U
About
Grafana Pyroscope is an open source continuous profiling platform designed to help engineering teams understand exactly how their applications consume CPU, memory, and other resources—right down to the specific line of code responsible. As part of the broader Grafana observability ecosystem, Pyroscope shares architectural principles with Grafana Mimir, Loki, and Tempo, enabling seamless correlation of profiling signals alongside metrics, logs, and traces in a single pane of glass. Pyroscope operates as a multi-tenant, horizontally scalable microservices system, making it suitable for both small teams and large enterprises. Applications can be instrumented using official SDKs across multiple languages or via auto-instrumentation through Grafana Alloy, lowering the barrier to adoption. Profiling data is stored continuously, allowing developers to query and analyze production performance at any point in time rather than relying on one-off snapshots. Visualization options include flame graphs for call-stack hot spot analysis, statistical tables for function-level detail, and time-series charts for trend comparison. Pyroscope's HTTP API and server-side configuration options (disk storage, tenant IDs, shuffle sharding, proxies) give platform teams granular control over deployment. It is the ideal tool for SREs, backend engineers, and DevOps teams who need deep application performance insights without the overhead of commercial APM solutions.
Key Features
- Continuous Profiling: Continuously captures CPU, memory, and resource usage profiles from running applications, enabling historical analysis rather than one-off snapshots.
- Flame Graph Visualization: Renders interactive flame graphs that map call relationships and instantly surface hot spots in application code.
- Grafana Observability Integration: Natively integrates with Grafana, Mimir, Loki, and Tempo to correlate profiling data alongside metrics, logs, and traces in a unified dashboard.
- Multi-Tenant Architecture: Horizontally scalable microservices design supports multiple tenants and high-availability deployments suitable for enterprise environments.
- Flexible Instrumentation: Supports both SDK-based manual instrumentation across major languages and zero-code auto-instrumentation via Grafana Alloy.
Use Cases
- Identifying CPU and memory hot spots in production microservices to reduce infrastructure costs and improve response times.
- Correlating a latency spike in Grafana traces with code-level profiling data to pinpoint the exact function causing the slowdown.
- Running continuous profiling in CI/CD pipelines to catch performance regressions before they reach production.
- Providing SRE and platform teams with a unified observability platform covering metrics, logs, traces, and profiling in one Grafana workspace.
- Analyzing long-term resource usage trends across multiple services to guide capacity planning and architectural decisions.
Pros
- Fully Open Source: Apache-licensed and community-driven, making it free to self-host with no vendor lock-in or licensing costs.
- Deep Grafana Ecosystem Fit: First-class integration with Grafana dashboards, Mimir, Loki, and Tempo enables true unified observability in a single tool.
- Scalable Microservices Design: Each component can scale independently, enabling the platform to handle high-throughput production workloads with ease.
- Low Instrumentation Barrier: Auto-instrumentation via Grafana Alloy means teams can start collecting profiles without modifying application code.
Cons
- Operational Complexity: Self-hosting a multi-component, distributed system requires significant infrastructure knowledge and ongoing maintenance effort.
- Grafana Dependency for Best Experience: While a standalone UI exists, the richest analysis and correlation features require a working Grafana instance.
- Limited Language SDK Coverage vs. Commercial APMs: Some niche or newer languages may have less mature SDK support compared to commercial alternatives like Datadog or New Relic.
Frequently Asked Questions
What is continuous profiling and how does Pyroscope support it?
Continuous profiling is the practice of collecting application performance data (CPU, memory, goroutines, etc.) non-stop in production. Pyroscope aggregates these profiles over time, letting you query and compare performance at any historical moment rather than during a single debugging session.
Is Grafana required to use Pyroscope?
No. Pyroscope ships with its own UI for querying and visualizing profiles. However, Grafana integration unlocks the ability to correlate profiling data alongside metrics (Mimir), logs (Loki), and traces (Tempo) in a single interface.
How do I instrument my application with Pyroscope?
You can instrument using official language SDKs (Go, Python, Java, Ruby, .NET, and others) for code-level control, or use Grafana Alloy for auto-instrumentation that requires no application code changes.
Does Pyroscope support multi-tenancy?
Yes. Pyroscope is designed as a multi-tenant system from the ground up, supporting tenant ID configuration so multiple teams or environments can share a single deployment with data isolation.
Is there a hosted version of Pyroscope?
Yes. Grafana Cloud Profiles is the managed, hosted version of Pyroscope available through Grafana Cloud, eliminating the need to self-host and operate the infrastructure yourself.
