Apache OpenWhisk: Building a Production-Ready Serverless Stack on/for Kubernetes

on YouTube ↗34m

" summary

Apache OpenWhisk runs on Kubernetes with a Helm chart and a Minikube-friendly deploy and measures concrete tradeoffs: a small 13-node test cluster with 10 invokers shows a 700 ms vs 2+ s cold start gap between Docker and Kubernetes container factories, 7 ms warm latency for Docker reuse, and unexpected log extraction stalls that reduce throughput. Architecture uses per-invoker Kafka queues, daemon-set invoker agents to suspend/resume containers (Kubernetes lacks suspend), and composition tools (Composer, Shell) that wire Travis CI webhooks to Slack.

AI-written summary · about

" chapters (17)

1. 0:00Intro
2. 0:26Agenda
3. 1:10What is OpenWhisk
4. 2:59OpenWhisk on Kubernetes
5. 4:23Assumptions
6. 6:49Container Life Cycle
7. 10:08Function Invocation
8. 13:44Container Factory SPI
9. 16:40Experiments
10. 17:55Latency Results
11. 20:23throughput Results
12. 20:56Key takeaways
13. 21:34Whats next
14. 21:47Dream big
15. 22:14Demo
16. 24:43Travis Slack Pipeline
17. 32:49Wrapup

" related talks

• 55mApache OpenWhisk: How it Works and How You Can BenefitCarlos Santana · ApacheCon North America 2017
• 56mApache OpenWhiskAnimesh Singh, John Zaccone · Devoxx US 2017
• 01h 00mServerless Java on KubernetesBurr Sutter, Kamesh Sampath · Voxxed Days Singapore 2018
• 41mOpenWhisk as an Apache Mesos FrameworkTyson Norris · MesosCon North America 2017
• 37mBuilding Serverless Applications (...)Lucio Grenzi · Codemotion Milan 2017
• 27mCloud Native, Event Drive, Serverless, Microservices Framework - OpenWhiskDaniel Krook · CloudNativeDay 2016
• 15mServerless Apps with Node.JS in OpenWhisk and Adobe I/O RuntimeShazron Abdullah · JSConf.Asia 2018
• 31mBirds of a Feather: Do You Really Need on Premises Serverless ?Igor Khapov · KubeCon + CloudNativeCon Europe 2019