OpenTelemetry profiles enters public alpha

📝 Discussion Summary (Click to expand)

Top 3 themes in the discussion

Skepticism about “low‑overhead” profiling
secondcoming doubts that anything from the OTel community can truly meet low‑overhead expectations: > "Continuously capturing low‑overhead performance profiles in production"
Maintainers address the performance concern
felixge (OTel Profiling SIG maintainer) acknowledges the critique and stresses ongoing optimisation:

"we’ve tried our best to make things efficient across the protocol and all involved components."
Practical enthusiasm and adoption
ollien reports real‑world success, noting that the Elixir implementation has been “exceptionally useful” in their work:

"Very excited for this. We've used the Elixir version of this at $WORK a handful of times and have found it exceptionally useful."

🚀 Project Ideas

Generating project ideas…

Continuously capture production‑grade performance profiles with sub‑1% overhead using eBPF and WebAssembly‑compiled Flux queries.
Provides a unified dashboard that visualizes hot functions, allocation sites, and resource‑intensive traces without code changes.

Key	Value
Target Audience	Production engineers, SREs, performance analysts
Core Feature	Real‑time, low‑impact profiling via eBPF agents that feed Flux‑based graphs to a cloud dashboard
Tech Stack	eBPF agents (Linux), WASM sandbox, Flux query engine, React front‑end, GraphQL API
Difficulty	High
Monetization	Revenue-ready: usage‑based tiered pricing (free up to 10 GB‑day, then $0.10/GB)

HN commenters lamented that most OTel profiling solutions are not truly low‑overhead; Flavory directly addresses this gap.
The integration of Flux into a profiling view would let engineers write concise “hot‑path” queries (e.g., “find functions with > 95th‑percentile CPU in last 5 min”) and instantly surface them on a graph.
Early adopters would love a SaaS that eliminates the need to self‑host agents or fiddle with custom OpenTelemetry pipelines.

A stand‑alone REPL for a expressive graph traversal language that lets developers explore execution graphs of distributed systems in real time.
Enables ad‑hoc diagnostics, performance debugging, and on‑the‑fly query testing without launching full applications.

Key	Value
Target Audience	Developers working with tracing pipelines, distributed systems architects
Core Feature	REPL that parses Flux‑style expressions, builds execution graphs, and streams live results to a UI canvas
Tech Stack	Rust VM, LLVM JIT, TypeScript UI, WebSockets for low‑latency updates
Difficulty	Medium
Monetization	Hobby

Community members (e.g., “Very excited for this”) have expressed interest in using such tools within Elixir pipelines, indicating strong uptake potential.
The REPL could be packaged as a desktop Electron app or a browser‑based playground, encouraging adoption among engineers who currently “haven’t heard of syslog” but crave more introspectable debugging.

An open‑source benchmarking platform that runs Jepsen‑like chaos experiments against rsyslogd (and alternatives) while simultaneously profiling CPU, I/O, and latency.
Generates actionable reports and performance graphs to help teams evaluate reliability under burst logging loads.

Key	Value
Target Audience	Ops teams, SREs, dev‑ops engineers using central syslog collectors
Core Feature	Configurable test harness that injects burst traffic, measures drop/latency, and outputs performance profiles
Tech Stack	Go test suite, k6 load generation, eBPF profilers, Prometheus + Grafana for visualization
Difficulty	Medium
Monetization	Hobby

Directly responds to “Has anyone profiled the performance and reliability characteristics of rsyslogd…” and the call for “Jepsen‑like stress test”.
Allows teams to prove or disprove that “fewer/no moving parts” solutions can actually meet production SLAs, satisfying commenters’ desire for concrete numbers.
Including a shareable dashboard would spark discussion on HN, as users could post results of their own deployments.