Autoscale your Kubernetes Workload with Prometheus - Frederic Branczyk, CoreOS

on YouTube ↗34m

" summary

Explains how Kubernetes moves from Heapster to the resource, custom and external metrics APIs (available since Kubernetes 1.8, external API alpha), with a Prometheus custom metrics adapter that translates HPA queries into PromQL. Shows HPA using per-pod rates, scrape intervals (15s) and targets, with cool-downs and a live demo that scales a blog to about 2000 requests per second up to a 30-replica max. Notes Heapster deprecation, CRD scaling in 1.11, and ongoing OpenMetrics/observability work.

AI-written summary · about

" chapters (11)

1. 0:00Start
2. 0:27What's to learn today?
3. 1:08Let's talk about autoscaling
4. 3:34Horizontal Autoscaling
5. 4:12Vertical autoscaling
6. 8:18Problem with Heapster over time
7. 11:50Resource & Custom Metrics APIs
8. 13:02Resource Metrics API
9. 16:44External Metrics API
10. 17:37Quick Prometheus intro
11. 21:00Monitoring pipeline overview

" related talks

• 33mDeep Dive: Kubernetes Metric APIs using PrometheusMatthias Loibl, Sergiusz Urbaniak · KubeCon + CloudNativeCon Europe 2019
• 25mAutoscaling Kubernetes Workloads Based On MetricsRakshit Gondwal · KCD Mumbai 2023
• 51mHorizontal Autoscaling with KubernetesRob Richardson · KCD Los Angeles 2023
• 34mScale your microservices by MicroProfile Metrics on KubernetesKenji Kazumura · EclipseCon 2022
• 29mKubernetes Autoscaling: Beyond One Size Fits AllJohn Platt · KubeDay Colombia 2024
• 08mGetting Creative with HPA Scaling in KubernetesDrew Rogers · Philly ETE 2023
• 19mKeynote: Monitoring Kubernetes Clusters with PrometheusFabian Reinartz · KubeCon + CloudNativeCon North America 2016
• 37mMonitoring Kubernetes with PrometheusTom Wilkie · DevFest Ukraine 2018