AI应用的容器化部署：从Docker到Kubernetes
前言
我们的产品从单服务器部署到多服务器部署时，遇到了很多问题：环境不一致、部署复杂、扩容困难。
后来我们引入了容器化和 Kubernetes，问题迎刃而解。今天，分享我们的实践经验。
一、容器化基础
1.1 Docker 入门
class DockerSetup: def create_dockerfile(self) -> str: """创建 Dockerfile""" return """ FROM python:3.10-slim WORKDIR /app COPY requirements.txt . RUN pip install --no-cache-dir -r requirements.txt COPY . . EXPOSE 8000 CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8000"] """1.2 Docker Compose
class DockerCompose: def create_config(self) -> dict: """创建 Docker Compose 配置""" return { "version": "3.8", "services": { "web": { "build": ".", "ports": ["8000:8000"], "depends_on": ["redis"] }, "redis": { "image": "redis:7-alpine" } } }二、Kubernetes 入门
2.1 Pod 配置
class KubernetesPod: def create_pod(self) -> dict: """创建 Pod 配置""" return { "apiVersion": "v1", "kind": "Pod", "metadata": {"name": "ai-app"}, "spec": { "containers": [{ "name": "app", "image": "ai-app:latest", "ports": [{"containerPort": 8000}] }] } }2.2 Deployment 配置
class KubernetesDeployment: def create_deployment(self) -> dict: """创建 Deployment""" return { "apiVersion": "apps/v1", "kind": "Deployment", "metadata": {"name": "ai-app"}, "spec": { "replicas": 3, "selector": {"matchLabels": {"app": "ai-app"}}, "template": { "metadata": {"labels": {"app": "ai-app"}}, "spec": { "containers": [{ "name": "app", "image": "ai-app:latest", "ports": [{"containerPort": 8000}] }] } } } }三、服务发现与负载均衡
3.1 Service 配置
class KubernetesService: def create_service(self) -> dict: """创建 Service""" return { "apiVersion": "v1", "kind": "Service", "metadata": {"name": "ai-app-service"}, "spec": { "selector": {"app": "ai-app"}, "ports": [{"port": 80, "targetPort": 8000}], "type": "LoadBalancer" } }3.2 Ingress 配置
class KubernetesIngress: def create_ingress(self) -> dict: """创建 Ingress""" return { "apiVersion": "networking.k8s.io/v1", "kind": "Ingress", "metadata": {"name": "ai-app-ingress"}, "spec": { "rules": [{ "host": "api.example.com", "http": { "paths": [{ "path": "/", "pathType": "Prefix", "backend": { "service": { "name": "ai-app-service", "port": {"number": 80} } } }] } }] } }四、CI/CD 集成
4.1 持续集成
class CICDWorkflow: def create_workflow(self) -> dict: """创建 CI/CD 工作流""" return { "name": "CI/CD", "on": {"push": {"branches": ["main"]}}, "jobs": { "build": {"steps": ["checkout", "build", "test"]}, "deploy": {"steps": ["deploy"]} } }4.2 自动部署
class AutoDeployment: def deploy(self, image_tag: str) -> dict: """自动部署""" return { "image": f"ai-app:{image_tag}", "strategy": "rollingUpdate", "success": True }五、最佳实践
5.1 容器化原则
- ✅单一职责：每个容器只运行一个进程
- ✅不可变容器：容器创建后不修改
- ✅健康检查：配置 liveness 和 readiness 探针
- ✅资源限制：设置 CPU 和内存限制
5.2 Kubernetes 最佳实践
- ✅副本管理：使用 Deployment 管理 Pod
- ✅服务发现：使用 Service 进行服务发现
- ✅配置管理：使用 ConfigMap 和 Secret
- ✅自动扩缩容：使用 Horizontal Pod Autoscaler
六、总结
容器化和 Kubernetes 是现代化部署的基石。关键在于：
- 标准化环境：确保开发、测试、生产环境一致
- 自动化部署：减少人工操作
- 弹性伸缩：根据负载自动调整
- 持续改进：不断优化部署流程
记住：容器化不是目的，是手段。
