Parca

As explained in Cloud Observability in Action

Parca is an open-source continuous profiling platform for collecting and analyzing profiles of running software.

Parca is an open-source continuous profiling platform that collects and analyzes running-software profiles without application instrumentation. It uses eBPF for zero-instrumentation collection and Prometheus service discovery to find profile sources. section 9.1.3

Parca’s practical profile workflow includes:

• Collect profiles from supported languages including C, C++, Rust, Haskell, Java, Python, .NET, and WebAssembly. section 9.1.3
• Scrape a profile-producing endpoint using a Parca configuration and inspect selected profiles as line graphs or Icicle Graphs. section 9.1.5

Profiling an OpenTelemetry Collector

A Docker Compose setup enables the Collector’s pprof extension, runs Parca on port 7070, and configures a two-second scrape interval. After profiles accumulate, selecting Memory In-Use Bytes and opening the line graph reveals an Icicle Graph for troubleshooting resource usage. section 9.1.5

Figure 9.4 Screenshot of the Parca demo environment

Use the Parca demo screenshot to orient yourself to its profile-analysis interface. section 9.1.3

How Cloud Observability in Action covers it

How deep it goes

Explains eBPF-based collection and scraping configuration, but does not detail eBPF internals.

Examples

Demonstrates Parca profiling an OpenTelemetry Collector through Docker Compose, pprof, profile selection, and an Icicle Graph.

What it assumes

Assumes familiarity with profile endpoints and Docker Compose while teaching zero-instrumentation collection and scrape configuration.

What you can do after

Operate/configure Parca to scrape profiles from an OpenTelemetry Collector and inspect resource-usage visualizations.

How it connects

Connects continuous profiling with eBPF, Prometheus service discovery, OpenTelemetry Collector, pprof, and Icicle Graphs.

liveBookmanning.com

Read Cloud Observability in Action on liveBook Get the book at manning.com

Related concepts

• eBPFuses · eBPF is a Linux kernel technology that safely executes verified, sandboxed programs in response to kernel, networking, system, or application events.
• Prometheususes · Prometheus is an open-source monitoring and alerting system that collects, stores, and queries time-series metrics.
• OpenTelemetry Collectoruses · The OpenTelemetry Collector is a vendor-neutral service that receives, processes, and exports telemetry data.
• continuous profilingkind of · Continuous profiling is the ongoing collection and analysis of execution or resource-usage profiles from running software over time.
• profilinguses · Profiling is the measurement and analysis of a program’s runtime behavior and resource use to characterize performance and identify bottlenecks.

This explainer was generated by Manning's knowledge graph from the text of Cloud Observability in Action. The section links point to the parts of the book it draws on.