda's and Dr teams together actually spawned us to be able to take the best of what Morantes had in the open stack environment for monitoring and logging and alerting and to do that integration in in a very short period of time so that now we've got it straight across the board for both the kubernetes world and the open stack world. Using the same tool sets >>warm. One other thing I wanna point out about that demo that I think there was some questions about our last go around was that demo was all about creating a managed workplace cluster. So the doctor enterprise Container Cloud managers were using those aws credentials provisioned it toe actually create new e c two instances installed Docker engine stalled. Doctor Enterprise. Remember all that stuff on top of those fresh new VM created and managed by Dr Enterprise contain the cloud. Nothing unique about that. AWS deployments do that on open staff doing on Parramatta stuff as well. Um, there's another flavor here, though in a way to do this for all of our long time doctor Enterprise customers that have been running Doctor Enterprise for years and years. Now, if you got existing UCP points existing doctor enterprise deployments, you plug those in to Dr Enterprise Container Cloud, uh, and use darker enterprise between the cloud to manage those pre existing Oh, working clusters. You don't always have to be strapping straight from Dr Enterprises. Plug in external clusters is bad. >>Yep, the the Cube config elements of the UCP environment. The bundling capability actually gives us a very straightforward methodology. And there's instructions on our website for exactly how thio, uh, bring in import and you see p cluster. Um so it it makes very convenient for our existing customers to take advantage of this new release. >>Absolutely cool. More thoughts on this wonders if we jump onto the next video. >>I think we should move press on >>time marches on here. So let's Let's carry on. So just to recap where we are right now, first video, we create a management cluster. That's what we're gonna use to create All our downstream were closed clusters, which is what we did in this video. Let's maybe the simplest architectures, because that's doing everything in one region on AWS pretty common use case because we want to be able to spin up workload clusters across many regions. And so to do that, we're gonna add a third layer in between the management and work cluster layers. That's gonna be our regional cluster managers. So this is gonna be, uh, our regional management cluster that exists per region that we're going to manage those regional managers will be than the ones responsible for spending part clusters across all these different regions. Let's see it in action in our next video. >>Hello. In this demo, we will cover the deployment of additional regional management. Cluster will include a brief architectural overview, how to set up the management environment, prepare for the deployment deployment overview, and then just to prove it, to play a regional child cluster. So looking at the overall architecture, the management cluster provides all the core functionality, including identity management, authentication, inventory and release version. ING Regional Cluster provides the specific architecture provider in this case, AWS on the L C M components on the d you speak cluster for child cluster is the cluster or clusters being deployed and managed? Okay, so why do you need original cluster? Different platform architectures, for example AWS open stack, even bare metal to simplify connectivity across multiple regions handle complexities like VPNs or one way connectivity through firewalls, but also help clarify availability zones. Yeah. Here we have a view of the regional cluster and how it connects to the management cluster on their components, including items like the LCN cluster Manager. We also machine manager. We're hell Mandel are managed as well as the actual provider logic. Okay, we'll begin by logging on Is the default administrative user writer. Okay, once we're in there, we'll have a look at the available clusters making sure we switch to the default project which contains the administration clusters. Here we can see the cars management cluster, which is the master controller. When you see it only has three nodes, three managers, no workers. Okay, if we look at another regional cluster, similar to what we're going to deploy now. Also only has three managers once again, no workers. But as a comparison is a child cluster. This one has three managers, but also has additional workers associate it to the cluster. Yeah, all right, we need to connect. Tell bootstrap note, preferably the same note that used to create the original management plaster. It's just on AWS, but I still want to machine Mhm. All right, A few things we have to do to make sure the environment is ready. First thing we're gonna pseudo into route. I mean, we'll go into our releases folder where we have the car's boot strap on. This was the original bootstrap used to build the original management cluster. We're going to double check to make sure our cube con figures there It's again. The one created after the original customers created just double check. That cute conflict is the correct one. Does point to the management cluster. We're just checking to make sure that we can reach the images that everything's working, condone, load our images waken access to a swell. Yeah, Next, we're gonna edit the machine definitions what we're doing here is ensuring that for this cluster we have the right machine definitions, including items like the am I So that's found under the templates AWS directory. We don't need to edit anything else here, but we could change items like the size of the machines attempts we want to use but the key items to ensure where changed the am I reference for the junta image is the one for the region in this case aws region of re utilizing. This was an open stack deployment. We have to make sure we're pointing in the correct open stack images. Yeah, yeah. Okay. Sit the correct Am I save the file? Yeah. We need to get up credentials again. When we originally created the bootstrap cluster, we got credentials made of the U. S. If we hadn't done this, we would need to go through the u A. W s set up. So we just exporting AWS access key and I d. What's important is Kaz aws enabled equals. True. Now we're sitting the region for the new regional cluster. In this case, it's Frankfurt on exporting our Q conflict that we want to use for the management cluster when we looked at earlier. Yeah, now we're exporting that. Want to call? The cluster region is Frankfurt's Socrates Frankfurt yet trying to use something descriptive? It's easy to identify. Yeah, and then after this, we'll just run the bootstrap script, which will complete the deployment for us. Bootstrap of the regional cluster is quite a bit quicker than the initial management clusters. There are fewer components to be deployed, but to make it watchable, we've spent it up. So we're preparing our bootstrap cluster on the local bootstrap node. Almost ready on. We started preparing the instances at us and waiting for the past, you know, to get started. Please the best your node, onda. We're also starting to build the actual management machines they're now provisioning on. We've reached the point where they're actually starting to deploy Dr Enterprise, he says. Probably the longest face we'll see in a second that all the nodes will go from the player deployed. Prepare, prepare Mhm. We'll see. Their status changes updates. It was the first word ready. Second, just applying second. Grady, both my time away from home control that's become ready. Removing cluster the management cluster from the bootstrap instance into the new cluster running a data for us? Yeah, almost a on. Now we're playing Stockland. Thanks. Whichever is done on Done. Now we'll build a child cluster in the new region very, very quickly. Find the cluster will pick our new credential have shown up. We'll just call it Frankfurt for simplicity. A key on customers to find. That's the machine. That cluster stop with three manages set the correct Am I for the region? Yeah, Same to add workers. There we go. That's the building. Yeah. Total bill of time. Should be about fifteen minutes. Concedes in progress. Can we expect this up a little bit? Check the events. We've created all the dependencies, machine instances, machines. A boat? Yeah. Shortly. We should have a working caster in the Frankfurt region. Now almost a one note is ready from management. Two in progress. On we're done. Trust us up and running. >>Excellent. There we have it. We've got our three layered doctor enterprise container cloud structure in place now with our management cluster in which we scrap everything else. Our regional clusters which manage individual aws regions and child clusters sitting over depends. >>Yeah, you can. You know you can actually see in the