인프런 영문 브랜드 로고
인프런 영문 브랜드 로고
BEST
Programming

/

Devops & Infra

Practical Docker: Creating your own deep learning cloud computer with Docker

You can create a deep learning analysis environment in the cloud using Docker. When Docker images managed by Google, MS, etc. are combined with the cloud, you can work with the latest deep learning analysis methods on your computer.

(4.6) 11 reviews

213 students

Docker
Virtualization
Python
Deep Learning(DL)
mlops
Thumbnail

This course is prepared for Basic Learners.

What you will learn!

  • Build the same data analysis environment as Kaggle using Docker

  • Various ways to connect your local to a cloud with powerful computing power

  • How to Minimize Costs When Using the Cloud

  • Linux for Understanding Docker

  • Using the container feature of IDE (VSCode, RStudio, Jupyter Notebook)

Docker + Cloud + Deep Learning = DevOps Data Scientist

Lecture focusing on Docker project practice

  • Apply Docker containers to synchronize the same data analysis environment on your local environment and Azure cloud computer.

  • Light analysis is performed on the local computer.

  • For resource-intensive analyses, perform them on cloud computers.


Visible speed difference (left: cloud vs. right: local environment)

  • 10% Spot Discount on regular price

  • Use NVIDIA GPU cloud computing at low cost

  • When using a cloud computer as a server for your team, you can completely isolate each team member's work environment with Docker containers.


Course Features

  • Use the Dynamic Link feature to quickly find relevant materials even after class.

  • Maintain the basic framework of the teaching materials

  • Link updates the latest materials and allows you to download class materials periodically after class to quickly access the latest information.

📖 Main Content List

Learning Docker Commands

Learn how to understand and apply Docker commands using the Docker help function (Section 4) .

  • docker run --help

  • Using chatgpt

Create a virtual machine in Azure

Create a Linux virtual machine in the Azure Cloud and connect the local environment to the cloud virtual machine in various ways. (Section 6)

  • key-based ssh (secure shell)

  • rdp (remote desktop protocol) in gui mode

  • Connecting to VS Code Environment via Remote Tunnel Extension

Docker for Python

Learn how to create a python analysis environment using the dev container VS Code extension (Section 7) .

Building and deploying Docker images



Docker for Python using dev containers and remote tunnel extentions allows you to use Docker without having to know the basic Docker syntax. This is a powerful advantage, but to help you understand the Docker syntax more deeply, we will walk through the process of installing the latest RStudio in addition to the Kaggle R Docker Image, covering the following (Section 8) :

  • COPY and ADD

  • Installing additional programs with the RUN command

  • Understanding Ports Connections

  • CMD and ENTRYPOINT

  • ARG

Setting up a file sharing system


When connecting the local environment and the cloud, which is the goal of the lab, connecting the computing environment is not enough. It may not feel inconvenient when learning, but in order to apply the knowledge you have learned to an actual project, setting up a file sharing system is essential. We will cover the following (Section 11) :

  • Create ADLS2 ( azure-data-lake-gen2 )

  • Connecting ADLS2 to a Linux VM via blob mount

  • Connecting ADLS2 to Windows environment via blobfuse method

  • Accessing ADLS2 Directory in Windows and MAC Environments via Microsoft Azure Storage Explorer

PySpark-cover

We will cover setting up a Spark analytics environment on your personal computer using Docker Compose and ADLS2 (Section 11) .

  • Conceptual understanding of Docker Compose

  • Connecting ADLS2 directly to a Docker container

  • Configuring the PySpark Analysis Environment

True virtualization implementation


Implementing true containerization, the goal of Docker, on Linux Server virtual machines in the Azure Cloud (Section 12)

We simulate the process of creating a server administrator and multiple users, all of whom share a common Docker Image, while completely isolating each other's Docker Container environments. This is the core of mlops.

Tools and hardware requirements used in class ⚒

  • Using Visual Studio Code

  • GPU settings apply to Linux virtual machines in the cloud and connect to local environments

  • The local environment can be Windows, Linux, or MAC, so you can proceed with the project in any environment.

What will I get after taking the course?

  • Confidence in projects using Docker

  • Escape the burden of cloud and Linux

  • High-end GPU-based computers available for around $10 per month

Do you have any questions?

Q. Do I need a GPU in my local environment?

Configuring the same GPU-based data analytics environment locally and in the cloud does not mean that you need a GPU locally.

If GPU settings are not set, the analysis environment will be set based on CPU.

Q. What operating systems can I use to produce the final result?

You can practice in all cases, whether your local environment's operating system is Windows, Linux, or MAC.

Connect to a Linux cloud computer from your local environment, regardless of the operating system of your local environment.

Q. I am a Python user. There is a section related to R in the lecture. Do I also need to learn R-related content?

As you will notice when you take the class, the process of creating a python Docker container becomes much easier when using the dev container extension of VS Code. In fact, this ease may make you stray further from the goal of learning Docker syntax. For example, even if you do not apply the volume mount that is emphasized in most Docker classes, the dev container automatically performs the volume mount process.

So, I intentionally included an R section to help students learn Docker syntax. Many Docker lectures are based on app-based applications like nodejs, but I included an R section to include data-based cases so that data engineers and scientists can approach it as easily as possible.

Rather than thinking of this as a section for learning R, please think of it as a course to solidify your understanding of Docker syntax.

Q. What is an appropriate cost for cloud usage?

For models used for practice, when the Spot Discount option is applied, it costs about $10 per month (if used for about 3 hours per day).

Building high-spec virtual machines for real projects may incur additional costs.

Q. Is player knowledge required?

No prior knowledge is required for this course.

Although no prior knowledge is assumed, the difficulty of the lecture itself is not easy. However, the content is structured to be repeated and in-depth throughout the various sections of the lecture.

Recommended for
these people!

Who is this course right for?

  • Data engineers, scientists, and analysts who want to learn Docker in a practical way

  • Developers and engineers who want to learn Docker through practical experience

  • Those who need a practical portfolio for the cloud

Hello
This is danielyouk

551

Students

53

Reviews

67

Answers

4.8

Rating

7

Courses

  • LLM 기반 AI 기업에서 Pod Lead로 활동

  • 서울대학교 기계항공 공학부 졸업

  • 유럽 소재 대학원에서 기계항공공학 석사

  • 독일 소재 공학 연구기관에서 박사 과정 연구 수행

  • 유럽 대형 에너지 기업에서 Senior Data Scientist 경험

  • 영국 소재 에너지 컨설팅 기업에서 Senior Consultant 활동

  • Databricks 기반 데이터 엔지니어링 프로젝트 수행

  • Kaggle 주식 거래 AI 대회 Top 3% 성과

  • AI Agent 개발팀장으로 현재 활동 중

More

Curriculum

All

66 lectures ∙ (10hr 27min)

Course Materials:

Lecture resources
Published: 
Last updated: 

Reviews

Not enough reviews.
Become the author of a review that helps everyone!