-
Couldn't load subscription status.
- Fork 1k
Description
Is your feature request related to a problem? Please describe.
Apache Iceberg is a data lake table format that has a lot of interesting features. Iceberg provides an SDK that makes it easy for engines to implement its specifications. Furthermore, the SDK emits metrics regarding table scan planning and commit handling (see: https://iceberg.apache.org/docs/nightly/metrics-reporting/).
This feature would introduce OpenTelemetry instrumentation for the Apache Iceberg SDK.
Describe the solution you'd like
In the figure, you can see how the Iceberg SDK handles metrics reporting. Custom metric reporters can be associated with TableScan objects during runtime, and they will be used to consume the emitted metrics (MetricReport). The desired solution would allow database engines that use the Iceberg SDK to consume the metrics emitted by this SDK and convert them into OpenTelemetry metrics.
Describe alternatives you've considered
To add an OpenTelemetry-based Metric Reporter directly into Iceberg SDK. However, this would not benefit the SDK versions before such an addition is introduced.
Additional context
Multiple existing open source database engines that implement the Iceberg specs already use OpenTelemetry, e.g., Snowflake, Trino, Starburst, and Doris. All of these engines, and others, would benefit from the proposed feature.
Tip
React with 👍 to help prioritize this issue. Please use comments to provide useful context, avoiding +1 or me too, to help us triage it. Learn more here.