Kubernetes for ML workloads in MLOps - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is Kubernetes in the context of ML workloads?

Kubernetes is a system that helps run and manage machine learning tasks on many computers. It makes sure ML programs run smoothly and can grow or shrink as needed.

beginner

Why use Kubernetes for machine learning model training?

Kubernetes helps by automatically managing resources, running training jobs in containers, and scaling up or down based on need. This saves time and avoids manual setup.

beginner

What is a Pod in Kubernetes?

A Pod is the smallest unit in Kubernetes. It holds one or more containers that run ML code or services together on the same machine.

intermediate

How does Kubernetes help with scaling ML workloads?

Kubernetes can add or remove Pods automatically based on how busy the ML workload is. This means your ML tasks get more power when needed and save resources when not busy.

intermediate

What role do Persistent Volumes play in ML workloads on Kubernetes?

Persistent Volumes store data like training datasets or model files outside of Pods, so data stays safe even if Pods stop or restart.

What does Kubernetes use to run ML code in isolated environments?

A Physical Servers

B Virtual Machines

C Containers

D Databases

Kubernetes runs ML code inside containers, which are lightweight and isolated environments.

Which Kubernetes object is the smallest unit that runs containers?

A Node

B Pod

C Service

D Deployment

A Pod is the smallest unit in Kubernetes that can hold one or more containers.

How does Kubernetes help when ML workloads need more computing power?

A It automatically scales Pods up or down

B It manually asks the user to add servers

C It pauses the workload

D It deletes old data

Kubernetes can automatically add or remove Pods to match the workload demand.

What is the purpose of Persistent Volumes in Kubernetes for ML?

A To store data safely outside Pods

B To monitor CPU usage

C To create network connections

D To run ML code faster

Persistent Volumes keep data safe even if Pods restart or stop.

Which of these is NOT a benefit of using Kubernetes for ML workloads?

A Isolation of ML tasks in containers

B Easy scaling of workloads

C Automatic resource management

D Manual hardware setup required

Kubernetes reduces the need for manual hardware setup by automating resource management.

Explain how Kubernetes manages machine learning workloads from running code to scaling resources.

Think about how Kubernetes runs and adjusts ML tasks automatically.

Describe the role of storage in Kubernetes for ML workloads and why it is important.

Consider what happens to data when ML tasks stop or restart.