Grafana Phlare: Open‑Source Continuous Profiling Database – Architecture, Features, and Kubernetes Deployment Guide

Grafana Phlare is an open‑source project for aggregating continuous profiling data, fully integrated with Grafana to correlate profiling information with other observability signals such as metrics, logs, and traces.

Continuous profiling continuously collects resource‑usage information from distributed cloud‑native applications, compresses it into time‑series data, and enables visualization of performance trends and hotspot analysis, forming the fourth pillar of observability.

Key features of Grafana Phlare include easy single‑binary installation, horizontal scalability, high availability through profile replication, cheap durable storage using object storage (compatible with S3, GCS, Azure Blob, OpenStack Swift, etc.), and native multi‑tenant isolation.

The system follows a micro‑service architecture where components such as Distributor , Ingester , and Querier can run independently; a -target flag selects which components to launch, allowing both a monolithic mode ( -target=all ) and a micro‑service mode.

Deployment is performed via Helm charts on a Kubernetes cluster. A namespace (e.g., phlare-test ) is created, the Helm repository added, and the chart installed either in monolithic mode with helm -n phlare-test install phlare grafana/phlare or in micro‑service mode using a custom values-micro-services.yaml file that configures component replica counts and resource limits.

After deployment, the Pods (agent, distributor, ingester, querier, MinIO, etc.) should reach the Running state. Grafana can then be installed in the same cluster, configured with a Phlare datasource pointing to http://phlare-querier.phlare-test.svc.cluster.local.:4100/ , and used to explore profiling data alongside metrics and logs.

Phlare also supports automatic pod scraping via annotations ( phlare.grafana.com/scrape="true" and phlare.grafana.com/port="8080" ) and uses standard relabel_config and kubernetes_sd_config mechanisms similar to Prometheus.

References: https://github.com/grafana/phlare https://grafana.com/blog/2022/11/02/announcing-grafana-phlare-oss-continuous-profiling-database/