Skip to content

[OSPP][Feature]Adding tracing mechanism to Apache Doris #9960

@jialiangCHOU

Description

@jialiangCHOU

Background:

The execution of most jobs in Apache Doris (such as imports, queries, etc.) is performed in a distributed environment. We need to have a Tracing framework to track the execution of these tasks in a distributed environment. Taking a query as an example, the user wants to be able to view the time-consuming of each stage and the statistics of some indicators during the entire execution process of the query, so as to quickly locate the slow node or optimize the query according to the statistical indicators. Currently, Doris provides Profiles to display statistical indicators at various stages of query tasks, but the collection method and data output format of Profiles cannot be well connected with the existing APM system, and operations such as data display and alarms cannot be easily implemented.
On the other hand, by being compatible with the OpenTracing solution, Doris can be used as a link in the enterprise data link, enabling enterprises to complete the distributed tracing of the entire data link.

Requirements:

  1. Investigate the integration solutions of APM systems such as OpenTelemetry and Apache SkyWalking, and produce storage, display, and alarm solutions for tracing data.

  2. Introduce the OpenTelemetry project in Doris, which can support query or import job tracking

  3. Further refine the tracing indicators, which can support the statistics of tracing indicators at the node level and operator level.

Project output requirements:

Tracing indicators of query jobs in Doris can be displayed through an APM system.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions