GPU Resource Management On JDOS 梁永清 liangyongqing1@jd.com 提供的服务 1. 用于实验的 GPU 容器 2.基于 Kubeflow 的机器学习训练服务 3.模型管理和模型 Serving 服务 Experiment Training Serving 均基于容器，不对业务方直接提供 GPU 物理机 GPU 实验 JDOS 常规的容器服务

placement of pods. This is used to spread pods across availability zones, while still respecting resource access and availability concerns. When Kubernetes runs on vSphere, the hypervisor platform also scheduling and resource management are active. Currently no automatic scheduling integration occurs, that is, Kubernetes is not aware of the underlying vSphere topology (sites, affinity groups, NUMA, etc etc.). This session will explain the options to gain better performance, resource optimization and availability through tuning of vSphere, and Kubernetes configuration and labeling. This is applicable

placement of pods. This is used to spread pods across availability zones, while still respecting resource access and availability concerns. When Kubernetes runs on vSphere, the hypervisor platform also scheduling and resource management are active. ​Currently no automatic scheduling integration occurs, that is, Kubernetes is not aware of the underlying vSphere topology (sites, affinity groups, NUMA, etc etc.). ​This session will explain the options to gain better performance, resource optimization and availability through tuning of vSphere, and Kubernetes configuration and labeling. This is applicable

#输出初始化配 置文件并编辑 # vi /etc/kubeadm-init.yaml apiVersion: kubeadm.k8s.io/v1beta2 bootstrapTokens: - groups: - system:bootstrappers:kubeadm:default-node-token token: abcdef.0123456789abcdef �l: 24h0m0s #输出初始化配 置文件并编辑 # vi /etc/kubeadm-init.yaml apiVersion: kubeadm.k8s.io/v1beta3 bootstrapTokens: - groups: - system:bootstrappers:kubeadm:default-node-token token: abcdef.0123456789abcdef �l: 24h0m0s v3.26.1/manifests/custom- resources.yaml #创建控制器 Install the Tigera Calico operator and custom resource defini�ons # kubectl create -f �gera-operator.yaml #如果需要更改镜像，可编辑此 文件 # more

pod ● Global search ● Login mechanism ● Settings page ● Support for Cron Jobs ● Redesigned resource creation ● ...and much much more. github.com/kubernetes/dashboard/releases In-progress work ly/securing-dashboard Securely running Dashboard is possible! “We operate a cluster with multiple end-user groups. Many of them would find the Dashboard extremely useful, but if we could ... have them [log in] huge win.” → Survey response → Cluster Operator, running Kubernetes in GCP and on-prem ● Custom Resource Definitions support ● Service topology view ● Mobile device support ● Cost estimates ● CI/CD

& KubeEdge� 华为公有云智能边缘平台IEF以开源的KubeEdge为内核打造� KubeEdge基础架构� KubeEdge设备管理理——Kubernetes Custom Resource Definitions (CRD)� 应⽤用场景：智能家居设备云端托管� KubeEdge设备管理理——从云端同步期望设备状态到边缘� 应⽤用场景：智能家居设备远程控制� KubeE Official Website：https://kubeedge.io� ➔ Slack：https://kubeedge.slack.com/� ➔ Mailing List: https://groups.google.com/forum/#!forum/kubeedge� ➔ Github Repository：https://github.com/kubeedge/kubeedge� ➔

(Dashboard) Dashboard 是⼀个Kubernetes集群通⽤、基于Web的UI。它允许⽤户管理/排错集群中应⽤程序以及集群本身。 Container Resource Monitoring（容器资源监控） Container Resource Monitoring 将容器的通⽤时序指标记录到⼀个中⼼化的数据库中，并提供⼀个UI以便于浏览该数 据。 Cluster-level Logging（集群级别的⽇志） 。 Stable等级： 版本名称为 vX ，其中 X 为整数。 稳定版本的功能将会出现在许多后续版本中。 API组 为了使Kubernetes API扩展更容易，我们实现了 API groups 。 API组可在序列化对象的 apiVersion 字段中使⽤⼀个 REST路径指定。 ⽬前可使⽤的⼏个API组： 1. “core”（由于没有明确的组名称，通常称为“legacy”）组，它的REST路径是 Namespace为Name（名称）提供了范围。在Namespace中，资源的名称必须唯⼀，但不能跨Namespace。 Namespace是⼀种在多种⽤途之间划分集群资源的⽅法（通过 resource quota ）。 在未来的Kubernetes版本中，同⼀Namespace中的对象默认有相同的访问控制策略。 没有必要使⽤多个Namespace来分隔稍微不同的资源，例如同⼀软件的不同版本，可使⽤

ongoing basis The VMware SIG How to Contribute Working Groups, GitHub, and more 5 The VMware SIG Charter Link to join group: https://groups.google.com/forum/#!forum/kubernetes-sig-vmware (This will inside Kubernetes clusters. Status within the Kubernetes project 10 How to contribute Working Groups in-tree vSphere cloud provider https://github.com/kubernetes/kubernetes/tree/master/pkg/cloudpr

Resource orchestration optimization of kubernetes cluster in large scale Patrickxie ( 谢谆志） Background Cloud has been the general trend. How to manage so many clusters ,resources and businesses How to ensure load balancing of cluster nodes 1 2 Improper resource requests 3 Multi-tenant resource preemption How to expand horizontally more quickly and flexibly 4 Region1 How do you manage K8S scheduling is based on the resource request of Pod. However, in many cases, some nodes have low resource requests but high load, while some nodes have high resource requests but low load. Dynamic-Scheduler

Efficient placement of service container and tasks When placed together, don’t affect each other Resource contention ���� ������ ���������� - Online workload low 1:00am – 6:00am - Offline jobs scale VTRON RPC Resource management VTRON: Virtual Total Resources Of Node cgroup �������� ������� Kubernetes YARN Online service usage Offline job resource usage Online service resource quota Offline Offline job resource quota �������� ������� Kubernetes YARN Online service usage Offline job resource usage Online service resource quota Offline job resource quota buffer Over- subscription ��������