Extend Flink to edge computing with much lower footprint
summary
A lightweight Flink-compatible runtime called Quick framework compiles jobs into statically linked executables that run with runtime memory below 10 MB and no JVM dependency. It reimplements Flink SQL planning and operators (coroutines + channel messaging), ships UDF bridging for Go/Java, and supports Kafka, MQTT and local filesystem state backends with two‑phase checkpoint commit. Binaries are <50 MB raw (~4 MB gzipped). Measured throughput is ~400,000 msgs/s on Raspberry Pi and >700,000 msgs/s on a cloud server.
AI-written summary · about
chapters (9)
- 0:00Introduction and motivation
- 1:27Challenges running Flink on edge devices
- 3:13Quick framework design and optimizations
- 6:14Job translation and execution process
- 9:14Supporting user-defined functions in Go
- 11:29Compatibility and supported features
- 12:38Performance and footprint results
- 13:45Deployment and future plans
- 14:50Demo and Q&A
related talks
18mFlexConnect: Mobile Computational Offloading
55mFog Computing and IoT Solution Architecture
27mControlling Distributed Energy Resources with Edge Computing and Go
48mInternet of Programmable Things
37mBuilding an Open IoT Solution with EdgeX Foundry and Zephyr Project
49mScaling Qt from Desktop to Servers and Micro-controllers
19mEdgeX Foundry - An Open Source Platform for Edge/IoT Solutions
37mBuilding an Open IoT Solution with EdgeX Foundry and Zephyr Project