About this talk
Pyroscope 2.0 brings a major architectural rewrite that cuts production infrastructure costs by up to 74%. Christian Simon and Alberto Soto from the Pyroscope team walk through exactly what was broken in v1, how the new design fixes it, and what it's already delivered after seven months running across all Grafana Cloud regions.
The v1 architecture placed stateful ingesters in both the read and write paths, meaning heavy queries directly degraded write availability. Series-hash sharding caused symbol data (stack traces, function names, line numbers) to be duplicated across every ingester — a 50-ingester cell stored more symbol data per second than it actually received as input. Scaling those ingesters took hours, not minutes.
Pyroscope v2 removes ingesters entirely. Stateless segment writers flush profiles directly to object storage every 500ms. A lightweight metastore handles query planning. Read-path backends are fully stateless and spin up in seconds. Sharding by service name instead of series hash means symbol data is stored once per service rather than spread everywhere, yielding a 95% reduction in symbol data volume in production. Total infrastructure cost dropped up to 74% across migrated cells.
The talk also demos three v2-only features: profile exemplars (drill into a single profile's flame graph directly from the aggregate view), a span CPU heatmap showing the full distribution of CPU usage across request traces, and bidirectional profiles-to-traces linking — clicking a slow profile now opens the matching Grafana Tempo trace for full request context.
Everything shown is fully open source. Grafana Cloud has a free tier.
0:00 Introduction
0:55 What Is Continuous Profiling and Why It Matters
5:41 Pyroscope v1 Architecture: Ingesters and Their Problems
9:59 Three Core Problems with v1: Availability, Cost, Operability
11:38 Pyroscope v2: Object-Store-First Architecture
14:54 Service-Based Sharding and 95% Symbol Data Reduction
16:57 v2 Read Path: Stateless, Elastic Query Backends
17:51 Production Results: 7 Months Across All Grafana Cloud Regions
20:33 New v2-Only Features: Recording Rules, Exemplars, Heatmaps
21:44 Live Demo: Exemplars, Span Heatmap, and Profiles-to-Traces
25:16 Conclusion: When and Why to Upgrade
Links/resources:
Learn about Pyroscope: https://grafana.com/oss/pyroscope/?src=yt
Read about Pyroscope 2.0: https://grafana.com/blog/pyroscope-2-0-release/?src=yt
Get started with the Grafana Cloud forever-free tier: https://grafana.com/g/cloud
Have a question? Ask Grot, your AI helper: https://grafana.com/grot/
Reach out in our community forums: https://gra.fan/communityyf
---
Thanks for watching!
👍 Was this video helpful? Like and subscribe to our channel for more videos.
Connect with Grafana Labs:
X: (https://www.twitter.com/grafana)
LinkedIn: (https://www.linkedin.com/company/grafana-labs/)
Facebook: (https://www.facebook.com/grafana)
#Grafana #Observability #Pyroscope #Profiles #DevOps #OSS
More from this event
See all 24 talks →
Grafana 13 Deep Dive: Suggested Dashboards, Dynamic Dashboards, SQL Expressions, and More!
1:23:01
Real-time ML Inside a $2M Electron Microscope for Nuclear Powered Data Centers | Theia Scientific
23:05
When Observability Meets a Tamagotchi
5:57
How to Speed Up Grafana Dashboards 100x with ASAPQuery
10:21