hierarchy the advantages that that presents for folks who have multiple locations where they'd like a geographic locations where they'd like to distribute their clusters so that you can access them or readily co resident with your development teams. Um and, uh, one of the other things I think that's really unique about it is that we provide that same operational support system capability throughout. So you've got stack light monitoring the stack light that's monitoring the stack light down to the actual child clusters that they have >>all through that single pane of glass that shows you all your different clusters, whether their workload cluster like what the child clusters or usual clusters from managing different regions. Cool. Alright, well, time marches on your folks. We've only got a few minutes left and I got one more video in our last video for the session. We're gonna walk through standing up a child cluster on bare metal. So so far, everything we've seen so far has been aws focus. Just because it's kind of easy to make that was on AWS. We don't want to leave you with the impression that that's all we do, we're covering AWS bare metal and open step deployments as well documented Craftsman Cloud. Let's see it in action with a bare metal child cluster. >>We are on the home stretch, >>right. >>Hello. This demo will cover the process of defining bare metal hosts and then review the steps of defining and deploying a bare metal based doctor enterprise cluster. Yeah, so why bare metal? Firstly, it eliminates hyper visor overhead with performance boost of up to thirty percent provides direct access to GP use, prioritize for high performance wear clothes like machine learning and AI, and support high performance workouts like network functions, virtualization. It also provides a focus on on Prem workloads, simplifying and ensuring we don't need to create the complexity of adding another hyper visor layer in between. So continuing on the theme Why communities and bare metal again Hyper visor overhead. Well, no virtualization overhead. Direct access to hardware items like F p g A s G p, us. We can be much more specific about resource is required on the nodes. No need to cater for additional overhead. We can handle utilization in the scheduling better Onda. We increase the performance and simplicity of the entire environment as we don't need another virtualization layer. Yeah, In this section will define the BM hosts will create a new project. Will add the bare metal hosts, including the host name. I put my credentials. I pay my address, Mac address on, then provide a machine type label to determine what type of machine it is. Related use. Okay, let's get started Certain Blufgan was the operator thing. We'll go and we'll create a project for our machines to be a member off. Helps with scoping for later on for security. I begin the process of adding machines to that project. Yeah. Yeah. So the first thing we had to be in post many of the machine a name. Anything you want? Yeah, in this case by mental zero one. Provide the IAP My user name. Type my password? Yeah. On the Mac address for the active, my interface with boot interface and then the i p m i P address. Yeah, these machines. We have the time storage worker manager. He's a manager. We're gonna add a number of other machines on will speed this up just so you could see what the process. Looks like in the future, better discovery will be added to the product. Okay, Okay. Getting back there. We haven't Are Six machines have been added. Are busy being inspected, being added to the system. Let's have a look at the details of a single note. Mhm. We can see information on the set up of the node. Its capabilities? Yeah. As well as the inventory information about that particular machine. Okay, it's going to create the cluster. Mhm. Okay, so we're going to deploy a bare metal child cluster. The process we're going to go through is pretty much the same as any other child cluster. So credit custom. We'll give it a name. Thank you. But he thought were selecting bare metal on the region. We're going to select the version we want to apply on. We're going to add this search keys. If we hope we're going to give the load. Balancer host I p that we'd like to use out of the dress range update the address range that we want to use for the cluster. Check that the sea idea blocks for the communities and tunnels are what we want them to be. Enable disabled stack light and said the stack light settings to find the cluster. And then, as for any other machine, we need to add machines to the cluster. Here we're focused on building communities clusters. So we're gonna put the count of machines. You want managers? We're gonna pick the label type manager on create three machines. Is a manager for the Cuban a disgusting? Yeah, they were having workers to the same. It's a process. Just making sure that the worker label host like you are so yes, on Duin wait for the machines to deploy. Let's go through the process of putting the operating system on the notes, validating that operating system. Deploying Docker enterprise on making sure that the cluster is up and running ready to go. Okay, let's review the bold events. We can see the machine info now populated with more information about the specifics of things like storage. Yeah, of course. Details of a cluster, etcetera. Yeah, Yeah. Okay. Well, now watch the machines go through the various stages from prepared to deploy on what's the cluster build, and that brings us to the end of this particular do my as you can see the process is identical to that of building a normal child cluster we got our complaint is complete. >>Here we have a child cluster on bare metal for folks that wanted to play the stuff on Prem. >>It's ah been an interesting journey taken from the mothership as we started out building ah management cluster and then populating it with a child cluster and then finally creating a regional cluster to spread the geographically the management of our clusters and finally to provide a platform for supporting, you know, ai needs and and big Data needs, uh, you know, thank goodness we're now able to put things like Hadoop on, uh, bare metal thio in containers were pretty exciting. >>Yeah, absolutely. So with this Doctor Enterprise container cloud platform. Hopefully this commoditized scooping clusters, doctor enterprise clusters that could be spun up and use quickly taking provisioning times. You know, from however many months to get new clusters spun up for our teams. Two minutes, right. We saw those clusters gets better. Just a couple of minutes. Excellent. All right, well, thank you, everyone, for joining us for our demo session for Dr Enterprise Container Cloud. Of course, there's many many more things to discuss about this and all of Miranda's products. If you'd like to learn more, if you'd like to get your hands dirty with all of this content, police see us a training don Miranda's dot com, where we can offer you workshops and a number of different formats on our entire line of products and hands on interactive fashion. Thanks, everyone. Enjoy the rest of the launchpad of that >>thank you all enjoy.

>>Hello, everyone. My name is Rick Pew. I'm a senior product manager at Mirant. This and I have been working on the Doctor Enterprise Container Cloud for the last eight months. Today we're gonna be talking about multi cloud kubernetes. So the first thing to kind of look at is, you know, is multi cloud rial. You know, the terms thrown around a lot and by the way, I should mention that in this presentation, we use the term multi cloud to mean both multi cloud, which you know in the technical sense, really means multiple public clouds and hybrid cloud means public clouds. And on Prem, uh, we use in this presentation will use the term multi cloud to refer to all different types of multiple clouds, whether it's all public cloud or a mixture of on Prem and Public Cloud or, for that matter, multiple on Prem clouds as doctor and price container. Cloud supports all of those scenarios. So it really well, let's look at some research that came out of flex era in their 2020 State of the cloud report. You'll notice that ah, 33% state that they've got multiple public and one private cloud. 53% say they've got multiple public and multiple private cloud. So if you have those two up, you get 86% of the people say that they're in multiple public clowns and at least one private cloud. So I think at this stage we could say that multi cloud is a reality. According to 4 51 research, you know, a number of CEO stated that the strong driver their desire was to optimize cost savings across their private and public clouds. Um, they also wanted to avoid vendor lock in by operating in multiple clouds and try to dissuade their teams from taking too much advantage of a given providers proprietary infrastructure. But they also indicated that there the complexity of using multiple clouds hindered the rate of adoption of doing it doesn't mean they're not doing it. It just means that they don't go assed fast as they would like to go in many cases because of the complexity. And here it Miranda's. We surveyed our customers as well, and they're telling us similar things, you know. Risk management, through the diversification of providers, is key on their list cost optimization and the democratization of allowing their development teams, uh, to create kubernetes clusters without having to file a nightie ticket. But to give them a self service, uh, cloud like environment, even if it's on prem or multi cloud to give them the ability to create their own clusters, resize their own clusters and delete their own clusters without needing to have I t. Or of their operations teams involved at all. But there are some challenges with this, with the different clouds you know require different automation. Thio provisioned the underlying infrastructure or deploy and operating system or deployed kubernetes, for that matter, in a given cloud. You could say that they're not that complicated. They all have, you know, very powerful consoles and a P I s to do that. But did you get across three or four or five different clouds? Then you have to learn three or four or five different AP ice and Web consoles in order to make that happen on in. That scenario is difficult to provide self service for developers across all the cloud options, which is what you want to really accelerate your application innovation. So what's in it for me? You know We've got a number of roles and their prizes developers, operators and business leaders, and they have somewhat different needs. So when the developer side the need is flexibility to meet their development schedules, Number one you know they're under constant pressure to produce, and in order to do that, they need flexibility and in this case, the flexibility to create kubernetes clusters and use them across multiple clouds. Now they also have C I C D tools, and they want them to be able to be normalized on automated across all of the the on prim and public clouds that they're using. You know, in many cases they'll have a test and deployment scenario where they'll want to create a cluster, deploy their software, run their test, score the tests and then delete that cluster because the only point of that cluster, perhaps, was to test ah pipeline of delivery. So they need that kind of flexibility. From the operator's perspective, you know, they always want to be able to customize the control of their infrastructure and deployment. Uh, they certainly have the desire to optimize their optics and Capex fans. They also want to support their develops teams who many times their their customers through a p I access for on Prem and public clouds burst. Scaling is something operators are interested in, and something public clouds can provide eso the ability to scale out into public clouds, perhaps from there on prem infrastructure in a seamless manner. And many times they need to support geographic distribution of applications either for compliance or performance reasons. So having you know, data centers all across the world and be able to specifically target a given region, uh, is high on their list. Business leaders want flexibility and confidence to know that you know, they're on prim and public cloud uh, deployments. Air fully supported. They want to be able, like the operator, optimize their cloud, spends business leaders, think about disaster recovery. So having the applications running and living in different data centers gives them the opportunity to have disaster recovery. And they really want the flexibility of keeping private data under their control. On on Prem In certain applications may access that on Prem. Other applications may be able to fully run in the cloud. So what should I look for in a container cloud? So you really want something that fully automates these cluster deployments for virtual machine or bare metal. The operating system, uh, and kubernetes eso It's not just deploying kubernetes. It's, you know, how do I create my underlying infrastructure of a VM or bare metal? How do I deploy the operating system? And then, on top of all that, I want to be able to deploy kubernetes. Uh, you also want one that gives a unified cluster lifecycle management across all the clouds. So these clusters air running software gets updated. Cooper Netease has a new release cycle. Uh, they come out with something new. It's available, you know, How do you get that across all of your clusters? That air running in multiple clouds. We also need a container cloud that can provide you the visibility through logging, monitoring and alerting again across all the clouds. You know, many offerings have these for a particular cloud, but getting that across multiple clouds, uh, becomes a little more difficult. The Doctor Enterprise Container cloud, you know, is a very strong solution and really meets many of these, uh, dimensions along the left or kind of the dimensions we went through in the last slide we've got on Prem and public clouds as of RG A Today we're supporting open stack and bare metal for the on Prem Solutions and AWS in the public cloud. We'll be adding VM ware very soon for another on Prem uh, solution as well as azure and G C P. So thank you very much. Uh, look forward, Thio answering any questions you might have and we'll call that a rap. Thank you. >>Hi, Rick. Thanks very much for that. For that talk, I I am John James. You've probably seen me in other sessions. I do marketing here in Miran Tous on. I wanted to to take this opportunity while we had Rick to ask some more questions about about multi cloud. It's ah, potentially a pretty big topic, isn't it, Rick? >>Yeah. I mean, you know, the devil's in the details and there's, uh, lots of details that we could go through if you'd like, be happy to answer any questions that you have. >>Well, we've been talking about hybrid cloud for literally years. Um, this is something that I think you know, several generations of folks in the in the I. A s space doing on premise. I s, for example, with open stack the way Miran Tous Uh does, um, found, um, you know, thought that that it had a lot of potential. A lot of enterprises believed that, but there were There were things stopping people from from making it. Really, In many cases, um, it required a very, ah, very high degree of willingness to create homogeneous platforms in the cloud and on the premise. Um, and that was often very challenging. Um, but it seems like with things like kubernetes and with the isolation provided by containers, that this is beginning to shift, that that people are actually looking for some degree of application portability between their own Prem and there and their cloud environments. And that this is opening up, Uh, you know, investment on interest in pursuing this stuff. Is that the right perception? >>Yeah. So let's let's break that down a little bit. So what's nice about kubernetes is through the a. P. I s are the same. Regardless of whether it's something that Google or or a W s is offering as a platform as a service or whether you've taken the upstream open source project and deploy it yourself on parameter in a public cloud or whatever the scenario might be or could be a competitor of Frances's product, the Kubernetes A. P I is the same, which is the thing that really gives you that application portability. So you know, the container itself is contained arising, obviously your application and minimizing any kind of dependency issues that you might have And then the ability to deploy that to any of the coup bernetti clusters you know, is the same regardless of where it's running, the complexity comes and how doe I actually spend up a cluster in AWS and open stack and D M Where and gp An azure. How do I build that infrastructure and and spin that up and then, you know, used the ubiquitous kubernetes a p I toe actually deploy my application and get it to run. So you know what we've done is we've we've unified and created A I use the word normalized. But a lot of times people think that normalization means that you're kind of going to a lowest common denominator, which really isn't the case and how we've attacked the the enabling of multi cloud. Uh, you know, what we've done is that we've looked at each one of the providers and are basically providing an AP that allows you to utilize. You know, whatever the best of you know, that particular breed of provider has and not, uh, you know, going to at least common denominator. But, you know, still giving you a ah single ap by which you can, you know, create the infrastructure and the infrastructure could be on Prem is a bare metal infrastructure. It could be on preeminent open stack or VM ware infrastructure. Any of the public clouds, you know, used to have a a napi I that works for all of them. And we've implemented that a p i as an extension to kubernetes itself. So all of the developers, Dev ops and operators that air already familiar operating within the, uh, within the aapi of kubernetes. It's very, very natural. Extension toe actually be able to spend up these clusters and deploy them >>Now that's interesting. Without giving away, obviously what? Maybe special sauce. Um, are you actually using operators to do this in the Cooper 90? Sense of the word? >>Yes. Yeah, we've extended it with with C R D s, uh, and and operators and controllers, you know in the way that it was meant to be extended. So Kubernetes has a recipe on how you extend their A P I on that. That's what we used as our model. >>That, at least to me, makes enormous sense. Nick Chase, My colleague and I were digging into operators a couple of weeks ago, and that's a very elegant technology. Obviously, it's a it's evolving very fast, but it's remarkably unintimidating once you start trying to write them. We were able toe to compose operators around Cron and other simple processes and just, >>you know, >>a couple of minutes on day worked, which I found pretty astonishing. >>Yeah, I mean, you know, Kubernetes does a lot of things and they spent a lot of effort, um, in being able, you know, knowing that their a p I was gonna be ubiquitous and knowing that people wanted to extend it, uh, they spent a lot of effort in the early development days of being able to define that a p I to find what an operator was, what a controller was, how they interact. How a third party who doesn't know anything about the internals of kubernetes could add whatever it is that they wanted, you know, and follow the model that makes it work. Exactly. Aziz, the native kubernetes ap CSTO >>What's also fascinating to me? And, you know, I've I've had a little perspective on this over the past, uh, several weeks or a month or so working with various stakeholders inside the company around sessions related to this event that the understanding of how things work is by no means evenly distributed, even in a company as sort of tightly knit as Moran Tous. Um, some people who shall remain nameless have represented to me that Dr Underprice Container Cloud basically works. Uh, if you handed some of the EMS, it will make things for you, you know, and this is clearly not what's going on that that what's going on is a lot more nuanced that you are using, um, optimal resource is from each provider to provide, uh, you know, really coherent architected solutions. Um, the load balancing the d. N s. The storage that this that that right? Um all of which would ultimately be. And, you know, you've probably tried this. I certainly have hard to script by yourself in answerable or cloud formation or whatever. Um, this is, you know, this is not easy work. I I wrote a about the middle of last year for my prior employer. I wrote a dip lawyer in no Js against the raw aws a piece for deployment and configuration of virtual networks and servers. Um, and that was not a trivial project. Um, it took a long time to get thio. Uh, you know, a dependable result. And to do it in parallel and do other things that you need to do in order to maintain speed. One of the things, in fact, that I've noticed in working with Dr Enterprise Container Cloud recently, is how much parallelism it's capable of within single platforms. It's It's pretty powerful. I mean, if you want to clusters to be deployed simultaneously, that's not hard for Doc. Aerated price container cloud to dio on. I found it pretty remarkable because I have sat in front of a single laptop trying to churn out of cluster under answerable, for example, and just on >>you get into that serial nature, your >>poor little devil, every you know, it's it's going out and it's ssh, Indian Terminals and it's pretending it's a person and it's doing all that stuff. This is much more magical. Um, so So that's all built into the system to, isn't it? >>Yeah. Interesting, Really Interesting point on that. Is that you know, the complexity isn't not necessarily and just creating a virtual machine because all of these companies have, you know, spend a lot of effort to try to make that as easy as possible. But when you get into networking, load balancing, routing, storage and hooking those up, you know, two containers automating that if you were to do that in terror form or answerable or something like that is many, many, many lines of code, you know, people have to experiment. Could you never get it right the first or second or the third time? Uh, you know, and then you have to maintain that. So one of the things that we've heard from customers that have looked a container cloud was that they just can't wait to throw away their answerable or their terror form that they've been maintaining for a couple of years. The kind of enables them to do this. It's very brittle. If if the clouds change something, you know on the network side, let's say that's really buried. And it's not something that's kind of top of mind. Uh, you know, your your thing fails or maybe worse, you think that it works. And it's not until you actually go to use it that you notice that you can't get any of your containers. So you know, it's really great the way that we've simplified that for the users and again democratizing it. So the developers and Dev ops people can create these clusters, you know, with ease and not worry about all the complexities of networking and storage. >>Another thing that amazed me as I was digging into my first, uh, Dr Price container Cloud Management cluster deployment was how, uh, I want I don't want to use the word nuanced again, but I can't think of a better word. Nuanced. The the security thinking is in how things air set up. How, um, really delicate the thinking about about how much credential power you give to the deploy. Er the to the seed server that deploys your management cluster as opposed thio Um uh or rather the how much how much administrative access you give to the to the administrator who owns the entire implementation around a given provider versus how much power the seed server gets because that gets its own user right? It gets a bootstrap user specifically created so that it's not your administrator, you know, more limited visibility and permissions. And this whole hierarchy of permissions is then extended down into the child clusters that this management cluster will ultimately create. So that Dev's who request clusters will get appropriate permissions granted within. Ah, you know, a corporate schema of permissions. But they don't get the keys to the kingdom. They don't have access to anything they don't you know they're not supposed to have access to, but within their own scope, they're safe. They could do anything they want, so it's like a It's a It's a really neat kind of elegant way of protecting organizations against, for example, resource over use. Um, you know, give people the power to deploy clusters, and basically you're giving them the power toe. Make sure that a big bill hits you know, your corporate accounting office at the end of the billing cycle, um so there have to be controls and those controls exist in this, you know, in this. >>Yeah, And there's kind of two flavors of that. One is kind of the day one that you're doing the deployment you mentioned the seed servers, you know, And then it creates a bastion server, and then it creates, you know, the management cluster and so forth, you know, and how all those permissions air handled. And then once the system is running, you know, then you have full access to going into key cloak, which is a very powerful open source identity management tool on you have dozens of, you know, granular permissions that you can give to an individual user that gives them permission to do certain things and not others within the context of kubernetes eso. It's really well thought out. And the defaults, you know, our 80% right. You know, there's very few people are gonna have to go in and sort of change those defaults. You mentioned the corporate directory. You know, hooks right upto l bap or active directory can suck everybody down. So there's no kind of work from a day. One perspective of having to go add. You know everybody that you can think of different teams and groupings of of people. Uh, you know, that's kind of all given from the three interface to the corporate directory. And so it just makes kind of managing the users and and controlling who can do what? Uh, really easy. And, you know, you know, day one day two it's really almost like our one hour to write because it's just all the defaults were really well thought out. You can deploy, you know, very powerful doctor and price container cloud, you know, within an hour, and then you could just start using it. And you know, you can create users if you want. You can use the default users. That air set up a time goes on, you can fine tune that, and it's a really, really nice model again for the whole frictionless democratization of giving developers the ability to go in and get it out of, you know, kind of their way and doing what they want to do. And I t is happy to do that because they don't like dozens of tickets and saying, you know, create a cluster for this team created cluster for that team. You know, here's the size of these guys. Want to resize when you know let's move all that into a self service model and really fulfill the prophecy of, you know, speeding up application development. >>It strikes me is extremely ironic that one of the things that public cloud providers bless them, uh, have always claimed, is that their products provide this democratization when in the experience, I think my own experience and the experience of most of the AWS developers, for example, not toe you know, name names, um, that I've encountered is that an initial experience of trying to start start a virtual machine and figuring out how to log into it? A. W s could take the better part of an afternoon. It's just it's not familiar once you have it in your fingers. Boom. Two seconds, right. But, wow, that learning curve is steep and precipitous, and you slip back and you make stupid mistakes your first couple 1000 times through the loop. Um, by letting people skip that and letting them skip it potentially on multiple providers, in a sense, I would think products like this are actually doing the public cloud industry is, you know, a real surface Hide as much of that as you can without without taking the power away. Because ultimately people want, you know, to control their destiny. They want choice for a reason. Um, and and they want access to the infinite services And, uh, and, uh, innovation that AWS and Azure and Google are all doing on their platforms. >>Yeah, you know, and they're solving, uh, very broad problems in the public clouds, you know, here were saying, you know, this is a world of containers, right? This is a world of orchestration of these containers. And why should I have to worry about the underlying infrastructure, whether it's a virtual machine or bare metal? You know, I shouldn't care if I'm an application developer developing some database application. You know, the last thing I wanna worry about is how do I go in and create a virtual machine? Oh, this is running. And Google. It's totally different than the one I was creating. An AWS I can't find. You know where I get the I P address in Google. It's not like it was an eight of us, you know, and you have to relearn the whole thing. And that's really not what your job is. Anyways, your job is to write data base coat, for example. And what you really want to do is just push a button, deploy a nor kiss traitor, get your app on it and start debugging it and getting it >>to work. Yep. Yeah, it's It's powerful. I've been really excited to work with the product the past week or so, and, uh, I hope that folks will look at the links at the bottoms of our thank you slides and, uh, and, uh, avail themselves of of free trial downloads of both Dr Enterprise Container, Cloud and Lens. Thank you very much for spending this extra time with me. Rick. I I think we've produced some added value here for for attendees. >>Well, thank you, John. I appreciate your help. >>Have a great rest of your session by bike. >>Okay, Thanks. Bye.

>>C G technology is focused on data where we have long believed that data is in our DNA. We help maximize humanity's potential by delivering world class, precision engineered data solutions developed through sustainable and profitable partnerships. Included in our offerings are hard disk drives. As I'm sure many of you know, ah, hard drive consists of a slider also known as a drive head or transducer attached to a head gimbal assembly. I had stack assembly made up of multiple head gimbal assemblies and a drive enclosure with one or more platters, or just that the head stacked assembles into. And while the concept hasn't changed, hard drive technology has progressed well beyond the initial five megabytes, 500 quarter inch drives that Seagate first produced. And, I think 1983. We have just announced in 18 terabytes 3.5 inch drive with nine flatters on a single head stack assembly with dual head stack assemblies this calendar year, the complexity of these drives further than need to incorporate Edge analytics at operation sites, so G Edward stemming established the concept of continual improvement and everything that we do, especially in product development and operations and at the end of World War Two, he embarked on a mission with support from the US government to help Japan recover from its four time losses. He established the concept of continual improvement and statistical process control to the leaders of prominent organizations within Japan. And because of this, he was honored by the Japanese emperor with the second order of the sacred treasure for his teachings, the only non Japanese to receive this honor in hundreds of years. Japan's quality control is now world famous, as many of you may know, and based on my own experience and product development, it is clear that they made a major impact on Japan's recovery after the war at Sea Gate. The work that we've been doing and adopting new technologies has been our mantra at continual improvement. As part of this effort, we embarked on the adoption of new technologies in our global operations, which includes establishing machine learning and artificial intelligence at the edge and in doing so, continue to adopt our technical capabilities within data science and data engineering. >>So I'm a principal engineer and member of the Operations and Technology Advanced Analytics Group. We are a service organization for those organizations who need to make sense of the data that they have and in doing so, perhaps introduce a different way to create an analyzed new data. Making sense of the data that organizations have is a key aspect of the work that data scientist and engineers do. So I'm a project manager for an initiative adopting artificial intelligence methodologies for C Gate manufacturing, which is the reason why I'm talking to you today. I thought I'd start by first talking about what we do at Sea Gate and follow that with a brief on artificial intelligence and its role in manufacturing. And I'd like them to discuss how AI and machine Learning is being used at Sea Gate in developing Edge analytics, where Dr Enterprise and Cooper Netease automates deployment, scaling and management of container raised applications. So finally, I like to discuss where we are headed with this initiative and where Mirant is has a major role in case some of you are not conversant in machine learning, artificial intelligence and difference outside some definitions. To cite one source, machine learning is the scientific study of algorithms and statistical bottles without computer systems use to effectively perform a specific task without using explicit instructions, relying on patterns and inference Instead, thus, being seen as a subset of narrow artificial intelligence were analytics and decision making take place. The intent of machine learning is to use basic algorithms to perform different functions, such as classify images to type classified emails into spam and not spam, and predict weather. The idea and this is where the concept of narrow artificial intelligence comes in, is to make decisions of a preset type basically let a machine learn from itself. These types of machine learning includes supervised learning, unsupervised learning and reinforcement learning and in supervised learning. The system learns from previous examples that are provided, such as images of dogs that are labeled by type in unsupervised learning. The algorithms are left to themselves to find answers. For example, a Siris of images of dogs can be used to group them into categories by association that's color, length of coat, length of snout and so on. So in the last slide, I mentioned narrow a I a few times, and to explain it is common to describe in terms of two categories general and narrow or weak. So Many of us were first exposed to General Ai in popular science fiction movies like 2000 and One, A Space Odyssey and Terminator General Ai is a I that can successfully perform any intellectual task that a human can. And if you ask you Lawn Musk or Stephen Hawking, this is how they view the future with General Ai. If we're not careful on how it is implemented, so most of us hope that is more like this is friendly and helpful. Um, like Wally. The reality is that machines today are not only capable of weak or narrow, a I AI that is focused on a narrow, specific task like understanding, speech or finding objects and images. Alexa and Google Home are becoming very popular, and they can be found in many homes. Their narrow task is to recognize human speech and answer limited questions or perform simple tasks like raising the temperature in your home or ordering a pizza as long as you have already defined the order. Narrow. AI is also very useful for recognizing objects in images and even counting people as they go in and out of stores. As you can see in this example, so artificial intelligence supplies, machine learning analytics inference and other techniques which can be used to solve actual problems. The two examples here particle detection, an image anomaly detection have the potential to adopt edge analytics during the manufacturing process. Ah, common problem in clean rooms is spikes in particle count from particle detectors. With this application, we can provide context to particle events by monitoring the area around the machine and detecting when foreign objects like gloves enter areas where they should not. Image Anomaly detection historically has been accomplished at sea gate by operators in clean rooms, viewing each image one at a time for anomalies, creating models of various anomalies through machine learning. Methodologies can be used to run comparative analyses in a production environment where outliers can be detected through influence in an automated real Time analytics scenario. So anomaly detection is also frequently used in machine learning to find patterns or unusual events in our data. How do you know what you don't know? It's really what you ask, and the first step in anomaly detection is to use an algorithm to find patterns or relationships in your data. In this case, we're looking at hundreds of variables and finding relationships between them. We can then look at a subset of variables and determine how they are behaving in relation to each other. We use this baseline to define normal behavior and generate a model of it. In this case, we're building a model with three variables. We can then run this model against new data. Observations that do not fit in the model are defined as anomalies, and anomalies can be good or bad. It takes a subject matter expert to determine how to classify the anomalies on classify classification could be scrapped or okay to use. For example, the subject matter expert is assisting the machine to learn the rules. We then update the model with the classifications anomalies and start running again, and we can see that there are few that generate these models. Now. Secret factories generate hundreds of thousands of images every day. Many of these require human toe, look at them and make a decision. This is dull and steak prone work that is ideal for artificial intelligence. The initiative that I am project managing is intended to offer a solution that matches the continual increased complexity of the products we manufacture and that minimizes the need for manual inspection. The Edge Rx Smart manufacturing reference architecture er, is the initiative both how meat and I are working on and sorry to say that Hamid isn't here today. But as I said, you may have guessed. Our goal is to introduce early defect detection in every stage of our manufacturing process through a machine learning and real time analytics through inference. And in doing so, we will improve overall product quality, enjoy higher yields with lesser defects and produce higher Ma Jin's. Because this was entirely new. We established partnerships with H B within video and with Docker and Amaranthus two years ago to develop the capability that we now have as we deploy edge Rx to our operation sites in four continents from a hardware. Since H P. E. And in video has been an able partner in helping us develop an architecture that we have standardized on and on the software stack side doctor has been instrumental in helping us manage a very complex project with a steep learning curve for all concerned. To further clarify efforts to enable more a i N M l in factories. Theobald active was to determine an economical edge Compute that would access the latest AI NML technology using a standardized platform across all factories. This objective included providing an upgrade path that scales while minimizing disruption to existing factory systems and burden on factory information systems. Resource is the two parts to the compute solution are shown in the diagram, and the gateway device connects to see gates, existing factory information systems, architecture ER and does inference calculations. The second part is a training device for creating and updating models. All factories will need the Gateway device and the Compute Cluster on site, and to this day it remains to be seen if the training devices needed in other locations. But we do know that one devices capable of supporting multiple factories simultaneously there are also options for training on cloud based Resource is the stream storing appliance consists of a kubernetes cluster with GPU and CPU worker notes, as well as master notes and docker trusted registries. The GPU nodes are hardware based using H B E l 4000 edge lines, the balance our virtual machines and for machine learning. We've standardized on both the H B E. Apollo 6500 and the NVIDIA G X one, each with eight in video V 100 GP use. And, incidentally, the same technology enables augmented and virtual reality. Hardware is only one part of the equation. Our software stack consists of Docker Enterprise and Cooper Netease. As I mentioned previously, we've deployed these clusters at all of our operations sites with specific use. Case is planned for each site. Moran Tous has had a major impact on our ability to develop this capability by offering a stable platform in universal control plane that provides us, with the necessary metrics to determine the health of the Kubernetes cluster and the use of Dr Trusted Registry to maintain a secure repository for containers. And they have been an exceptional partner in our efforts to deploy clusters at multiple sites. At this point in our deployment efforts, we are on prem, but we are exploring cloud service options that include Miranda's next generation Docker enterprise offering that includes stack light in conjunction with multi cluster management. And to me, the concept of federation of multi cluster management is a requirement in our case because of the global nature of our business where our operation sites are on four continents. So Stack Light provides the hook of each cluster that banks multi cluster management and effective solution. Open source has been a major part of Project Athena, and there has been a debate about using Dr CE versus Dr Enterprise. And that decision was actually easy, given the advantages that Dr Enterprise would offer, especially during a nearly phase of development. Cooper Netease was a natural addition to the software stack and has been widely accepted. But we have also been a work to adopt such open source as rabbit and to messaging tensorflow and tensor rt, to name three good lab for developments and a number of others. As you see here, is well, and most of our programming programming has been in python. The results of our efforts so far have been excellent. We are seeing a six month return on investment from just one of seven clusters where the hardware and software cost approached close to $1 million. The performance on this cluster is now over three million images processed per day for their adoption has been growing, but the biggest challenge we've seen has been handling a steep learning curve. Installing and maintaining complex Cooper needs clusters in data centers that are not used to managing the unique aspect of clusters like this. And because of this, we have been considering adopting a control plane in the cloud with Kubernetes as the service supported by Miranda's. Even without considering, Kubernetes is a service. The concept of federation or multi cluster management has to be on her road map, especially considering the global nature of our company. Thank you.

>>welcome to the session. Long live swarm with containers and kubernetes everywhere we have this increasing cloud complexity at the same time that we're facing economic uncertainty and, of course, to navigate this. For most companies, it's a matter of focusing on speed and on shipping and iterating their code faster. Now. For many, Marantz is customers. That means using docker swarm rather than kubernetes to handle container orchestration. We really believe that the best way to increase your speed to production is choice, simplicity and security. So we wanted to bring you a couple of experts to talk about the state of swarm and Docker enterprise and how you can make best use of both of you. So let's get to it. Well, good afternoon or good morning, depending on where you are on and welcome to today's session. Long live swarm. I am Nick Chase. I'm head of content here at Mantis and I would like to introduce you to our two Panelists today eight of Manzini. Why don't you introduce yourself? >>I am a van CNI. I'm a solutions architect here at Moran Tous on work primarily with Docker Enterprise System. I have a long history of working with support team. Um, at what used to be Ah Docker Enterprise, part of Docker Inc. >>Yeah, Okay. Great. And Don Power. >>I, um Yeah, I'm Don Power on the docker. Captain Docker, community leader. Right now I run our Dev Ops team for Citizens Bank out of Nashville, Tennessee, and happy to be here. >>All right, Excellent. So All right, so thank you both for coming. Now, before we say anything else, I want to go ahead and kind of name the elephant in the room. There's been a lot of talk about the >>future. Yeah, that's right. Um, swarm as it stands right now, um, we have, ah, very vested interest in keeping our customers on who want to continue using swarm, functional and keeping swarm a viable alternative or complement to kubernetes. However you see the orchestration war playing out as it were. >>Okay? It's hardly a war at this point, but they do work together, and so that's >>absolutely Yeah, I I definitely consider them more of like, complimentary services, um, using the right tool for the job. Sort of sense. They both have different design goals when they were originally created and set out so I definitely don't see it as a completely one or the other kind of decision and that they could both be used in the same environment and similar clusters to run whatever workload that you have. >>Excellent. And we'll get into the details of all that as we go along. So that's terrific. So I have not really been involved in in the sort of swarm area. So set the stage for us where we kind of start out with all of this. Don I know that you were involved and so guys said, set the stage for us. >>Sure, Um I mean so I've been a heavy user of swarm in my past few roles. Professionally, we've been running containers in production with Swarm for coming up on about four years. Now, Um, in our case, we you know, we looked at what was available at the time, and of course you had. Kubernetes is your biggest contender out there, but like I just mentioned, the one of the things that really led us to swarm is it's design goals were very different than kubernetes. So Kubernetes tries to have an answer for absolutely every scenario where swarm tries to have an answer for, like, the 80% of problems or challenges will say that you might come across 80% of the workloads. Um, I had a better way of saying that, but I think I got my point across >>E Yeah, I think I think you hit the nail on the head. Um, Kubernetes in particular with the way that kubernetes itself is an a P I I believe that kubernetes was, um, you know, written as a toolkit. It wasn't really intended to be used by end users directly. It was really a way to build platforms that run containers. And because it's this really, really extensible ap I you can extend it to manage all sorts of resource is swarm doesn't have that X sensibility aspect, but what it was designed to do, it does very, very well and very easily in a very, very simple sort of way. Um, it's highly opinionated about the way that you should use the product, but it works very effectively. It's very easy to use. It's very low. Um, not low effort, but low. Ah, low barrier to entry. >>Yes. Yes. Absolutely. I was gonna touch on the same thing. It's very easy for someone to come in. Pick up swarm. You know they don't They don't have to know anything about the orchestrator on day one. Most people that are getting into this space are very familiar with Docker. Compose um, and entering from Docker compose into swarm is changing one command that you would run on the command line. >>Yeah, very, very trivial to if you are already used to building docker files using composed, organize your deployment into stacks of related components. It's trivial to turn on swarm mode and then deploy your container set to a cluster. >>Well, excellent. So answer this question for me. Is the swarm of today the same as the swarm of, you know, the original swarm. So, like when swim first started is that the same is what we have now >>it's kind of ah, complicated story with the storm project because it's changed names and forms a few times. Originally in is really somewhere around 2014 in the first version, and it was a component that you really had to configure and set up separately from Docker Ah, the way that it was structured. Ah, you would just have docker installed on a number of servers are machines in your cluster. And then you would organize them into a swarm by bringing your own database and some of the tooling to get those nodes talking to each other and to organize your containers across all of your docker engines. Ah, few years later, the swarm project was retooled and baked into the docker engine. And, um, this is where we sort of get the name change from. So originally it was a feature that we called swarm. Ah. Then the Swarm Kit project was released on Get Hub and baked directly into the engine, where they renamed it as swarm mode. Because now it is a motile option that you just turn on as a button in the docker engine and because it's already there the, um, the tuning knobs that you haven't swarm kit with regard to how what my time outs are and some of these other sort of performance settings there locked there, they're there. It's part of the opinionated set of components that builds up the docker engine is that we bring in the Swarm Kit project with a certain set of defaults and settings. And that is how it operates in today's version of Docker engine. >>Uh, okay for that, that makes sense. That makes sense. So ah, so don, I know you have pretty strong feelings about this topic, but it is swarm still viable in a world that's sort of increasingly dominated by Kubernetes. >>Absolutely. And you were right. I'm very passionate about this topic where I work. We're we're doing almost all of our production work lives on swarm we only have out of Ah, we've got something like 600 different services between three and 4000 containers. At any given point in time. Out of all of those projects, all of those services we've only run into two or three that don't kind of fit into the opinionated model of swarm. So we are running those on KUBERNETES in the same cluster using Moranis is Docker enterprise offering. But, um, no, that's a very, very small percentage of services that we didn't have an answer for in swarm with one. The one case that really gets us just about every time is scaling state full services. But you're gonna have very few staple services in most environments for things like micro service architecture, which is predominantly what we build out. Swarm is perfect. It's simple. It's easy to use you, don't you? Don't end up going for miles of yamma files trying to figure out the one setting that you didn't get exactly right? Um yeah, the other Thea the other big piece of it that way really led us to adopting it so heavily in the beginning is, you know, the overlay network. So your networks don't have to span the whole cluster like they do with kubernetes. So we could we could set up a network isolation between service A and service B, just by use using the built in overlay networks. That was a huge component that, like I said, let us Teoh adopting it so heavily when we first got started. >>Excellent. You look like you're about to say something in a >>Yeah, I think that speaks to the design goals for each piece of software. On the way that I've heard this described before is with regard to the networking piece the ah, the docker networking under the hood, um, feels like it was written by a network engineer. The way that the docker engine overlay networks communicate uses ah, VX lan under the hood, which creates pseudo V lands for your containers. And if two containers aren't on the same Dylan, there's no way they can communicate with each other as opposed to the design of kubernetes networking, which is really left to the C and I implementation but still has the design philosophy of one big, flat sub net where every I p could reach every other i p and you control what is allowed to access, what by policy. So it's more of an application focused Ah design. Whereas in Docker swarm on the overlay networking side, it's really of a network engineering sort of focus. Right? >>Okay, got it. Well, so now how does all this fit in with Docker enterprise now? So I understand there's been some changes on how swarm is handled within Docker Enterprise. Coming with this new release, >>Docker s O swarm Inside Docker Enterprise is represented as both the swarm classic legacy system that we shift way back in 2014 on and then also the swarm mode that is curly used in the docker engine. Um, the Swarm Classic back end gives us legacy support for being able to run unmanaged plane containers onto a cluster. If you were to take Docker ce right now, you would find that you wouldn't be able to just do a very basic docker run against a whole cluster of machines. You can create services using the swarms services, a p I but, um, that that legacy plane container support is something that you have to set up external swarm in order to provide. So right now, the architecture of Docker Enterprise UCP is based on some of that legacy