Cloud Native Computing Foundation KubeCon + CloudNativeCon Europe 2021, 15 May 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Sponsored Session: Huawei - Zero Change Transitioning to Multi-Cloud Architecture

Description

Don’t miss out! Join us at our upcoming event: KubeCon + CloudNativeCon North America 2021 in Los Angeles, CA from October 12-15. 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.

Sponsored Session: Huawei - Zero Change Transitioning to Multi-Cloud Architecture

Speakers: Hongcai Ren

Multi-cloud has become the dominant enterprise strategy, but cloud-native multi-cloud is challenging: cumbersome and repetitive setup due to too many clusters, fragmentation of workloads, limited by boundary of clusters etc.
In this presentation, Hongcai will demo how to use kubernetes native api and a set of attached policies to manage workloads across clouds with automatic HA spreading, fault-migration enabled.

A

Hello, everybody I'm home tyrang from huawei and my github id is rainbow mango.

A

Maybe you have to manage multiple clusters to support development, testing and all kinds of business, and maybe you are using cncd2 to handle that stuff today, I want to show you a different system that will make this thing easier with the system you can use and manage multi cloud. Just like a single cluster, you don't need to change your infrastructure and don't need to change your application configurations.

A

The system is gear moda. It's an open cloud native and multi-cloud orchestration engine.

A

The mother project aims to of ability to easy to easily use and manage multiple clusters. People use multi clusters just like a single kubernetes cluster before the demo. I want to give you a quick overview of the architecture and the main concepts.

A

The main components of the project include kiamata api server, kr motor controllers and nkl mother scheduler. The key amount. Api server is essentially a cube api server, so people can use control, command or client go to xx8.

A

Okay, kmart api server used to store users, workload, configurations and kmart api objects.

A

It takes e2cd, as is back-end the key among the scheduler in charge of select clusters for resource template according to policy apis, the care model provides several apis that implemented by care model controllers.

A

The cluster api that implemented by cluster controller, provides a pre, provides a representation of the mimo cluster. That's registered to care moda.

A

The propagation policy api that implemented by policy controller, provides a representation of the rules where results templates through the propagation tool. The resource binding api that implemented by binding controller, is an internal api which binds a resource template and a list of targeted clusters.

A

The work api wraps a resource template. The execution controller will apply resource template to memo clusters.

A

The override policy api provides a occupy competitive compatibility to override resource templates before they propagated to memory clusters.

A

Here is an example of a propagation policy. It represents a policy that propagates all deployments that labeled with multi-zoom replication and the target clusters should locate in three zones.

A

Here is the example of override policy it represents when a deployment should be propagated to clusters with label. Reading dc1.

A

The key amana controller should replace its image registry.

A

This is a resource. Template example, don't be surprised, it's exactly the same yaml that people applied to kubernetes people can use group control command to apply it to kmart api server.

A

Okay, I'm going to show you how to join cluster to clear mother and then apply a result, template and then apply a propagation policy to clear mother. To the last we will see how to check status from kia mada.

A

Here I prepared a german system and a kubernetes cluster named member one: okay, let's register the member one cluster to kill mother.

A

Before we join the club, the cluster to kill mother, we choose the setback, fun, config, first to point to mother api.

A

A

Okay, we can see all the api that.

A

Is started by camada.

A

Here is the cluster resource when, uh when we join a cluster okay, our mother will create a classic object for it.

A

Now, let's join number one, um we have a kia mother kingdom come on. We can use it to join member one with its group.

A

A

A

Check, yes, the cluster has been joined, they can see their version mode and.

A

A

Here is a more detailed status of the one cluster about the r api, enablements and.

A

Know the summary information.

A

Okay, now I'm going to deploy a propagation policy.

A

A

A

Examples and deploy progression policy yeah. The policy shows that it will deploy a deployment, a development named index to to cluster number one. Okay, let's apply it.

A

Yeah now I'm going to deploy a result template let's check.

A

It's a simple deployment.

A

Now, before we deploy the deployment deployment, we can watch the development on member one cluster yeah. First export could config.

A

Yes, that's what we.

A

Expected yes, yeah: let's apply the deployment.

A

Here after we deploy a result, template um the clear mother.

A

If, if there is a policy matches the the results, template the result, template the will be propagated to mama clusters and after that, okay, ahmad aps kia mada will collect status from memory cluster and we can get cluster guide deployment status from kia mother yeah and after that. If, if, if wait, the avoid deletes the development, the resource.

A

If it deletes the development, the demand will be removed from my memory classes.

A

Yeah. Okay, thank you! That's the demo. Okay, that's all welcome to try it out by yourself and if you have any questions, please feel free to contact the materials by the github issue or slack. Thank you. Bye.