Grafana's Pyroscope 2.0 Makes Continuous Profiling Practical at Scale
Scored daily by a customisable AI persona to surface the most relevant engineering leadership news.
Pyroscope 2.0 improves continuous profiling, relevant for performance engineering.
Grafana Labs released Pyroscope 2.0, a complete rearchitecture of its continuous profiling database that eliminates write-path replication (reducing storage from 3x to 1x) and makes the read path fully stateless for elastic scaling. The new design stores profiles once in object storage, deduplicates symbolic information to cut symbol storage by 95%, and enables new capabilities like metrics derived from profiles and single-profile inspection. Deployments that took 8-12 hours in v1 now complete in minutes, and the architecture supports bursty query patterns from LLM-powered agents.
Adopt Pyroscope 2.0 to identify function-level bottlenecks in your agentic workloads and reduce observability storage costs by up to 95%.
For engineers building agentic systems and optimizing cloud costs, Pyroscope 2.0 makes continuous profiling practical at scale, enabling function-level optimization of LLM agent workloads and reducing idle capacity overhead.