Summary of best practices for ACK managed clusters and ACK dedicated clusters - Container Service for Kubernetes

This topic summarizes best practices for using ACK managed clusters and ACK dedicated clusters in a variety of scenarios.

Category	References
LLM inference	Deploy a DeepSeek distilled model inference service on ACK Get started with large language models in ACK Use vLLM to deploy a Qwen model inference service Use rtp-llm to deploy a Qwen model inference service Use Triton to deploy a Qwen model inference service Use TGI to deploy a Qwen model inference service
AIGC	Accelerate AI image building and pulling Deploy a production-grade Stable Diffusion service based on Knative Use CPUs to accelerate text-to-image inference for Stable Diffusion XL Turbo
Authorization	Authorization best practices
Clusters	Recommended configurations for ECS instance types Recommended configurations for a high-availability cluster architecture Network planning for ACK managed clusters Recommendations for using large-scale clusters
Nodes and node pools	Best practices for node pool deployment sets Best practices for node pools of spot instances Migrate standalone nodes to a node pool Comparison of containerd, sandboxed containers, and Docker runtimes
Networking	Optimize the extensibility of NetworkPolicy in large-scale Terway clusters Use an ALB to expose Services through the Gateway API DNS best practices Recommendations for using Nginx Ingress Controller
Application Management	Recommended configurations for workloads Accelerate AI image building and pulling Deploy Jenkins in an ACK cluster to build and deploy applications Use GitLab CI to run GitLab Runner and execute a pipeline
Knative	Enable monitoring and alerts for Knative Services Implement log collection on Knative Use ECI resources Use HPA in Knative Implement automatic service scaling based on the number of traffic requests View the Knative service monitoring dashboard Deploy a production-grade Stable Diffusion service based on Knative
Storage	Use disks as ephemeral storage volumes Recommended high-availability configurations for disk volumes Best practices for block storage throttling Best practices for disk storage data security Migrate stateful applications that use disk volumes across zones Use multi-attach and reservations for NVMe disks Implement persistent storage for StatefulSets Best practices for read/write splitting for OSS storage Automatically collect JVM dump files from abnormal exits using CNFS
Observability	Use KubeSkoop to locate network issues Use Prometheus Client to monitor applications Best practices for configuring alert rules in Prometheus Use SysOM to locate container memory issues
Cost management suite	Use Cost Insight for cost management and optimization
Auto Scaling	Implement second-level container scaling based on ack-autoscaling-placeholder Optimize elasticity with custom images Implement rapid and simultaneous elastic scale-out across multiple zones Implement workload scaling based on UnitedDeployment Implement HPA-based elastic scaling through ARMS Application Monitoring Implement elastic scaling based on GPU metrics
Service Mesh	Manage ECI pod applications on ACK virtual nodes using ASM Manage applications in registered clusters using ASM Use an ASM serverless gateway to implement a multi-cluster ingress Disaster recovery scenarios for multiple ACK clusters in different VPCs (VPC networks connected through CEN) Use ASM metrics to implement automatic elastic scaling for workloads Access gRPC services within a mesh through an ASM ingress gateway
Graphical applications	Best practices for Linux graphical applications Best practices for Unity remote rendering