Pets vs Cattle: Fleet Monitoring for Connected Devices

on YouTube ↗31m

" summary

A working proof of concept uses a Yocto layer that deploys a fluent-bit client on devices, a custom output plugin that sends in-band JSON to an Elasticsearch + Kibana server (docker-compose provided), and visualizations for CPU, memory and device metadata. Fluent‑bit is chosen for its tiny footprint (about 3 MB flash, 650 KB RAM) compared with alternatives; a stress test shows live CPU spikes in Kibana. Source and build instructions are published on GitHub.

AI-written summary · about

" chapters (7)

1. 0:00Start
2. 2:12Pets vs Cattle
3. 4:47Fleet Monitoring - Intro
4. 6:18Fleet Monitoring - Details
5. 8:04Fleet Monitoring General
6. 10:33Discarded Options
7. 13:40Considered Options

" related talks

• 22mLogging: Fluentd & Fluent BitEduardo Silva, Masahiro Nakagawa · KubeCon + CloudNativeCon North America 2020 Virtual
• 34mGetting Started With Logging in KubernetesEduardo Silva · KubeCon + CloudNativeCon China 2018
• 34mFluentd and Fluent BitEduardo Silva · KubeCon + CloudNativeCon North America 2021
• 34mPower of O11y with FLUENT BIT in AWS EKS environmentDharmendra Ahuja · Conf42 Observability 2025
• 36mDeep Dive: Fluent BitEduardo Silva, Masahiro Nakagawa · KubeCon + CloudNativeCon China 2018
• 28mBuilding Observability Pipeline with Fluent-BitChao Xu · Open Observability Day North America 2022
• 30mFluent Bit, the engine to power Chat OpsPatrick Stephens · Open Source Monitoring Conference 2024
• 13mFluent BitEduardo Silva · KubeCon + CloudNativeCon Europe 2021