Continuous Delivery Foundation cdCon 2021, 6 Jul 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: No K8s Expertise? No Problem. How to Use Open Source Ketch to Deploy Cloud Native... - Bruno Andrade

Description

No K8s Expertise? No Problem. How to Use Open Source Ketch to Deploy Cloud Native Applications Without a K8s Skillset - Bruno Andrade, Shipa.io
Speakers: Bruno Andrade
Ketch is a new and fully open source project for deploying and managing cloud native applications on any Kubernetes cluster *without* writing YAML configuration files. While Kubernetes is now the gold standard orchestrator for containerized and microservices-based applications, operating it requires overcoming a steep learning curve. Traditionally, successful Kubernetes deployments require developers to become experts in Kubernetes concepts, objects, and YAML file writing and management. Open source Ketch eliminates this complexity by deploying applications directly to Kubernetes clusters – rendering YAML files unnecessary and flattening the learning curve for developers. Ketch enables developers to focus on writing code rather than on infrastructure, and to deploy and run applications on Kubernetes with no expertise necessary.

For more Continuous Delivery Foundation content, check out our blog: https://cd.foundation/blog/

A

Hey folks, thanks for joining me here at citycon today, I hope you're enjoying the event. The purpose of this talk today is to talk a little bit about how we can make the deployment and management of our applications easier on kubernetes right. I think we can all agree that kubernetes is pretty much the de facto engine when deploying and managing your applica your containerized applications today, and that a lot of organizations they are either starting or they're scaling the kubernetes initiatives.

A

But one thing that remains open is: how can we improve the developer experience and productivity kubernetes is awesome is a strong tool, but it also introduces a lot of challenges, especially when it comes to the development side. A little bit about me before we get started. um I'm assis admin and I say it's his admin because I had an opportunity to start um back when we were dealing with a uh bare metal service uh servers and then vms and so on and so forth, creating shell scripts to automate stuff.

A

We didn't have any of the terraform or even ansible, and any of these things um I'll say I'm a lover of pizza and I'm a lover of music. My original goal was to be a musician. I realized. I was definitely not good enough for that. I was not going to cut it, so I decided to go into cis admin and software engineering.

A

Sometimes I still think I'm not good enough as well, but can't make the change now and I really like to work on um making things easy right, especially when it comes towards developer centric solutions.

A

I think we already have to deal with a lot during our day to day and deploying and managing our applications should definitely not be in our way as as a complex task, and if we look at the uh the the adoption story for kubernetes right, we see this across a bunch of companies. um That's how it starts for a lot of folks. um You, you see people talking about kubernetes everywhere, right and you're like well, I'm excited! I, I gotta try this thing and when you talk to your teammates, did you do you?

A

Have you use kubernetes? Do you like it and there the usual response is. It is awesome because you get so many things for free, you get service, uh discovery, scheduling, self-healing and, and so many other things. You should definitely jump into this bandwagon.

A

You decide to give it a try and you spin up a kubernetes cluster in your laptop or on a cloud managed service. You deploy your first uh hello world application, and- and this is easy and you're like I'm I'm doing great and- and I should go deeper into that- I really meant I really nailed this kubernetes thing and but, as you start bringing your systems then or your applications are more complex applications. That's when you run into that moment that you start learning about deployments.

A

Services are back pvc and so many other things, and that's the awakening moment right. If I start bringing it into kind of a day-to-day task. um You have your happy developer. um My organization is uh adopting kubernetes, I'm a developer here in the team. I want to jump into that bandwagon.

A

um I have a website I'll start with a website. It shouldn't be too complex um and I'll get it deployed on kubernetes you'll copy, someone's helm chart and you look into the chart. Yellow file looks easy. It's pretty much a description in a version and then you open values, yaml file here you start seeing new concepts where you're like well, okay repository it's my image, I I know where my image is located, but what are the resources that I should assign to my application and what is readiness probe and what is k native deploy?

A

What what are all these things and you kind of work your way around it? But, as you start digging deeper and you go into the deployment emo file, we're like wow wait. This is not as easy as expected right and, as you start keeping going uh as you keep going through your uh your helm chart. Basically you go from being that developer. That is optimistic about implementing kubernetes to basically this. Why are we even doing this thing right? So an easy task for us just deploying a website was super easy.

A

Now it becomes super complicated, deploying and managing my my website and the updates and the the end of the data scenario that ends up is that we as developers, we are no longer just applying our application code, but now we have to learn: terraform, helm, charts, kubernetes objects and the deployment of applications that were easy before now becomes super complex on the developer side and managing this at scale becomes even harder right and the scenario that we see folks looking for out there is, I want to deploy my application doesn't matter if I'm, if I'm deploying it from get from ci pipeline or from a terminal.

A

But I need something in between my kubernetes and my pipeline or my application code, something that allows me to deploy my application, like I use, deploy at least as close as possible to the way I used to deploy my applications before and not have application deployment and management as a complex task and a barrier in the way, and that's where catch comes in as a project catch allows you to do exactly that.

A

Catch allows you to deploy your applications and we're going to go through a demo right now, but deploy your applications from git pipeline or your terminal in a way that you are focusing on your application code rather than creating and maintaining a bunch of kubernetes related objects.

A

Catch in turn receives your application, coder application image. It builds all the objects automatically and deploy these objects in the cluster, and it gives you automatically visibility into your application: how to access it. The status of your application without having to learn all the complexities of kubernetes. So let's see it in action all right. So, let's see what we have here. um I I'm going to be using a gke cluster um for this demo. You can use aks, eks or mini key or k3s. For example.

A

Catch is not limited in any ways to gke, I'm just using that for the purpose of the demo. Today, I'm going to be following the documentation available on learn, dot, the catch, dot, io and it is available and open here and we're going to be basically going through the process of creating a pool, deploying an application and accessing that application.

A

If we follow through the the getting started, the first step is to install the catch cli. I already have cache cli installed here on my machine, so I don't need to do that.

A

The next step is to install cert manager and ingress traffic or istio, for example, and cluster issuer.

A

In my case, since this is an existing cluster, I already have istio install and I have cluster issuer insert manager already in place. If that's your case as well, you don't need to do that. You can use the existing ones that you have in your cluster. If it is a brand new cluster, then you can install it by following these uh these instructions.

A

The last step for for you is to install the catch controller, which you can do directly from this command. Here. I already have catch as a controller running here locally, so I can see um if there are any existing applications. In my case, I don't have any application so great for you to install and start using catch, then cli first step.

A

If you don't have one install cert manager, ingress and cluster issuer. If you do, then you're gonna see how you can link catch back to those and install the catch controller. It's a fairly easy process and quick. So, let's see catch, has a concept of pool a pool, basically maps to a namespace in your cluster. They are the same thing: um the why catchy uses pool for your namespaces, so you can limit where your applications and the objects are being deployed, and I'm going to be deploying this application from my cli.

A

But we can also see how this can be deployed directly from your ci pipeline as well. The first step is add a pool to catch so catch understands and connects to a specific namespace in your cluster.

A

For us to do that, we're gonna need to give catch the ingress service endpoint.

A

A

A

Okay, so I have my istio ingress service endpoint right here, so let's do that catch pull, add dev.

A

And then my ip is going to be this one and ingress type istio. Just so. Catch knows how to connect to it.

A

Istio there you go. If we do catch pull list. For example, we should see we have a catch, uh a pool called dev and it's in a status created, and you can see that when you create a name, a pool on catch, it automatically creates a namespace for you in your cluster.

A

Next up for us is to create the application itself. So before we can deploy our image, we have to create an application. So catch knows that an application code will be deployed there. So we're doing catch, app, create bulletin board and we're saying it should use the pool dev, the pool that we just created.

A

If we do catch app list.

A

I can see that I have my bulletin board here and it's in dev the style. The state of my application is just created because we have not deployed anything yet, but I can see it already gives me an address here.

A

Now we're going to do the app deploy right, I'm just saying catch app, deploy the name of the app that we created bulletin board and where my image is located.

A

Great now we can see that catch will start the deployment automatically. You can see right away that I haven't created any deployment file, any service file or anything else from a kubernetes perspective me as a developer. I'm just pushing my uh my application image to catch and catch in turn will work with a uh with kubernetes on the back to have our image for any.

A

If you see cube, ctl get name space. For example, you can see catch dev, which is the name space tied to the pool we created with catch, and if we do cube, ctrl get pods catch dev we can see the application is even already running if we do catch app list. For example, now we see our application is running.

A

If I get this address and let's see if I try to go to my browser.

A

There you go, I see my application running and again, I didn't have to do any kubernetes deployment or or service files or understand so now, you're getting your developers to be able to deploy their applications quickly and maintain those applications quickly. They can go further and add things such as add a cname to their applications.

A

Even though catch gives you this address by default, you can actually add, for example, catch cname and, for example, you can say app1, dot, cat, the catch dot, io and then the name of the application, which is bulletin board in this case, catch automatically, connect to istio, create the uh the certificates, apply and add that cname to your application as available that's great now, you can quickly deploy your applications and now with cache, you have a wrap, a replicable environment that you can deploy using catch in a local cluster and you can have cache installed in your company's cluster or in your team's cluster.

A

So the deployment of your images are exactly the same. You don't need to customize deployment files or yellow files or anything like that to have your applications deploy. So you develop it once and you test it anywhere and you deploy it anywhere with catch, but we know that you, don't you don't apply only from your cli right. You also deploy your your applications from your ci pipeline.

A

um Let me try to find one here.

A

Great here is here's an example: I'm running catch with my github actions, but the same thing applies to um any other ci pipeline that you have running. My github actions is basically pulling from master. It's it's building my image.

A

It's installing requirements that I have and it's storing my doctor image in a google container registry and as soon as this process is done, I'm actually setting up catch, which is basically installing the uh the cli during the pipeline and down here, I'm adding a pool using github actions secrets, I'm passing the name of the pool, the ingress ip. Basically, what we did before through our cli I'm creating an application dynamically from my ci pipeline and I'm deploying that image right.

A

So you can do exactly the same thing from your pipeline that we just did from our cli. So now you can get your developers or your team to deploy your application consistently either from your pipeline or from your cli. For example, you have that repeatable process and again you're focused only on your application code rather than on kubernetes xml files, so focusing back again on developer productivity or or the developer experience itself.

A

There are many other things that you can do with catch if you're deploying specific applications that you need deployment hooks, for example, or if you need to tie specific ports to your application, so they can run properly on kubernetes.

A

You can do this directly from here using a catch.emo file that you can pass along with your image when you're running deployments, but this this is also available to you in your team. You can find more information about catch directly on catch github.

A

It is available here, like the project, would love to see you collaborating and participating, sending questions, sending ideas as well as you can learn more about catch directly from the cache documentation and join our community would love to have you and participate, but again, let's see how together we can improve the developer, experience and developer productivity, I'm always open to any questions, and I really appreciate your time today. Thank you.