Kubernetes AI 推理服务最佳实践

3. TensorFlow Serving 部署

3.1 准备模型

# 下载示例模型
mkdir -p models/mnist/1
wget -O models/mnist/1/saved_model.pb https://storage.googleapis.com/download.tensorflow.org/models/official/20181001_resnet/savedmodels/resnet_v2_fp32_savedmodel_NHWC_jpg.tar.gz
# 创建模型存储
kubectl create -f - <<EOF
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: model-pvc
  namespace: default
spec:
  accessModes:
  - ReadWriteOnce
  resources:
    requests:
      storage: 10Gi
EOF

3.2 部署 TensorFlow Serving

deployment.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: tf-serving
  namespace: default
spec:
  replicas: 2
  selector:
    matchLabels:
      app: tf-serving
  template:
    metadata:
      labels:
        app: tf-serving
    spec:
      containers:
      - name: tf-serving
        image: tensorflow/serving:latest
        ports:
        - containerPort: 8500
        - containerPort: 8501
        env:
        - name: MODEL_NAME
          value: mnist
        volumeMounts:
        - name: model-volume
          mountPath: /models
      volumes:
      - name: model-volume
        persistentVolumeClaim:
          claimName: model-pvc

service.yaml

apiVersion: v1
kind: Service
metadata:
  name: tf-serving
  namespace: default
spec:
  selector:
    app: tf-serving
  ports:
  - port: 8501
    targetPort: 8501
  type: LoadBalancer

# 部署服务
kubectl apply -f deployment.yaml
kubectl apply -f service.yaml
# 测试推理服务
MODEL_SERVICE=$(kubectl get svc tf-serving -o jsonpath='{.status.loadBalancer.ingress[0].ip}')
curl -d '{"instances": [[[0.0 for _ in range(28)] for _ in range(28)]]}' -X POST http://$MODEL_SERVICE:8501/v1/models/mnist:predict

4. Triton Inference Server 部署

4.1 安装 Triton Inference Server

deployment.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: triton-server
  namespace: default
spec:
  replicas: 2
  selector:
    matchLabels:
      app: triton-server
  template:
    metadata:
      labels:
        app: triton-server
    spec:
      containers:
      - name: triton-server
        image: nvcr.io/nvidia/tritonserver:23.08-py3
        ports:
        - containerPort: 8000
        - containerPort: 8001
        - containerPort: 8002
        resources:
          limits:
            nvidia.com/gpu: 1
          requests:
            nvidia.com/gpu: 1
        volumeMounts:
        - name: model-volume
          mountPath: /models
      volumes:
      - name: model-volume
        persistentVolumeClaim:
          claimName: model-pvc

service.yaml

apiVersion: v1
kind: Service
metadata:
  name: triton-server
  namespace: default
spec:
  selector:
    app: triton-server
  ports:
  - port: 8000
    targetPort: 8000
  - port: 8001
    targetPort: 8001
  - port: 8002
    targetPort: 8002
  type: LoadBalancer

# 部署服务
kubectl apply -f deployment.yaml
kubectl apply -f service.yaml
# 检查服务状态
kubectl get pods -l app=triton-server

5. 性能优化

5.1 模型优化

1. 模型量化：将模型从 FP32 量化为 INT8 或 FP16
2. 模型剪枝：移除冗余的神经元和连接
3. 模型蒸馏：使用大模型训练小模型

5.2 推理服务优化

配置批处理

apiVersion: apps/v1
kind: Deployment
metadata:
  name: tf-serving-batched
  namespace: default
spec:
  replicas: 2
  selector:
    matchLabels:
      app: tf-serving-batched
  template:
    metadata:
      labels:
        app: tf-serving-batched
    spec:
      containers:
      - name: tf-serving
        image: tensorflow/serving:latest
        ports:
        - containerPort: 8501
        env:
        - name: MODEL_NAME
          value: mnist
        - name: TF_FORCE_GPU_ALLOW_GROWTH
          value: "true"
        - name: BATCH_SIZE
          value: "32"
        resources:
          limits:
            nvidia.com/gpu: 1
          requests:
            nvidia.com/gpu: 1

5.3 自动缩放

HPA 配置

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: tf-serving-hpa
  namespace: default
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: tf-serving
  minReplicas: 2
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: cpu
    target:
      type: Utilization
      averageUtilization: 70
  - type: Resource
    resource:
      name: memory
    target:
      type: Utilization
      averageUtilization: 80

