OpenTelemetry announces support for profiling

2024-03-19 · Publisher : OpenTelemetry · Domain : Cloud & DevOps

The OpenTelemetry project announces that profiling becomes its fourth signal, after traces, metrics and logs. Elastic and Splunk donate their profiling agents to the project.

### Background On 19 March 2024, the **OpenTelemetry** project announced that it supports **profiling** as its fourth observability signal, alongside traces, metrics and logs. The profiling working group had been created in Valencia at KubeCon Europe 2022, two years earlier. The profiling data model has been merged and the work aims at a stable specification and implementations in 2024. ### Key facts - Contributors named include Datadog, Google, Grafana Labs and Red Hat, among others. - Elastic pledges to donate its eBPF-based continuous profiling agent; Splunk begins donating its .NET profiler. - Profiling is meant to provide bidirectional links with the other signals: going from a resource problem to the function, or even the line of code, responsible. - The project reports more than 1,000 unique monthly contributors from more than 180 companies, and more than 30 million monthly downloads across its main repositories. ### Why it matters Continuous profiling in production used to be the domain of proprietary tools. Its arrival in OpenTelemetry promises a format shared across vendors, hence less dependence on a single agent and profile data that can be correlated with traces. For a services firm, it is a criterion to build into observability requirements, provided the actual maturity of the signal is tracked. ### Takeaway With profiling, OpenTelemetry extends its standard from basic instrumentation to fine-grained analysis of code in production.

Tags : Observabilité, Open Source, Cloud natif