Telemetry Showdown: Fluent Bit Vs. OpenTelemetry Collector - a Comprehensive...
summary
Benchmark compares Fluent Bit 2.x against the OpenTelemetry Collector using Hipster Shop and OpenTelemetry demo apps, tracing, logs and Prometheus scraping. Fluent Bit handles logs with low cost (Fluent Bit ≈15 MB), Collector offers full trace/metric transforms and relabeling but consumes more memory and CPU (initial ~95 MB, two‑hour tests up to 1.1–1.5 GB) and shows a 24‑hour soak jumping to ~9.5 GB and crashes (Collector leak to 9.5GB). Adding a target allocator restores stability (~800 MB).
AI-written summary · about
chapters (16)
- 0:00Introduction
- 1:06Talk objectives
- 2:30Fluent Bit vs OpenTelemetry
- 3:56Design and pipeline
- 9:10Logging features
- 10:36Metrics support
- 12:22Tracing capabilities
- 13:30Benchmark setup
- 15:57Performance logs and CPU
- 16:38Performance traces and memory
- 17:00Metrics under load
- 17:42Soak test results
- 19:27Target allocator
- 20:12Processing location impact
- 20:54Recommendations
- 21:38Q&A
related talks
24mTelemetry Showdown: Fluent Bit Vs. OpenTelemetry Collector - A Comprehensive Benchma...
57mTelemetry Showdown: Fluent Bit vs. OpenTelemetry Collector - A Comprehensive Benchmark Analysis
19mTelemetry Showdown: Fluent Bit vs. OpenTelemetry Collector - A Comprehensive Benchma...
07mFluent Bit vs. OpenTelemetry Collector - Why Not Both?
15mJourney from Fluent Bit, Fluentd and Prometheus to Open...
06mFluent Bit V5
27mLog Support in OpenTelemetry
36mFluent Bit: Logs, OpenMetrics, and OpenTelemetry all-in-one