Skip to content

[WIP] Add Optimizer (query and job recommendations) documentation#37961

Draft
kevinzenghu wants to merge 2 commits into
masterfrom
kevinzenghu/add-optimizer-recommendations-docs
Draft

[WIP] Add Optimizer (query and job recommendations) documentation#37961
kevinzenghu wants to merge 2 commits into
masterfrom
kevinzenghu/add-optimizer-recommendations-docs

Conversation

@kevinzenghu

Copy link
Copy Markdown
Contributor

Summary

  • Documents Optimizer: cost/performance recommendations for Spark/Databricks jobs and Snowflake/BigQuery/Trino queries.
  • Zero prior public documentation despite a real customer-facing surface and being referenced (unlinked) in the top-level DO overview page's "Optimize cost & performance" bullet.

Evidence

  • Backend: data-observability-optimizer — public API (/api/ui/data-observability-optimizer/recommendations, list/detail/status-update), FOR_REVIEW→REVIEWED→IGNORED→RESOLVED lifecycle.
  • Frontend: web-ui/packages/apps/data-jobs/.../JobsRecommendationsPage, RecommendationsSidePanel, RecommendationsSection — confirmed a real UI surface, not just a backend system.
  • Phase: gated per-recommender (Snowflake/Databricks SQL/Trino each have their own flag) plus a Balto org allowlist — Preview per the DO Product Features tracker.

Open items

  • Verify the exact navigation entry point against the live product. Confirmed evidence shows recommendations surface inline in Jobs Monitoring job/cluster tables via a side panel — I did not find a definitive standalone top-nav breadcrumb and deliberately didn't invent one. Please confirm/correct before merging.
  • Confirm current phase (Preview) and which recommenders are actually enabled-by-default vs. allowlist-only, so the docs don't overpromise availability.
  • No screenshot included — flagging per the pipeline's screenshot policy.

Status

Work in progress — not ready for review yet. Draft, no reviewers requested.

AI assistance

Content drafted by Claude Code, grounded in the cited backend API and frontend components — flagging per the AI-assistance disclosure in CONTRIBUTING.md.

Optimizer had a full customer-facing surface (JobsRecommendationsPage,
per-job/cluster recommendation side panels, a public
/api/ui/data-observability-optimizer/recommendations API with a
FOR_REVIEW/REVIEWED/IGNORED/RESOLVED status lifecycle, and a "Fix with
Bits" AI-assisted-fix action) but zero public documentation, despite
being alluded to in the top-level DO overview page's "Optimize cost &
performance" bullet.

Marked Preview per the DO Product Features tracker: individual
recommenders (Snowflake, Databricks SQL, Trino) are gated behind their
own feature flags plus a Balto org allowlist, and notifications are
separately gated off by default.

Deliberately did not claim a specific top-level navigation breadcrumb —
confirmed evidence shows recommendations surface inline in Jobs
Monitoring job/cluster tables via a side panel
(RecommendationsSidePanel, RecommendationsSection), not a fully
independent standalone page reachable from primary nav. Flagged as an
open item to verify the exact entry point against the live product.

AI assistance: drafted by Claude Code, grounded in
data-observability-optimizer (backend) and the data-jobs app's
Recommendations components (frontend).
@github-actions

github-actions Bot commented Jul 5, 2026

Copy link
Copy Markdown
Contributor

Same class of bug just found and fixed for Airbyte: this PR's new
top-level page and its _index.md nextlink were correctly added, but the
separate sidebar-nav config (config/_default/menus/main.en.yaml) was
not — this branch predates discovering that file exists. Added a
top-level entry as a sibling of Quality Monitoring / Jobs Monitoring.
@github-actions github-actions Bot added the Architecture Everything related to the Doc backend label Jul 5, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Architecture Everything related to the Doc backend

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant