Cloud Native Computing Foundation KubeCon + CloudNativeCon North America 2017 (Austin), 15 Dec 2017

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Keynote: Manage the App on Kubernetes - Brandon Philips, CTO, CoreOS

Description

Keynote: Manage the App on Kubernetes - Brandon Philips, CTO, CoreOS

Kubernetes has yet to close the developer gap from source code to app running in a production Kubernetes cluster. Many build bespoke tools. How can the Kubernetes community come together to build decomposable solutions that help people define their app, deploy it, and manage its lifecycle over time? Learn about the progress we are making together to elevate the conversation from container orchestration to application lifecycles management.

About Brandon Philips
Brandon Philips is helping to build modern Linux server infrastructure at CoreOS as CTO. Prior to CoreOS, he worked at Rackspace hacking on cloud monitoring and was a Linux kernel developer at SUSE. As a graduate of Oregon State's Open Source Lab he is passionate about open source technologies.
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

Wow zoo-oo dia, so I'm gonna be talking about managing applications on kubernetes, I'm, Brandon, Phillips, CTO of core OS and I have five minutes. So the goal here I think in kubernetes is to make it easier to build, deploy, operate, upgrade decommission applications over time and to ideally do this with minimum toil removing the grudge rework of getting this getting your application in production.

A

Now, there's some good news and some bad news. The good news is that we're making it easier and easier to get apps into production. The bad news is that there's going to be more and more apps that we need to manage in production, and this is kind of an economic effect that is was observed during the Industrial Revolution. The price of coal went down, and then people started to use more coal, and this is holds true and pretty much any economy.

A

If you lower the prices people will people will use more of it if you make it easier, they'll use more of it and I think the fact that my voice is echoing in this room of all of you indicates that there is a lot of demand for making deploying apps easier, so we're gonna end up with more apps.

A

But if you have a lot of apps there's a lot of things that you need to be accountable for what versions of those apps exist where they deployed? Are they healthy or not? Who actually owned me out? Who put this on the cluster? There's just a lot of questions that you need to answer over time and pre kubernetes apps made it into production by being bunch of binaries from scripts and then boom production and in reality it look more like this, which is a bunch of binaries some scripts and then what exactly exists in production.

A

Our current state is that we have a bunch of containers, some manifests and then those containers end up in production, but still it's really just a bunch of containers all over the place across our kubernetes cluster and then there's all these questions of where the dashboards for this thing, who owns it, where the docks who has the SMAS and it's usually on a really poorly maintained, wiki page internally or just tribal, and all where you like. How can somebody tell me where the health monitoring is for this app?

A

And so what I'd like to see and what a lot of us in the kubernetes would like ecosystem would like to see is a state where we start to define. Containers Plus manifests are a distinct version of the application, and then those versions ends up in a catalog where you say this is the app it's some internal name and the app has a number of versions and then you're able to say these instances of that app tied to those versions. And so it's a lot less confusing.

A

When you can go back and say all these containers map to this version of this app, which was was built in this way, and then you can start to do things like oh I need to have a temporary penetration test environment or a temporary debug environment, and then what we'd like to do is tie all those things together with all the metadata that makes an app an app and not just a bunch of containers. So a quick demo here.

A

This is a tectonic console and we've created a catalog of services called open cloud services and we're using this idea. So you come in and you enable apps like vault, and this is all driven via the kubernetes api. It's just. The console gets a little old after a while doing demos, and we can do things like resolve dependencies, so vault depends on @cd, so we deploy those things and immediately inside of the namespace where I deployed them.

A

We see the couple of base things required to spin up these vault instances, and then this application is available on-site inside of this namespace, and we can do things like configure an instance of this application, click create and it will go off and deploy all of the pieces and parts necessary in order to get a vault application going, and so we've deployed that and now we configured it and now all the pieces and parts to get that app up and running are deployed, and the cool thing is is that we can tie all these things together.

A

We can start to say well what are the all the resources in kubernetes that exist for this instance of this app and similarly we can start my things together, like inside of the TV cluster Wordle, the pods that exist to support that a TV Custer on this app. So this is kind of a vision of what we see happening in the future, where we use kubernetes the whole. All this metadata together and we want to create a shared toolkit using kubernetes api is accessible via coop cuddle that lets.

A

You move all this stuff together, the catalogs and the instances that I just showed you are actual real kubernetes resources and wait. What about helm and everything else? I think these are solving good problems, but by creating a toolkit instead of tools. I think we're able to solve this problem in a better way, and we can answer these questions in a consistent way.

A

If you want to get involved, there's the app theft working group that is starting to tackle some of these fundamental problems and as well say gaps if you'd like to chat more about this come see us at the chorus booth and thanks.