Главная
Study mode:
on
1
Introduction
2
Problem description
3
What is a job
4
Batch workloads
5
Flexible on multiple dimensions
6
The problem
7
Microbatch jobs
8
What is Job Queueing
9
What we want to solve
10
Why propose a new controller
11
The concept of a queue
12
Design principle
13
Job API
14
Resource Model
15
Multiple namespaces
16
Multiple cluster queues
17
Borrowing cohorts
18
Queue API
19
Cluster Queue API
20
Dakota
21
Borrowing Cohort
22
How it works
23
suspend
24
queuing controller
25
existing controller
26
handling custom workloads
27
workload API
28
workload controller
29
autoscaling
30
why
31
integration
32
capacity requests
33
pod management
34
release
35
questions
36
how does the queue interact with index jobs
Description:
Explore Kueue, a Kubernetes-native job queueing system, in this informative conference talk. Dive into the challenges of managing batch workloads in Kubernetes and learn how Kueue addresses limitations in pod-centric scheduling. Discover the concept of job queueing and its benefits for resource management, especially for GPU-intensive tasks. Understand Kueue's design principles, API structure, and integration with existing Kubernetes components. Gain insights into resource allocation across multiple namespaces and clusters, borrowing cohorts, and autoscaling capabilities. Learn about the workload API, custom workload handling, and capacity management. Explore how Kueue interacts with index jobs and enhances overall cluster efficiency for batch processing tasks.

Kueue: Kubernetes-Native Job Queueing for Batch Workloads

CNCF [Cloud Native Computing Foundation]
Add to list
0:00 / 0:00