检测到您已登录华为云国际站账号，为了您更好的体验，建议您访问国际站服务网站 https://www.huaweicloud.com/intl/zh-cn
不再显示此消息
检测到您已登录华为云国际站账号，为了您更好的体验，建议您访问国际站服务网站 https://www.huaweicloud.com/intl/zh-cn
不再显示此消息
Why Is the Mount Point of a Docker Container in the Kunpeng Cluster Uninstalled? What Can I Do If a Layer Is Missing During Image Pull? Why the File Permission and User in the Container Are Question Marks? Parent Topic: Workload
NUMA Affinity Scheduling In non-uniform memory access (NUMA) architecture, a NUMA node is a fundamental component that includes a processor and local memory. These nodes are physically separate but interconnected through a high-speed bus to form a complete system.
Volcano Scheduling Volcano Scheduling Overview Scheduling Workloads Resource Usage-based Scheduling Priority-based Scheduling AI Performance-based Scheduling Queue Scheduling NUMA Affinity Scheduling Application Scaling Priority Policies Parent Topic: Scheduling
NUMA affinity scheduling (numa-aware) weight After this function is enabled, the parameter defaults to 1. NUMA Affinity Scheduling Load-aware scheduling (usage) weight After this function is enabled, the parameter defaults to 5.
AI computing is 3 to 5 times better with NUMA BMSs and high-speed InfiniBand network cards. Highly Available and Secure HA: CCE supports three control plane nodes on the cluster management plane. These nodes run in different regions to ensure cluster HA.
Node or node pool management Heterogeneous resource management: available across various Huawei Cloud computing instances, including VMs, BMSs, Kunpeng instances, Ascend computing, and GPU-based instances. For details, see Node Overview and Node Pool Overview.
AI Performance-based Scheduling NUMA affinity scheduling Volcano targets to lift the limitation to make scheduler NUMA topology aware so that: Pods are not scheduled to the nodes that NUMA topology does not match. Pods are scheduled to the most suitable node for NUMA topology.
Table 27 Kunpeng ultra-high I/O ECS features ECS Type Compute Network Supported Cluster Type kI1 vCPU to memory ratio: 1:4 Number of vCPUs: 8 to 64 Kunpeng 920 processor Base frequency: 2.6 GHz Ultra-high PPS throughput Higher compute specifications for better network performance
CCE adapts to various Huawei Cloud compute instances like Kunpeng instances and supports GPUs and Ascend compute. It provides GPU virtualization, shared scheduling, and resource-aware scheduling optimization.
Resource isolation measures (such as CPU pinning, NUMA affinity, tidal affinity, and network bandwidth control) ensure the resource-sensitive services to meet their SLOs.
