pertool

Pertool is a modular open-source toolkit for performance engineering, intended to simplify instrumentation, profiling, tracing, and reliability testing across software systems. It provides a core runtime, a plugin system, and language adapters to collect and analyze runtime data from applications, services, and infrastructure.

The architecture centers on a lightweight core that sequences data collection and processing, with plugins that extend instrumentation for particular languages, runtimes, or platforms. Pertool collects CPU and memory profiles, I/O, and custom events, and models data as events, samples, and traces. Export backends include standard observability stacks such as OpenTelemetry and Prometheus, as well as proprietary or other third-party sinks. A built-in visualizer supports flame graphs, latency distributions, traces, and throughput charts. The tool is designed to be non-invasive and can operate in development, staging, and production environments with minimal overhead.

Usage and workflows: Developers instrument code directly via pertool APIs or run a standalone agent as a sidecar or daemon. The CLI and REST API enable configuration, collection, and querying of data. Typical workflows include CPU and memory profiling during development, distributed tracing across services, and anomaly detection through baselines and alerting.

History and status: The project was announced in 2023 by a coalition of contributors and has since evolved through community involvement. It aims to complement existing tools by offering a cohesive, extensible framework for performance instrumentation.

See also: OpenTelemetry, Prometheus, profiling, tracing, observability.