6. 监控与可观测性

6.1 监控配置

Prometheus 配置

apiVersion: monitoring.coreos.com/v1
kind: ServiceMonitor
metadata:
  name: tf-serving-monitor
  namespace: monitoring
spec:
  selector:
    matchLabels:
      app: tf-serving
  endpoints:
  - port: 8501
    path: /v1/monitoring/prometheus
    interval: 15s

6.2 日志管理

日志配置

apiVersion: apps/v1
kind: Deployment
metadata:
  name: tf-serving
  namespace: default
spec:
  # ...
  template:
    spec:
      containers:
      - name: tf-serving
        image: tensorflow/serving:latest
        # ...
        env:
        - name: TF_CPP_MIN_LOG_LEVEL
          value: "0"
        - name: TF_ENABLE_GPU_GARBAGE_COLLECTION
          value: "true"
        args:
        - --model_name=mnist
        - --model_base_path=/models/mnist
        - --enable_batching=true
        - --batching_parameters_file=/models/batching_parameters.txt

7. 安全最佳实践

7.1 模型安全

1. 模型加密：使用加密技术保护模型文件
2. 访问控制：使用 RBAC 限制模型访问
3. 模型版本管理：追踪模型版本和变更

7.2 网络安全

网络策略

apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
  name: ai-inference-network-policy
  namespace: default
spec:
  podSelector:
    matchLabels:
      app: tf-serving
  policyTypes:
  - Ingress
  - Egress
  ingress:
  - from:
    - podSelector:
        matchLabels:
          app: api-gateway
    ports:
    - protocol: TCP
      port: 8501
  egress:
  - to:
    - podSelector:
        matchLabels:
          app: monitoring
    ports:
    - protocol: TCP
      port: 9090

8. 实际应用场景

8.1 多模型部署

多模型配置

apiVersion: apps/v1
kind: Deployment
metadata:
  name: triton-multi-model
  namespace: default
spec:
  replicas: 2
  selector:
    matchLabels:
      app: triton-multi-model
  template:
    metadata:
      labels:
        app: triton-multi-model
    spec:
      containers:
      - name: triton-server
        image: nvcr.io/nvidia/tritonserver:23.08-py3
        ports:
        - containerPort: 8000
        - containerPort: 8001
        - containerPort: 8002
        resources:
          limits:
            nvidia.com/gpu: 1
          requests:
            nvidia.com/gpu: 1
        volumeMounts:
        - name: model-volume
          mountPath: /models
      volumes:
      - name: model-volume
        persistentVolumeClaim:
          claimName: models-pvc

8.2 A/B 测试

A/B 测试配置

apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: ai-inference-ingress
  namespace: default
  annotations:
    nginx.ingress.kubernetes.io/canary: "true"
    nginx.ingress.kubernetes.io/canary-weight: "20"
spec:
  rules:
  - host: inference.example.com
    http:
      paths:
      - path: /v1/models
        pathType: Prefix
        backend:
          service:
            name: tf-serving-v2
            port:
              number: 8501

9. 故障排查

9.1 常见问题解决

# 查看 GPU 使用情况
kubectl exec -it <pod-name> -- nvidia-smi
# 查看推理服务日志
kubectl logs -l app=tf-serving
# 检查模型状态
curl http://<service-ip>:8501/v1/models/mnist
# 测试推理服务
curl -d '{"instances": [[[0.0 for _ in range(28)] for _ in range(28)]]}' -X POST http://<service-ip>:8501/v1/models/mnist:predict

9.2 调试技巧

1. 启用详细日志：设置 TF_CPP_MIN_LOG_LEVEL=0
2. 使用 GPU 分析工具：nvidia-smi、nvprof
3. 检查网络连接：确保服务可以正常访问
4. 验证模型格式：确保模型格式正确

10. 总结

Kubernetes 为 AI 推理服务提供了强大的部署和管理能力。通过合理配置 GPU 资源、优化模型和服务参数，可以构建高性能、可靠的 AI 推理服务。

关键要点：

• 正确配置 GPU 资源管理
• 选择适合的推理框架
• 优化模型和服务性能
• 实施安全最佳实践
• 建立完善的监控和可观测性

通过以上最佳实践，可以充分发挥 Kubernetes 的优势，构建更加高效、可靠的 AI 推理服务。