ACK Lingjun clusters support the same management capabilities as ACK Pro clusters. Use the Container Service for Kubernetes (ACK) console to manage authorization, networking, applications, security, observability, and scheduling across your clusters.
| Area | Topics |
|---|---|
| Authorization management | Authorization management |
| Network management | Service management<br>Ingress management<br>DNS Service discovery |
| Component management | Manage components |
| Application management | Workloads<br>Application scheduling<br>Configuration management |
| Security management | Security |
| Observability | Log management<br>Monitoring management |
| Scheduling | Task scheduling<br>Task scheduling overview<br>Work with gang scheduling<br>Use Capacity Scheduling<br><br>GPU scheduling<br>Use eGPU to share and schedule GPU resources<br><br>Topology-aware GPU scheduling<br>Because Kubernetes is unaware of GPU resource topology, it schedules GPU resources randomly, causing variable training job acceleration. ACK supports topology-aware GPU scheduling, which selects multiple GPUs from GPU-accelerated nodes to achieve optimal GPU acceleration for training jobs.<br>Overview<br>Work with topology-aware GPU scheduling (TensorFlow edition)<br>Work with topology-aware GPU scheduling (PyTorch edition) |
该文章对您有帮助吗？