Smarter sampling: We need your insights!
We’re exploring new ways to sample requests more intelligently, and we’d love your input.
Today, Blackfire’s request sampling is straightforward. If you define a sample rate of 50%, we instrument every other request. It’s clean, predictable, and purely mathematical.
Is it the best we can do?
This strategy lacks nuance. All requests are treated equally, regardless of their actual value. We might be missing out on high-value signals while oversampling low-impact noise.
That means lost opportunities to:
- Catch rare but critical slowdowns,
- Focus on key user journeys,
- Optimize where it matters most.
And ultimately, it means leaving insights on the table.
We believe we can do better.
We’re exploring possible smarter sampling strategies. The goal? Maximize signal, minimize cost.
But we can’t build this alone. We want to learn from you:
- What data would you always want to keep?
- What patterns do you see in your own traffic?
- How do you balance observability and cost?
Join the conversation!
Share your thoughts, ideas, or even frustrations:
- Send me a message: thomas.diluccio@platform.sh
- Book a quick chat
Let’s co-create a smarter way to sample data collection. Together.
To better observability and beyond