Explore the challenges and solutions of running Apache Spark on Kubernetes in this conference talk. Discover how to build a large-scale Spark Service on Kubernetes, focusing on autoscaling in multi-tenant environments. Learn about advanced features like physical isolation, capacity settings, bin-packing, and scale controls. Gain insights into significant CPU and memory utilization improvements for Spark on Kubernetes. Understand the elastic architecture, node group layouts, cluster autoscaling, and production status of this solution. Get key takeaways and future directions for optimizing Spark workloads in cloud-native environments.
Spark on Kubernetes: Building an Elastic Service - Apple's Approach
CNCF [Cloud Native Computing Foundation] via YouTube
Overview
Syllabus
Introduction
Bowen Li Apple
Outline
Why Cloud
Design Principles
Elastic Spark Architecture
Open Source
Node Groups Layout
Cluster Autoscaling
Production Status
Key takeaways
Whats next
Taught by
CNCF [Cloud Native Computing Foundation]
Related Courses
-
Beyond Experimental - Spark on Kubernetes
-
Getting Started with Apache Spark on Kubernetes
-
Apache Spark on Kubernetes - Lessons Learned from Launching Millions of Spark Executors
-
Downscaling Apache Spark Clusters - Challenges and Solutions
-
Capacity Scheduling for Elastic Resource Sharing in Kubernetes
-
Managing Multi-Cloud Apache Spark on Kubernetes
