简体繁体 English

有关在Google容器集群内使用的节点的困惑

[英]Confusion about nodes used inside a Google Container Cluster

原文 2016-03-18 21:35:19 9 2 google-app-engine/ containers/ kubernetes/ google-kubernetes-engine

When a Google Container Engine cluster is created, Container Engine creates a Compute Engine managed instance group to manage the created instances. 创建Google Container Engine集群后，Container Engine将创建一个Compute Engine管理的实例组来管理创建的实例。 These instances are from Google Compute engine, which means, they are Virtual machines. 这些实例来自Google Compute引擎，这意味着它们是虚拟机。

But we read in the doc page: "VMs are heavyweight and non-portable. The New Way is to deploy containers based on operating-system-level virtualization rather than hardware virtualization" isn't a contradiction? 但是我们在doc页面中读到：“ VM是重量级且不可移植的。新方法是基于操作系统级虚拟化而不是硬件虚拟化来部署容器”不是矛盾吗？ correct me if I'm wrong. 如果我错了纠正我。 We use containers because they are extremely fast (either in boot time or tasks execution) compared to VMs, and they save a lot of space storage. 我们之所以使用容器，是因为与VM相比，它们的运行速度非常快（无论是在启动时还是在执行任务方面），而且它们还节省了大量空间。 So if we have one node(vm) that can supports 4 containers max, our clients can rapidly lunch 4 containers, but beyond this number, gcloud autoscaler will need to lunch a new node(vm) to support upcoming containers, which incurs some tasks delay. 因此，如果我们有一个最多可支持4个容器的节点（vm），我们的客户可以迅速吃掉4个容器，但是超出此数目，gcloud autoscaler将需要吃一个新节点（vm）来支持即将到来的容器，这将导致一些任务延迟。

Is it impossible to launch containers over physical machines? 在物理机器上启动容器是不可能的吗？

And what do you recommend for running critical time execution tasks? 您对运行关键时间执行任务有何建议？

2 个解决方案

It is definitely possible to launch containers on physical machines. 绝对有可能在物理计算机上启动容器。 In fact, according to the Borg paper ( the design of which heavily influenced Container Engine/Kubernetes ), this is the norm within Google's own infrastructure: 实际上，根据Borg的论文（其设计对Container Engine / Kubernetes产生了重大影响），这是Google自己的基础架构中的规范：

Each task maps to a set of Linux processes running in a container on a machine [62]. 每个任务都映射到在计算机上的容器中运行的一组Linux进程[62]。 The vast majority of the Borg workload does not run inside virtual machines (VMs), because we don't want to pay the cost of virtualization. Borg的大部分工作负载都无法在虚拟机（VM）中运行，因为我们不想支付虚拟化的费用。 Also, the system was designed at a time when we had a considerable investment in processors with no virtualization support in hardware. 此外，该系统是在我们对处理器进行了大量投资而没有硬件虚拟化支持的时候设计的。

Since Container Engine is hosted within GCP, VMs are used to facilitate dynamic provisioning. 由于容器引擎托管在GCP内，因此使用VM来促进动态配置。 However, these VMs are long lived compared to the lifetime of containers scheduled onto them. 但是，与计划在其上的容器的生存期相比，这些VM的生存期很长。 Pods of containers may be scheduled on and off of these VMs and jobs run to completion. 可以在这些VM的上/下计划容器的容器，然后作业完成。 However, VMs are torn down when clusters are upgraded or re-sized. 但是，在升级集群或调整集群大小时，VM将被拆除。

Kubernetes can be installed on both virtual and physical machines (there are multiple getting started guides for bare metal ). Kubernetes可以安装在虚拟机和物理机上（有多个关于裸机的入门指南）。 Google's Cloud Platform only offers virtual machines as a service which is why Google Container Engine is built on top of virtual machines. Google的Cloud Platform仅提供虚拟机作为服务，这就是为什么Google Container Engine构建在虚拟机之上的原因。