Arve Knudsen
Staff Software Engineer @ Grafana Labs
Lachen, Switzerland
Actions
Arve Knudsen is an experienced software engineer, developing Grafana Mimir. He is also a Prometheus maintainer, responsible for OpenTelemetry support.
Links
Area of Expertise
The Future of Metadata in Prometheus: Enhancing Storage and Usability
Prometheus has revolutionized the way we handle monitoring and alerting, yet the management of metadata within the system remains a challenge. Currently, Prometheus stores series metadata per time-series only in memory, with no persistence on disk. As the use of Prometheus expands, the demand for more sophisticated metadata management has become evident. Metadata use cases now extend beyond a single instance per time-series, encompassing metadata that holds independent significance or provides additional context to series or groups of series.
In this talk, we will explore the evolving landscape of metadata in Prometheus. We will discuss the emerging use cases that necessitate a more robust metadata management solution and our vision for addressing these needs. Our goal is to design a persistent metadata store that can efficiently handle diverse metadata requirements, enhancing the flexibility and functionality of Prometheus.
Join us as we delve into the challenges and opportunities of building a metadata storage layer. We will share our initial design considerations, anticipated benefits, and invite the community to contribute to shaping the future of metadata in Prometheus. This session is ideal for developers, system architects, and anyone interested in the inner workings of Prometheus and the future of observability.
Practical OpenTelemetry with Prometheus 3.0
A lot of hard work has gone into making Prometheus 3.0 a great OpenTelemetry (OTel) metrics store. In this talk we dive into the new features and what you need to effectively store and query OTel data inside Prometheus.
OTLP ingestion is now GA.
* This means no more feature flag. But you need to enable it via another flag.
* This means we are committed to making this work at scale and is a major focus of future development.
* OTLP performance has been improved massively:
* Benchmarks show ~49% faster request translation on average.
* Effectively storing and querying OTLP data
* To make OTLP useful, copy some resource attributes. Here is a good starting list.
* Don't copy too many as it will cause issues.
* To make this even more seamless, there is the experimental info function.
* Proper metadata storage is being worked on too.
* Support for delta temporality
* OTLP is not just cumulative, but there is not the delta to cumulative processor to help convert your deltas before storing.
* Proper support for delta in Prometheus is being worked on as well.
* A community mixin for popular semantic conventions
* There is also a community mixin now available (TBD but will likely exist by Sept.)
3.0 is not the end, we are actively working on more improvements. Our goal is to make Prometheus the best OSS store for OTel metrics.
* up metric implementation in push
* . in metric names and how that'll look like
Arve Knudsen
Staff Software Engineer @ Grafana Labs
Lachen, Switzerland
Links
Actions
Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.
Jump to top