Skip to content

Complete the pluggable data warehouse abstraction #225

@vinaysrao1

Description

@vinaysrao1

Is your feature request related to a problem? Please describe.
Step two in abstracting the Osprey codebase away from assuming Druid downstream. Follow-up to #188.

What solution would you like?
Finish the decoupling work so the neutral interface is the only path to the analytics database.

Key changes expected:

  1. Extract the Druid read path into a first-class backend plugin registered through the same hookspec as any other backend
  2. Migrate the bulk-label path off Druid-specific query objects and onto the neutral backend interface
  3. Check and complete all migrations (validators, UDFs etc) so no shared type parameterizes on a Druid-shaped model
  4. Move the Druid client library to an optional install extra so ClickHouse-only deployments don't pull it in

Describe alternatives you have considered
Treat Druid as a "privileged" backend. Rejected — leaves the system unable to actually retire Druid and defeats the goal of #188.

Do you have any additional context?
This is prerequisite work for Step 3

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No fields configured for Task.

    Projects

    Status

    Todo

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions