Cloud Native Computing Foundation KubeCon + CloudNativeCon Europe 2018 (Copenhagen), 14 Nov 2015

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: YAML is for Computers. ksonnet is for Humans - Bryan Liles, Heptio (Any Skill Level)

Description

Want to view more sessions and keep the conversations going? Join us for KubeCon + CloudNativeCon North America in Seattle, December 11 - 13, 2018 (http://bit.ly/KCCNCNA18) or in Shanghai, November 14-15 (http://bit.ly/kccncchina18).

YAML is for Computers. ksonnet is for Humans - Bryan Liles, Heptio (Any Skill Level)

YAML as a configuration language for Kubernetes means the configuration is accessible to the API service and mostly accessible to humans. The open source project, ksonnet, provides an intuitive way to create and edit Kubernetes configuration files in a declarative fashion. It also allows configuration of applications across multiple environments. In this session, the audience will be introduced to the world of easier Kubernetes configurations. They will learn how to simplify their deployments, take advantage of reusable components, and integrate with other tools. They will be able to do this while reducing complexity by separating parameters from resources. Finally, we will review why GitOps is an important technique for ensuring your organization always knows what is in production. At the conclusion the audience will be able to immediately start integrating ksonnet into their stacks.

About Bryan
Bryan Liles is developer with Heptio. He currently works on the ksonnet project looking for ways to make configuring Kubernetes easier for users. Previously, Bryan was an early engineer at DigitalOcean, worked on disease discovery models, and has spent 20 years in the greater open source community.
Join us for KubeCon + CloudNativeCon in Barcelona May 20 - 23, Shanghai June 24 - 26, and San Diego November 18 - 21! Learn more at https://kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy and all of the other CNCF-hosted projects.

A

Alright, so we have 14 seconds before we get started. According to my computer I'd like to start this talk off with a little bit of a rant of traveling to foreign countries, with a touch bar iMac or a my Mac. I have a dongle over here. That is way bigger than my little clicker thing. I have a dongle over here in my bag, I have a dongle for my power, which is Donald again, so I can get this fancy European power. So this is a struggle to get here today.

A

So I hope you all appreciate the effort. I went through.

A

So the first thing I want to start off with is a lesson life lesson you want to write talks. You want to talk at conferences, you want to be like me. The first thing you need to learn how to do is make a provocative title. The animals for computers case on it is for people. What does that even mean? That makes no sense, but the people who were reviewing and I have reviewed for a lot of conferences, love fancy titles. They also like a lot of content whenever you're whenever you're talking about your synopsis.

A

So this is a little bit of a lesson. Make your titles a little bit provocative and you will rise to the least of the mid pack, so hello.

A

This is me. The reason I put my picture in my slides is because one day these slides will end up on the internet and I want people to know that this guy created them. So I'm, Bryan, Liles I'm, an engineer at hep. Do I write software I like cars and hip-hop, I notice, I, don't say software I do that for money, it's fun, but I like cars and hip-hop I code at Brunel on github and I talk smack and Brian Hill on Twitter.

A

So here we go so I think any project anything you're going to do like anything you're going to do you need to have a goal. If you don't have a goal, what are you actually aiming for? What is the end?

A

What is the end of your task so actually, for this talk, I put in some goals and we're going to talk about get ops and we're going to talk about managing any application and then we're going to talk about making application to manage or declarative application management easy easy when air quotes, because this is a Tech Talk. So this is not easy for your non-technical partner and then also democratizing the management of applications, as you notice, I use a lot of big words here.

A

This is a proper syllabus and but we'll get into it and I'll explain everything. So the first thing here is uh get ops. What is that anyone here know what get ops is actually I'm actually challenging you all. That's a made-up term. There is other than we've the term get ops. It's just the thing, it's something that you think it is and I think most of us in this room are on the same page. So basically forget ops. What we're thinking is, there's a person or whatever you know, black emoji, that they make a change.

A

They take that change and they put it in their git repository and they created a PR or whatever your system does is pull request, and then there is a review and approve a cycle and then, after that, it gets committed to whatever your master or upstream branches, and then it gets to magically applied to your your whatever that's what I'm thinking of get ops.

A

So basically it's taking something taking some kind of configuration describing it as a text and then making the updates to text and then reviewing it like you, review your software and then automatically applying it to wherever it needs to be applied. Just wanted to put that out there so on to the controversy yeah working with kubernetes is working with the mo, and it's funny my mind. My boss is literally one of the people who created kubernetes and I like the guy, but I look at him every single time like mo this is a long troll.

A

Yeah mo is actually great. It's and I'll go into why it's great later, but it's a long troll. So let me show you why it's a long troll, so we have as a deployment- and this is just a standard deployment I a few extra things in here just to make it a little bit more of a road for both. But what we're doing here is we're deploying this docker image called HTTP been everyone knows. Http, then, is right.

A

Well, if not, you will know in about four or five slides, but we're going to play this app to our cluster and we're done right. Well, of course not. We can't just deploy an appointment with the deployer service because we want to actually be able to talk to this web app that we've created. So we also go ahead and we create a we create this service and once again, services I really do enjoy a service configuration, but still it's more yamo and we're gonna put the service on port 80.

A

But now how do our users to get ahold of it? Well, we could have put it on a node port and and and let that go, but we did not do that in this case. We're going to define an ingress and the ingress type is not important here.

A

This actual code would work with nginx ingress, it would work a contour ingress and the reason I say contour is because hep do makes contour so I have to actually use it sometimes, but it's not important, but what we have here is this ingress and we're going to serve this site on HTTP bin dot.

A

My k8s and the secret here is that this is the example is on my mini cube the locally I've configured, my DNS, so anything in the my Kas on my case domain goes to where I think mini cube is at that particular time. So that's a deployment and just as a little thing here, I see a lot of people do it, but this is for posterity. You're going to see me run Kay a lot during this demo or during this talk. I want you to know that K is just an alias for cube control.

A

It's not an alias for cube, cuddle or cube CTL. It's cube control just putting that out there. So let's look at this I'm, a horrible, typer and I. If you look at my tweets, you'll notice, I miss words I misspell words, I'm actually I understand it, but between my brain and my finger, something that's my spinal cord is lossy or something so I lose out.

A

So all my demos here, I'm, not one I'm, not a Kelsie, Hightower I'm going to do this live I did this live and then I recorded the best, take and sped it up, because who has time for that? So let's actually do this, and this is just the. This is just an example of a deploy with with cute control and, as you can see here, I just do it really quickly and I. Do it over and over again it's just apply this directory. You know we've seen this.

A

This is actually the appeal of kubernetes I can actually take this declarative. Manifest of an out of deployments. Ingress service on disk I can run this one command with this dash F and you what do I get well I get this. This HTTP been website running on my local cluster on the mini cube. Not on this thing. This is just my travel Mac, but on my other other Mac at my house, and really what do we have here? Well, we have it running on that URL and we have this website and the reason I show.

A

This is because I think anyone who works in the web domain needs to know about this app. It's HD Ben, it's just HD, it's if you're doing things with HTTP you're gonna want to maybe spoof some traffic or maybe see if you can actually see some traffic.

A

So it's a great app and it's super simple super small, so I like to use it so I hope you all were paying attention to the manifests that I created earlier, because if you come to my talk, you're gonna get quizzed: um what makes face were those objects deploy to yell it out? It's fine default! That's why I didn't I didn't actually add one there, but what? If you wanted multiple versions of the deployment service and ingress in your cluster? Well, what could you do?

A

Well, you could be like hey Brian, you know: that's smart I'll. Take that directory I will copy that directory to another directory. I will go into that thing and then I will add a namespace to each one of those and I will do tube control, apply our apply F again you could. But what, if you wanted to do again? Well, Brian I could copy that directory to another directory change the namespace to whatever I wanted to, and do it again?

A

Well, yes, but it's that optimal I don't know, maybe it is for you I want to give you options. The next thing is what, if you wanted to deploy the service deployment and ingress to another cluster and here's the thing I love a lot of things about kubernetes somewhere under here, I have a kubernetes hoody I have a kubernetes patch I have lots kubernetes shirts, but one thing that I am NOT happy about in kubernetes LAN is the cube config file and not that it's bad because it works.

A

But the problem is: if you touch two clusters, it's great. If you touch 15 clusters and then you delete four of those clusters and then you add three more clusters and you delete one cluster, because gke makes it really easy to bring clusters up and down. What do you have? Well, you have a mess. So what if you want it's a ploy to another cluster? Well, you could actually just chain, you guys, go cube control and then you can go config and you can set another context and you can just apply problem-solve well. What?

A

If you wanted to use a different ingress setup in production versus development and what I mean by this is well what if you want it to? Actually, just change the ingress out like in my staging environment, I'm might be using engine X, but you know we want to make sure that we're. Yet we have contour and production, or maybe we're actually using engine X in our in our development environment and we're moving to contour. So we want to actually figure out how they use that.

A

How do we actually make those changes, or even better yet I used to work in a bank and there's all sorts of crazy firewalls that you can get at banks and in production we ran all sorts of web application, firewalls sometimes more than one, but we never ran that in staging. So what? If? How do you? How do you manage that configuration at the Floyd time and here's another one of my favorites? What, if you wanted to apply a node affinity rule to your deployment, node affinities are actually not hard.

A

You can do with them in like four lines of code, but most people don't get them right. The first time and note affinity is actually the easy one. What if you wanted to do a pod affinity? I cannot do that from memory, because I know that one of the keys is literally this long when I type it out on my computer, and it's like preferred during something something something something something you users don't need to know that that is actually a kubernetes thing and then also we'll get to this final piece.

A

What, if I, want to change this and generally what it is is my host is h3 Ben, not Maya, Kate's? What, if I want it to be something else? How do I actually change that and the answer is I- can just go edit, my files, but we're in declarative world now so we're not. We should not be doing it this way. There are better ways.

A

So real quick here infrastructure is code. I think we're all in agreement that this is a good idea. We saw what happened with the cfengine I mean it might predate some of you all, but we had some puppet. We had some chef, we had some salt and then we had some ansible. Those were all good. I actually really enjoyed the environments that I was able to build with those, but one problem that we came into is that will be realized before with all the ones that I just names up to where we are now.

A

We realized that procedural was a problem, whatever we were doing when we were doing our infrastructure deploys and not that it's bad or wrong, it just happens is that when you're doing a procedural, you tend to generate cruft over time. So we went to declarative and kubernetes went declarative, what their object files and then terraform once declarative as well. So just I'm not talking trash about ansible I, know red hats.

A

Here, I actually really do like ansible I use it, but it's not right for all problems and specifically where it's not right is for creating kubernetes objects. I enjoy that someone actually went out and created this stuff, but the problem is it only works within their context. It only works in that narrow band that you can do it, and also with terraform terraform is great for bringing up nodes and everywhere I like to talk about terraform, because I use my software.

A

So my software is on lots of computers, but when it comes to actually doing kubernetes notice that in every single case, where you're bringing a kubernetes objects, they actually have resources like kubernetes config map, they actually have kubernetes service kubernetes namespace. We don't want that. We want flexibility in our code. So what's the answer, all right, I'm gonna talk about case on it, but I want to show you case on it, real, quick here. So what I've done here is sped this up.

A

A lot is I've created this case on an app and I am deploying my deployment, my service and my ingress, and what have you noticed here is that, besides this last line here, because I'm I did this on purpose with this JQ weirdness?

A

What I've done here is I've touched no software and I'm not saying that I have this magical tool here. That can actually solve all your problems with a generator, because that's silly but here's the start, I want easy things to be easy, hard things to be possible, and what we're doing here is we're trying to remove the the need from yamo from your day to day not get rid of it, but just remove some of the need, so just to show you we have our site and I just want. This is more of my proof.

A

This is social proof that this all works. So what we've done here is this is a new site, new or URL different than another URL, but, back to my questions, what if you wanted to actually deploy multiple versions of this deployment, this ingress in this service on your cluster? Well, guess what JSON? It has an answer for that as well.

A

So let me show you something here once again: not changing any code with Cleburne case on it. We have this concept of environments. Really. What we're saying is that an environment is a cluster and a namespace. You can have n of them n greater than equal 1 with 0. It probably doesn't work very well, but what we can do is now that we know something about your cluster. We can, or we know something about your environments.

A

We can add a new one, so I added a new one in a name, same namespace, same cluster, and now what I've done here without changing any code without doing any any crazy thing, changing any mo? What we've done here once you? Actually, this is the Safari issue. I've moved to Safari recently because fire phox too much memory, but safaris brain-dead, so you actually got a type HTTP in there.

A

So what I'm, showing you is that I deployed this thing to my cluster again without changing anything, but now you're, saying hey, Brian you're, just showing me magic. This is all canned. What is this important so now I think I have some of you intrigued. Let me show you talk a little about what JSON it is. Okay, so on it is.

A

What we will be have is a case on an app I'm, not sure why we call it a case on it app if you actually can't execute it, but is a collection of things that you can use to describe your deployments to your kubernetes clusters, not just cluster. So generally, what we have or components and I'll go into that. We have libraries of code, we will pre generate some code for you or you could have code from somewhere else. We'll talk about that one in a second and then we have vendor code.

A

Vendor code is code that you may have created for another project or someone in the community created. You can actually use that will slightly go over that, and then we have this environments thing where we actually talk about the environments that we're working in so the goal here is really to do this. We want to take your application, break it down to the basic parts, so not just have one application in here. Maybe we have five applications in here and then we want to take those five applications.

A

Put them together in whatever way, seems relevant or is important at that point and then applying them to environments. So right now we have in this environment prod, and we have this environment dev and one more slide on this. What we would also like to do is inside of these components is like actually break it down. So in this case we have, we have an ingress, a deployment and a service and we're actually going to apply those and notice we haven't touched anything here.

A

So I raised these points earlier and but what case on it can do is it actually gets rid of the need to know what namespace the namespaces, the environment? You know you can actually remove that. You can remove that piece of variable code from your from your configurations. But what, if you wanted? Multiple versions? You see that I just deployed multiple versions. You can just create a new environment, deploy it.

A

What, if you want it's the deployment, the service Gress in another cluster once again use an environment and what, if you had a different ingress setup? Well, one of the power of Kasaan and environments. Is that if you define an application that has component a B and C in dev, what you could also do in production is define and you could define ABCD and E and use composition to bring them all together. So how does this all work? Well, I have this little gag that I like to run and I?

A

Don't I'm only halfway serious about this, but most of the things that we do in modern computer science come from lots of hard work at Google. If anyone wants to debate me on this, I have a long list and you probably go to the end and you'll realize it came from Google and well I'll. Give you an easy one. Kubernetes came from google, but json are actually another one Prometheus modelled left or something on Google. um A lot of these data things were doing model off, there's something on Google.

A

Not that Google is the smartest out there. They just have a lot of great ideas, but one thing they did is came up with this idea: around scripting, bork and I. Don't know what Borg configuration language looks like because that's kind of cheating but I do know that an engineer at Google put together something called JSON II, which is an embodiment of this board, configuration language. But what is it well?

A

Here's something that predates me at Ft, oh, why is case on it called case Ani I like to think just because we couldn't think of a better name but JSON and JSON it rhyme. They are similar they're, not the same thing, but let me show you some some JSON it here. So I have a few examples of what it looks like this is the smallest JSON. It document like I, think of think about JSON. It is basically an extension.

A

It's a data templating language on top of Jason, so any valid JSON objects or item or literal or whatever is valid JSON it so on the Left I have some JSON it on the right, it'll, either evaluate to Jason or Gamal, depending on what what's the output is, so this one I have an object: empty object left right.

A

Let's this one now I actually have a key and a value, and you notice on the right side that it actually rendered to a gamble, object. Json. It also has a standard library. I know know before I started working on this project. If I ever wondered, if Jason could have a standard library, but now that I see what JSON has in it, I'm actually wondering. Why doesn't my data libraries have standard libraries and the reason I bring? This up is because you have data.

A

We all will understand some principles of functional programming on data like just modifying data, creating something new JSON. It has this built in. So in this next example, what I'm going to do is I'm actually gonna.

A

Take this object and instead of spitting out the whole object, I'm going to spit out just the fields of the object as a list Lloris an array, and in this example, what I've done is I have two things here: I have a function and a conditional so now something that else I thought I would never want, but now that I have it I find it amazing. I can actually create functions. Inside of my of my Chase of my data and I can make that I take those functions and I can create new data.

A

So the example here is that I have I'm returning an array and one time I'm, calling it with hello with no arguments and other time I'm calling it what's an argument and those the outputs are different. Based on the conditional.

A

Another thing that I can do here- and this is where we start getting interesting- is we can take objects and then we can combine them with other objects.

A

So I have this o one which has a foo bar and then my output is a 1 plus this new object I'm using a little bit of syntactic sugar in here and and not in the lighting, the the plus symbol, but that's what I'm doing and as my final example and this in this realm here is I'm, actually taking an object and I'm taking and going inside of this object and I'm actually modifying a key inside of an embedded object so Brian. Why do I show this well?

A

What we do inside of Kasaan is we use these environments to actually build kubernetes objects. Remember your kubernetes objects are just shamble, so we're actually using constructs from programming and other things, not data, templating and but we're actually building new things. So the power that we're bringing to the table means that we can have a lot of things we can make. We can make hard transformations, easy and impossible transformations possible or impossible transformations possible so to bring this back to tube con.

A

What I have here on the left is a deployment it's pretty long, but just showing you that Jason definitely does the right thing, and just here is I'm actually now I'm introducing parameters where I have this parameter or I say my replicas are 42 and then inside of my my tip on my um spec I, actually say that look I can I can say that it's 42 and notice that when it renders it out, it just runs out the object. No more Jinja, no more go templates, I am actually I.

A

Just want to take a little sidebar here. I am very grateful for all the great projects that use go templates, but as a as a seasoned gopher go template, suck they're, not great they're great for I, don't know, but it's um I'm thankful that our our go friends created them, but I'm not gonna, use them anymore. So why do I talk like this? Well, here's something! Does anyone know what this this is? I'm sure some of you it's it's the it's just a spec! It's it's!

A

What's in swagger that Jason, if you actually go, look at it, it's the Oakland, API spec and what we've done and what we've done as a community as we've described all the endpoints inside of our kubernetes api and we've made them jason. Well, because we've done that what I can now do is take that and I can convert it to JSON II, and why would I do that?

A

Well, here's why I would do it now, and this is something this is something that we're exploring inside a case on Atlantis now, I can use code to actually build my objects and I'm not advocating this all the time. I actually prefer people to stay similar, it's a simple and use simple JSON base or something closer, but think about what this allows you to do and I have a great example.

A

So imagine if you were earlier I mentioned note, affinities and applying a node affinity to an existing object is not hard. Like I said, it's I think it's about four lines of EML that you have to go find and indent in. But what, if you wanted to do this, and what, if you wanted to make this available in a generic way? Well, what we can do is inside of JSON. It is.

A

We can actually write a node affinity, library and this actually, this files probably 100 lines long just to make sure that we capture all the cases. But then we can export this object as a function and then what we can do is we can add it as a node affinity and the best thing that we can do after.

A

That is that we can add another value as well, and this is one of the powers and notice I'm not talking about case on and I'm actually talking about the powers of JSON here, and this is why we've used JSON as the base of case on it, because what it does when you are freed from the burdens of actually editing Yambol, you can create some pretty interesting abstractions that can be applied in novel ways.

A

So answering that question, but moving on so what about the existing ecosystem, so I've been up here for 25 minutes talking about the animal is not great. It's hard we're using this JSON it thing well, why do I do that? Well, cuz. He animal is for computers and I did ask Joe beta about this and he balks every time. I say hey, you know.

A

Why did you long troll us with the Y Amal and he says: well, you know really yeah yeah yeah, but he says what else could we have done at the time and actually, we've done a lot with the Amal, so the animal is for computers. Humans can deal with it. But ya know is for computers, the animals, not bad, but it's not great. For me at least I can't speak for all the humans, but one thing that I will can look.

A

What I will say is that it's in heavy use- and it would be awkward for me to say that case on it- is this interesting project for doing all sorts of configuration for everything kubernetes, but then to not be able to actually use the existing ecosystem.

A

So I said you know what challenge accepted this right here is a config map in yeah Mille, it's arbitrary I, don't think you could actually use this for anything, but what I've done here is I'm going to show you something pretty interesting, and this is this is um case on it.

A

0.10 came out last week notice that you can list the components and then what I've done is, since all of our, we, like the dash F I added an important commands to case on it and now notice what I've done I've imported that llamo, what it did is it actually brought the camel in it, put it in the right place and the right component directory, and then it just it, and then it created an empty parameters that we could fill in later and that's all I did but notice that I can actually render Gamal with alongside my JSON II, and this is important, because any new project that comes along should be able to.

A

We have to be able to embrace the past if we're going to move forward in the future. We do not want to alienate so JSON our case on. It actually does have fully animal support. You can drop it in and it will actually render whatever you give it so moving along actually wants to talk about case on it in your environment. So remember earlier, when I was talking about the pulling case on it, see from your dev to your pride well, there's reasons why we like this.

A

A lot of companies do actually not deploy a lot of teams. A lot of projects actually don't deploy their their projects to just one place. They actually deployed in the multiple places and I used to work in this Bank and I'm not going to talk about it because they don't pay me anymore, so they get no name drops. But I worked at this Bank and every city.

A

We used something called CMDB a lot of enterprises, you CMDB, and what we had to do was he had a tag, every single asset with a project the owner and it's in its purpose. So what we did also put in the case on it is the ability to say, use an environment, not only use that to deploy to a particular location in a cluster or at or a particular namespace at a particular cluster.

A

We can also tag items all the items, so we have global labeling per environment inside of case on it, but we can also do even more I didn't create slides for this, but I want to talk about our registries. So I was talking about being able to create software for others, so with case on it registry, so you can actually create code like I created the node affinity stuff earlier. What you can actually do is take that code package. It up share it.

A

Other people can take advantage of it and the thing is, it doesn't have peered publicly. You can be shared privately as well. The best thing is the code that I didn't have to write. We have this concept in case sonic called prototypes, which are basically starter files. So if you don't want to actually write a deployment out, will write it for you and then you can fill out the rest of it it. It goes along to the point that some kubernetes is not accessible.

A

I actually look up at the point at most time when I'm, creating it a config map tube, so I, don't where all the keys we should be able to create some of that for you and then get you started on your way. It's all about leveling, the playing field, democratizing development on kubernetes. Another neat feature is that we can validate your objects before we apply them. We actually see if they're valid, you can apply invalid objects with cube control and it'll. Tell you kiss on it will as well.

A

But the neat thing is it'll tell you in the context of where it was applied, so I look at case on it as a project that was introduced some time last year, but what I'm really looking at case on it is where we can go next, and this next slide is a big one and what it is is we are going to in this next iteration. So sometime in June, we're gonna support, helm, charts and case on it, and this is a weird one, because before like Brian, why would you do that you're competing projects?

A

Well, if you say that I would actually say that you're not paying attention to where the industry is going. This is open source. We write software that we hope people will use, because it is useful. We are a lot of our phones run Linux. No one here is writing new operating systems anymore, that isn't Google or some something weird in Russia or China. No one's doing that anymore, Microsoft has even decided. You know what that Linux thing is not that bad.

