In the dynamic realm of cloud-native application observability, OpenTelemetry has recently celebrated a significant milestone that marks a new era in its developmental journey. Reflecting on two fruitful years since the inception of the Profiling Special Interest Group (SIG) at KubeCon + CloudNativeCon Europe 2022 in Valencia, we stand at a pivotal juncture. The latest achievement, the introduction of a profiling data model OTEP, paves the way for a stable specification and implementation set to unfold within this year. This development is a giant leap towards realising the long-standing vision of embedding observability into cloud-native applications – a goal OpenTelemetry has nurtured from its outset.
Collaborative Milestones and Profiling’s Evolution:
The journey of enhancing OpenTelemetry’s profiling capabilities has been a collective effort guided by the dedicated Profiling SIG. This group has thrived on the rich diversity of its contributors, drawing from the expertise of profiling vendors and users alike. Noteworthy contributions have come from industry stalwarts such as Felix Geisendörfer of Datadog, Alexey Alexandrov of Google, Dmitry Filimonov and Ryan Perry of Grafana Labs, and Jonathan Halliday of Red Hat. The central task has been to pinpoint the optimal data format for profiling—a mission that has sparked extensive discussions and proposals within OpenTelemetry’s vibrant community.
OpenTelemetry’s path to this milestone has been marked by several key achievements, notably aligning the profiling vision in August 2022 and proposing v1 and v2 of the profiling data models in the latter stages of 2023. These strides underscore the commitment to a development ethos that is inclusive and collaborative, welcoming a spectrum of insights and expertise.
Demystifying Profiling:
But what is profiling, and why does it matter? Profiling is a dynamic analysis method that delves into the behaviour and performance of application code during its execution. Continuous profiling uncovers granular insights into code-level resource utilisation, supporting storing, querying, and analysing profiling data over time and across different dimensions. This approach is a boon for developers and performance engineers, offering a deep dive into the mechanics of code operations. OpenTelemetry’s foray into profiling is pioneering, linking profiling data with other telemetry types to provide a holistic view of resource use and pinpoint performance bottlenecks to the offending lines of code.
A Unified Observability Experience:
The introduction of OpenTelemetry’s profiling capabilities has been met with enthusiasm across the industry, bolstered by significant contributions from Elastic and Splunk. These partnerships have supercharged progress, integrating profiling technologies to enrich OpenTelemetry’s arsenal.
This enhancement heralds a more interconnected observability experience for users. Profiling data will be seamlessly linked with other telemetry types, such as logs, metrics, and traces. This integration simplifies the observability landscape, allowing users to transition effortlessly from general telemetry data to in-depth profiling insights, streamlining their understanding of application performance.
Looking Ahead:
The momentum behind OpenTelemetry is a testament to the dedication of its contributors. With over 1,000 developers from over 180 companies contributing monthly and code repositories amassing over 30 million downloads a month, OpenTelemetry sits at the core of an expanding observability movement. The future is bright as OpenTelemetry continue to grow, improve, and diversify tools throughout 2024 and beyond.
For more details and to join this transformative project, visit the OpenTelemetry website: https://opentelemetry.io/
Original insights by Austin Parker (Honeycomb), Tuesday, March 19, 2024.
