┌─────────────────────────────────┐  ┌─────────────────────────────────┐
│       Virtual Machine           │  │         Container               │
├─────────────────────────────────┤  ├─────────────────────────────────┤
│  App A  │  App B  │  App C      │  │  App A  │  App B  │  App C      │
│  Bins   │  Bins   │  Bins       │  │  Bins   │  Bins   │  Bins       │
│  Guest  │  Guest  │  Guest      │  ├─────────────────────────────────┤
│   OS    │   OS    │   OS        │  │      Container Runtime          │
├─────────────────────────────────┤  │        (Docker Engine)          │
│       Hypervisor (Type 1/2)     │  ├─────────────────────────────────┤
├─────────────────────────────────┤  │         Host OS Kernel          │
│         Host OS / Hardware      │  ├─────────────────────────────────┤
└─────────────────────────────────┘  │           Hardware              │
                                     └─────────────────────────────────┘

┌──────────┐     REST API     ┌──────────────────────┐
│ Docker   │ ───────────────→ │   Docker Daemon       │
│  CLI     │                  │   (dockerd)           │
└──────────┘                  │                       │
                              │  ┌─── containerd ───┐ │
                              │  │  ┌── runc ──┐    │ │
                              │  │  │Container │    │ │
                              │  │  └──────────┘    │ │
                              │  └──────────────────┘ │
                              └──────────────────────┘

┌─────────────────────────────┐  ← 쓰기 가능 (컨테이너 레이어)
├─────────────────────────────┤
│ Layer 4: COPY app.py /app   │  ← 읽기 전용
│ Layer 3: RUN pip install    │  ← 읽기 전용 (캐시 대상)
│ Layer 2: RUN apt-get update │  ← 읽기 전용
│ Layer 1: FROM python:3.12   │  ← 베이스 이미지
└─────────────────────────────┘

FROM python:3.12-slim AS base

# 시스템 의존성
RUN apt-get update && apt-get install -y --no-install-recommends \
    curl \
    && rm -rf /var/lib/apt/lists/*

# 비root 사용자
RUN groupadd -r appuser && useradd -r -g appuser appuser

WORKDIR /app

# 의존성 먼저 (캐시 활용)
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt

# 소스코드
COPY . .

USER appuser
EXPOSE 8000

HEALTHCHECK --interval=30s --timeout=5s --retries=3 \
    CMD curl -f http://localhost:8000/health || exit 1

CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8000"]

# ---- 빌드 스테이지 ----
FROM python:3.12 AS builder
WORKDIR /build
COPY requirements.txt .
RUN pip install --prefix=/install --no-cache-dir -r requirements.txt

# ---- 런타임 스테이지 ----
FROM python:3.12-slim AS runtime
COPY --from=builder /install /usr/local
WORKDIR /app
COPY . .
CMD ["python", "serve.py"]

# ---- 빌드 스테이지 ----
FROM nvidia/cuda:12.4.1-devel-ubuntu22.04 AS builder
RUN apt-get update && apt-get install -y python3-pip
COPY requirements.txt .
RUN pip install --prefix=/install --no-cache-dir -r requirements.txt

# ---- 런타임 스테이지 (runtime은 devel보다 ~2GB 작음) ----
FROM nvidia/cuda:12.4.1-runtime-ubuntu22.04 AS runtime
RUN apt-get update && apt-get install -y python3 && rm -rf /var/lib/apt/lists/*
COPY --from=builder /install /usr/local
WORKDIR /app
COPY . .
CMD ["python3", "train.py"]

.git
__pycache__
*.pyc
.env
data/
models/
*.tar.gz
.vscode
node_modules

# 학습 데이터 바인드 마운트 + 모델 출력용 Named Volume
docker run --gpus all \
    -v /data/datasets:/data/input:ro \
    -v model-output:/data/output \
    training-image python train.py

# docker-compose.yml — ML 서비스 스택 예시
services:
  mlflow:
    image: ghcr.io/mlflow/mlflow:v2.21.3
    ports:
      - "5000:5000"
    volumes:
      - mlflow-data:/mlflow
    environment:
      - MLFLOW_BACKEND_STORE_URI=sqlite:///mlflow/mlflow.db
      - MLFLOW_DEFAULT_ARTIFACT_ROOT=/mlflow/artifacts
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:5000/health"]
      interval: 30s
      timeout: 5s
      retries: 3

  model-server:
    build: ./serving
    ports:
      - "8000:8000"
    depends_on:
      mlflow:
        condition: service_healthy
    environment:
      - MLFLOW_TRACKING_URI=http://mlflow:5000
      - MODEL_NAME=production-model
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]

  jupyter:
    image: quay.io/jupyter/pytorch-notebook:latest
    ports:
      - "8888:8888"
    depends_on:
      mlflow:
        condition: service_healthy
    environment:
      - MLFLOW_TRACKING_URI=http://mlflow:5000
    volumes:
      - ./notebooks:/home/jovyan/work

volumes:
  mlflow-data:

networks:
  default:
    name: ml-stack

# vLLM으로 LLM 모델 서빙하는 컨테이너 실행 예시
docker run --gpus all \
    -v /data/models/llama3:/model \
    -p 8000:8000 \
    vllm/vllm-openai:latest \
    --model /model \
    --max-model-len 8192 \
    --tensor-parallel-size 1