Design & Deployment of FaaS apps at the edge-cloud
summary
PHYSICS (H2020) provides a Node-RED based design environment that generates Docker images and deployable OpenWhisk actions via Jenkins, with a multi-cluster backend using Kubernetes/OpenShift, Submariner and Open Cluster Management. A custom Node-RED palette adds semantics like split-join parallelization, affinity/locality and sizing annotations; a function generator builds artifacts and an OpenWhisk pipeline runs them. Measurements show ~100 ms extra delay for container deployments versus local Node-RED runs and a misconfigured OpenWhisk memory (2 GB vs 32 GB) explains long Azure wait times. Schedulers target performance vs energy and cold-start minimization.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 1:04FAST project overview
- 3:56FAST challenges
- 6:05FAST partners timeline
- 6:46FAST architecture
- 8:10FAST pilot apps
- 9:56Design environment
- 13:54Artifact lifecycle
- 15:19UI and function creation
- 17:25Complex workflow patterns
- 19:15Parallelization in FAST
- 21:27Semantic nodes rules
- 22:32Kubernetes setup
- 24:17Function registration
- 26:27Performance measurements
- 28:35Network latency
- 30:48Scheduling optimizations
- 32:12Reusable artifacts
related talks
31mFaaS Meets Node.js Frameworks: Developing Cloud Native Node.js Applications at Speed
15mServerless Apps with Node.JS in OpenWhisk and Adobe I/O Runtime
27mCloud Native, Event Drive, Serverless, Microservices Framework - OpenWhisk
39mFaaS: A Contemporary, Comparative Study
31mBirds of a Feather: Do You Really Need on Premises Serverless ?
36mServerless and Event-driven Applications with Cloud Foundry and OpenWhisk
34mApache OpenWhisk: Building a Production-Ready Serverless Stack on/for Kubernetes
55mApache OpenWhisk: How it Works and How You Can Benefit