Shulou(Shulou.com)05/31 Report--

Most people do not understand the knowledge points of this article "what are the characteristics of tensile-kube?", so the editor summarizes the following content, detailed content, clear steps, and has a certain reference value. I hope you can get something after reading this article. Let's take a look at this "what are the characteristics of tensile-kube" article.

The origin of the name tensile-kube

Tensile is called extensible, extensible, and tensile also contains the meaning of tencent and tenc.

Background of tensile-kube birth

Tencent Games Tenc (container) computing platform has dozens of K8s clusters, these clusters have some fragmented resources, which can not be used effectively. The common scenario is that a job needs N resources, but the remaining resources of existing clusters A, B, C and so on do not satisfy N, while the sum of resources of cluster A, B, C can satisfy N. In addition, we have another scenario where a business has one type of business that needs to be published to multiple clusters at the same time.

In fact, the industry has a similar demand for a long time, but also gave birth to Federation and other programs, but these for the existing K8s will have a larger transformation cost, while adding a lot of complicated CRD. Therefore, we independently developed a lighter virtual-kubelet-based solution, tensile-kube.

Https://github.com/kubernetes-sigs/kubefed

Introduction to virtual-kubelet (VK)

VK is a basic library open source with Microsoft, and its implementation is often called provider. Early VK included a variety of provider, but now the VK community has separated it into virtual-kubelet, node-cli and provider. Virtual-kubelet provides an abstract capability. By implementing provider, the user implements a virtual K8s node to which the user's Pod can be scheduled. The core function of virtual-kubelet is the life cycle management of Node and Pod.

Tensile-kube architecture

The core components of tensile-kube are 4:

Virtual-node

The function of K8s provider is realized based on virtual-kubelet, and a number of controller are added to synchronize ConfigMap, Sercret, Service and other resources.

Webhook

Convert the fields in Pod that may interfere with the upper-level cluster scheduling, write them into annotation, and then restore them when virtual-node creates the Pod, so as to avoid affecting the scheduling results expected by users.

Descheduler

To avoid the impact of resource fragmentation, we have redeveloped this component based on the community's descheduler to make it more suitable for this scenario.

Multi-scheduler

It is mainly used to connect the API Server of the lower-level cluster to avoid the impact of resource fragments and to determine whether the resources of the virtual-node are sufficient during scheduling. (if it is not shown in the figure, the multi-scheduler will be heavy and is not recommended. If you want to use it, replace the upper cluster kube-scheduler with multi-scheduler. )

Characteristics of tensile-kube

No intrusion to native K8s cluster

In tensile-kube, the connection between upper and lower clusters is established through virtual-node, and the upper Pod is dispatched to virtual-node through the scheduler, and then the Pod is synchronized to the lower cluster to create. The whole scheme does not invade K8s.

The reconstruction cost of K8s upper platform is low.

From the point of view of the upper K8s platform, the cluster connected through tensile-kube is a large cluster, which can completely shield the differences, so there is almost no need to modify the platform.

Based on native virtual-kubelet development

Tensile-kube is based on native virtual-kubelet, without too much extra dependency, while maintaining many features of native virtual-kubelet, making it easy to update and upgrade.

These make it easy for tensile-kube to implement the scheduling management of multiple clusters, and it is suitable for most scenarios.

Main capabilities of tensile-kube

Integration of cluster fragment resources

For offline clusters, there are more or less resource fragments. With tensile-kube, you can synthesize this write fragment into a large pool of resources, making full use of idle resources.

Service multi-cluster scheduling

Combined with Affinity, we can realize the multi-cluster scheduling of Pod. For clusters of 1.16 and above, more fine-grained cross-level group scheduling can also be realized based on TopologySpreadConstraint.

Service cross-cluster communication

Tensile-kube natively provides the ability to communicate across clusters over Service. But the premise is: the networks between clusters can communicate with each other through Pod IP, for example, private clusters share a flannel, and so on.

Convenient operation and maintenance cluster

When we upgrade or change a single cluster, we often need to notify business migration, cluster shielding scheduling, and so on. With the support of tensile-kube, we only need to set virtual-node to impossible scheduling.

Native kubectl capability

Tensile kubectl supports all native kubectl operations, including kuebctl logs and kubectl exec. This is very convenient for the operation and maintenance staff to manage the pod on the lower cluster, and there is no additional learning cost.

Other possible applications

Edge computing

Hybrid cloud

The above is about the content of this article on "what are the characteristics of tensile-kube". I believe we all have a certain understanding. I hope the content shared by the editor will be helpful to you. If you want to know more about the relevant knowledge, please pay attention to the industry information channel.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.