Skip to content

Proposal - OTel Arrow - Phase 3 #3258

@lquerel

Description

@lquerel

Pre-filing checklist

  • I searched existing issues and didn't find a duplicate

Component(s)

Rust OTAP dataflow (rust/otap-dataflow/)

Description

For reference and discussion, I’m copying here the Phase 3 section as originally described in the blog post.

Phase 3 is currently under discussion. It is expected to target a first stable,
production-usable release, with significant improvements in several areas:

  • Pipeline-level control mechanisms: inter-pipeline topics, tenant-aware
    resource governance, live reconfiguration with rollback support, and possible
    controller extension support.
  • Core component ecosystem: receivers, processors, exporters, and extensions
    covering the majority of common telemetry pipelines, maintained within
    OpenTelemetry.
  • Extensibility and processing: a WASM-based extension model for specialized
    components, including exploration of how selected existing OpenTelemetry
    Collector (and collector contrib) components could run inside the Dataflow
    Engine. This also includes a new OTAP-native transform processor with
    OTTL-compatible transformations and the experimental OPL language.
  • SDK-level OTAP export: prototyping OTAP exporters in OpenTelemetry SDKs,
    starting with OTel Rust, to evaluate how much efficiency is gained when
    telemetry starts in an Arrow-friendly representation at the SDK boundary
    instead of being converted later in the pipeline.
  • Ecosystem validation and guidance: collaboration with the OpenTelemetry
    Demo and relevant Blueprint efforts to validate OTAP and the Dataflow Engine
    in realistic end-to-end scenarios.
  • OpenTelemetry Profiles: add support for OpenTelemetry profiles in the OTAP
    representation.

OPL, or OpenTelemetry Processing Language, is currently being specified and
implemented. It is intended to be stream-oriented, strongly typed, and safe for
processing signals that comply with the OpenTelemetry data model. This
processing language could also enable predicate and projection pushdown,
allowing receivers to apply filtering and field-selection requirements closer to
the source when supported.

We are also exploring native integration with agentic workflows, enabling agents
to discover engine capabilities, configuration possibilities, and internal
telemetry, and to safely validate generated configurations. During Phase 3, we
will begin a second round of discussions with the OpenTelemetry Governance
Committee to evaluate how this engine could integrate into the broader
OpenTelemetry ecosystem.

Additional Context

No response

Metadata

Metadata

Assignees

No one assigned

    Type

    No type
    No fields configured for issues without a type.

    Projects

    Status
    No status

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions