No More Chaos: Audit and Inspect Kubernetes at Scale
summary
Describes how Alibaba and Ant Financial monitor and troubleshoot Kubernetes at web scale by designing stability-focused metrics, consuming those metrics via streaming alarms, and aggregating/analyzing metric, event and resource data to detect faults. Covers concrete practices and a real-case demo of metric aggregation and analysis, plus an automated real-time inspection and analysis system used in Alibaba for Kubernetes clusters.
AI-written summary · about
chapters (14)
- 0:00Start
- 0:37We have A Large Scale Kubernetes Cluster
- 4:46Considerations on An Inspection Service
- 6:06What We Build in Version 1
- 10:43Knative: The Killer Component
- 11:57We Build V2 in Serwerless with knative
- 15:42Some Problems in Business Stability
- 18:01Changes When We Moving to Cloud Native and Kubernetes?
- 20:54Boundary of Inspection System
- 26:38Self-healing Inspection System - Overview
- 28:20User-driven System Inspecting
- 30:18Core Engine for Self-Healing Inspection System
- 32:38Examples & Results
- 34:35Future: Alps-driven Self-Healing inspection System
related talks
35mNo More Chaos: Audit and Inspect Kubernetes at Scale
29mHow We Discover and Locate k8s Cluster Problems Before Users at Alibaba
40mThe Good, the Bad, and the Ugly of Managing Production Systems with Kubernetes
11mKubernetes Observability
41mImprove observability for your container workloads
42mDistributed Monitoring: How to understand the Chaos
32mImplementing Chaos Engineering in Kubernetes
56mBest practices for optimizing Kubernetes applications on AWS