code from about five or six years ago. Okay. Ah, that gives us ability to deploy plane containers for use cases that require it as well as swarm services for those kinds of workloads that might be better served by the built in load balancing and h A and scaling features that swarm provides. >>Okay, so now I know that at one point kubernetes was deployed within Docker Enterprise as you create a swarm cluster and then deploy kubernetes on top of swarm. >>Correct? That is how the current architecture works. >>Okay. All right. And then, um what is what is where we're going with this like, Are we supposed to? Are we going to running Swarm on top of kubernetes? What's >>the the design goals for the future of swarm within branches? Stocker Enterprise are that we will start the employing Ah, like kubernetes cluster features as the base and a swarm kit on top of kubernetes. So it is like you mentioned just a reversal of the roles. I think we're finding that, um, the ability to extend kubernetes a p I to manage resource is is valuable at an infrastructure and platform level in a way that we can't do with swarm. We still want to be able to run swarm workloads. So we're going to keep the swarm kit code the swarm kit orchestration features to run swarm services as a part of the platform to keep the >>got it. Okay, so, uh, if I'm a developer and I want to run swarm, but my company's running kubernetes what? What are my one of my options there? Well, I think >>eight touched on it pretty well already where you know, it depends on your design goals, and you know, one of the other things that's come up a few times is Thea. The level of entry for for swarm is much, much simpler than kubernetes. So I mean, it's it's kind of hard to introduce anything new. So I mean, a company, a company that's got most of their stuff in kubernetes and production is gonna have a hard time maybe looking at a swarm. I mean, this is gonna be, you know, higher, higher up, not the boots on the ground. But, um, you know, the the upper management, that's at some point, you have to pay for all their support, all of it. What we did in our approach. Because there was one team already using kubernetes. We went ahead and stood up a small cluster ah, small swarm cluster and taught the developers how to use it and how to deploy code to it. And they loved it. They thought it was super simple. A time went on, the other teams took notice and saw how fast these guys were getting getting code deployed, getting services up, getting things usable, and they would look over at what the innovation team was doing and say, Hey, I I want to do that to, uh, you know, so there's there's a bunch of different approaches. That's the approach we took and it worked out very well. It looks like you wanted to say something too. >>Yeah, I think that if you if you're if you're having to make this kind of decision, there isn't There isn't a wrong choice. Ah, it's never a swarm of its role and your organization, right? Right. If you're if you're an individual and you're using docker on your workstation on your laptop but your organization wants to standardize on kubernetes there, there are still some two rules that Mike over Ah, pose. And he's manifest if you need to deploy. Coop resource is, um if you are running Docker Enterprise Swarm kit code will still be there. And you can run swarm services as regular swarm workloads on that component. So I I don't want to I don't want people to think that they're going to be like, locked into one or the other orchestration system. Ah, there the way we want to enable developer choice so that however the developer wants to do their work, they can get it done. Um Docker desktop. Ah, ships with that kubernetes distribution bundled in it. So if you're using a Mac or Windows and that's your development, uh, system, you can run docker debt, turn on your mode and run the kubernetes bits. So you have the choices. You have the tools to deploy to either system. >>And that's one of the things that we were super excited about when they introduced Q. Burnett ease into the Docker Enterprise offering. So we were able to run both, so we didn't have to have that. I don't want to call it a battle or argument, but we didn't have to make anybody choose one or the other. We, you know, we gave them both options just by having Docker enterprise so >>excellent. So speaking of having both options, let's just say for developers who need to make a decision while should I go swarm, or should I go kubernetes when it sort of some of the things that they should think about? >>So I think that certain certain elements of, um, certain elements of containers are going to be agnostic right now. So the the the designing a docker file and building a container image, you're going to need to know that skill for either system that you choose to operate on. Ah, the swarm value. Some of the storm advantage comes in that you don't have to know anything beyond that. So you don't have to learn a whole new A p I a whole new domain specific language using Gamel to define your deployment. Um, chances are that if you've been using docker for any length of time, you probably have a whole stack of composed files that are related to things that you've worked on. And, um, again, the barrier to entry to getting those running on swarm is very low. You just turn it on docker stack, deploy, and you're good to go. So I think that if you're trying to make that choice, if you I have a use case that doesn't require you to manage new resource is if you don't need the Extensible researchers part, Ah, swarm is a great great, great viable option. >>Absolutely. Yeah, the the recommendation I've always made to people that are just getting started is start with swarm and then move into kubernetes and going through the the two of them, you're gonna figure out what fits your design principles. What fits your goals. Which one? You know which ones gonna work best for you. And there's no harm in choosing one or the other using both each one of you know, very tailor fit for very various types of use cases. And like I said, kubernetes is great at some things, but for a lot of other stuff, I still want to use swarm and vice versa. So >>on my home lab, for all my personal like services that I run in my, uh, my home network, I used storm, um, for things that I might deploy onto, you know, a bit this environment, a lot of the ones that I'm using right now are mainly tailored for kubernetes eso. I think especially some of the tools that are out there in the open source community as well as in docker Enterprise helped to bridge that gap like there's a translator that can take your compose file, turn it into kubernetes. Yeah, Mel's, um, if if you're trying to decide, like on the business side, should we standardize on former kubernetes? I think like your what? What functionality are you looking at? Out of getting out of your system? If you need things like tight integration into a ah infrastructure vendor such as AWS Azure or VM ware that might have, like plug ins for kubernetes. You're now you're getting into that area where you're managing Resource is of the infrastructure with your orchestration. AP I with kube so things like persistent volumes can talk to your storage device and carve off chunks of storage and assign those two pods if you don't have that need or that use case. Um, you know, KUBERNETES is bringing in a lot of these features that you maybe you're just not taking advantage of. Um, similarly, if you want to take advantage of things like auto scaling to scale horizontally, let's say you have a message queue system and then a number of workers, and you want to start scaling up on your workers. When your CPU hits a certain a metric. That is something that Kubernetes has built right into it. And so, if you want that, I would probably suggest that you look at kubernetes if you don't need that, or if you want to write some of that tooling yourself. Swarm doesn't have an object built into it that will do automatic horizontal scaling based on some kind of metric. So I always consider this decision as a what features are the most I available to you and your business that you need to Yep. >>All right. Excellent. Well, and, ah, fortunately, of course, they're both available on Docker Enterprise. So aren't we lucky? All right, so I am going to wrap this up. I want to thank Don Bauer Docker captain, for coming here and spending some time with us and eight of Manzini. I would like to thank you. I know that the the, uh, circumstances are less than ideal here for your recording today, but we appreciate you joining us. Um and ah, both of you. Thank you very much. And I want to invite all of you. First of all, thank you for joining us. We know your time is valuable and I want to invite you all Teoh to take a look at Docker Enterprise. Ah, follow the link that's on your screen and we'll see you in the next session. Thank you all so much. Thank you. >>Thank you, Nick.