A

um So what I've looked at is that I could actually go out and create this whole hierarchy, and all this stuff for importing like I, want to have a my sequel thing inside of my case on it project, I could actually create that infrastructure. Now what if I've done is I've actually fragmented, the community and really what in the apps community for kubernetes we're all trying to accomplish the same thing but we're all attacking it at different levels.

A

So what I wanted to do here say that hey, you know what the helmets I got a chart for everything and they're vetted, so we're gonna use this. The next thing we're going to do is have linting and the reason we want this is because, like I, keep on bringing up the the pod affinity and the reason, why is because a lot of issues with people maintaining pod affinities, they just type them in wrong? Even if you go to the kubernetes documentation right now, you might not get all the options.

A

If you read it long enough, I think you might get it but who but we're solving problems, we're not trying to figure out pod affinities. We actually always have a bigger problem so case on it. Well, actually, because we know the structure of your software will just lend it for you. We know what it looks like we and we can do that with yam. Oh, we can do that with helm charts. We can do that with JSON it and another thing that I want to introduce, and actually we have it working.

A

But it's not in the main line. Right now is the concept of overlays Google. Has a project I, don't know what it's called this day easily called conflate I'm, not quite sure what it's called right. Now but customize is it with a case still alright customize with the K Google? Has this thing called customize with the K and the novel feature of it is that it can overlay? You create you, you start with your Y Amal and, depending on the context you can actually overlay case on it.

A

Has this ability where we could actually say I want to actually change. All I wants to actually add a new pod without an initialization controller to ever a sidecar to everyone. We could just overlay it in there. You can basically overlay anything and then also deaf. What is the difference between what I have on disk versus what just in my cluster?

A

We will be making that amazing and the last two items here are obvious code integration. We do have a vs code plug-in it's, we wrote it for JSON it, but not for case on it. Next month we will be releasing AVS code for extension for case on it, which is first class, and the final thing here that I would like to add is a case on. It is notice that I was I, didn't go in the case on it case on.

A

It is like a two hour topic: I had half an hour, so I just want to talk about. I was trying to actually show the vision of where we're trying to be about declarative management, but really what we want to do is do a case on it. Spec talking about really when you're talking about the clarity of applications. What does it look like?

A

How do you actually build them, and then also, instead of leading with your problems, will tell you problems that we think case on it's resolved and then we can point your direction from there. So this is meant to be an introduction of case on it, just showing you that hey, we are actually out here. Thinking about this problem and actually developing tools that are for the user. I invite you all to go to their website. But, more importantly- and this is this- is the most important fit. This is about people.

A

This is about software, about helping people. This is not software to get github stars or anything like that. So come to get case on it, slack I'm, Brian Elam there I will meet with everyone who wants to talk with me to tell me what's wrong with declarative management and applications and kubernetes, because I'm interested in the problem, not the software. The software is only a tool. So please use me to help you be better and that's all I can ask. Thank you.

A

So we have one minute: I can take two questions. If any one right here and I'll repeat it.

A

So the question was: can I deploy custom resources which case on it? Yes, actually you can you, if kate, the custom resource, so you needed your own controller, but a custom resource is only a mo, so you can deploy that we can manage. It doesn't really matter to us right now. You won't have all the great introspection, because getting that schema back is not something that it's built inside a case on it, but we're looking at how to do that. We're currently working with the cube flow team.

A

Actually, if you go look at their demo in the google booth they're using case on it right now and the our go team and the workflow team, the the workflow tool from intuit and athletics, those are both using it and I have certain manager to play it all over the place using crts. One more question right here.

A

Why don't I use? What do you say fabricate.

A

So is fabricate that tool that was used for generating configurations before is it fairly old or I'm are thinking of something else.

A

So why don't we use? Why don't we use it? I actually don't have a great answer. We just don't use it, but if you come up here and tell me a good reason, I I will actually go look at it and then I'll have a great fronts for why we don't use it. So, thank you all I know I'm standing between you and drink time come talk to me now or when you were all out drinking later I'm here to answer your questions. Thank you.