Data Governance · Operating in Public

The framework,
running.

Most portfolios say "I designed data governance frameworks." This page is the framework. Every number on djtakerun.jp comes from BigQuery. Every master record has a Sheets-shaped editor. Every transformation is in Git. Every sync runs nightly. Click around.

① Lineage — where data comes from.

Every asset tracked end-to-end. Click any node to see full metadata — owner, schedule, quality, and upstream / downstream dependencies.

SOURCES
SOURCE
Google Sheets
dim_* · 5 master tables
on edit ~520 rows
SOURCE
Strava API
OAuth 2.0 · activities
nightly 23:00 JST ~1,200
SOURCE
Mixcloud API
public endpoint · mixes
nightly 23:00 JST 69 mixes
SOURCE
Recordbox CSV
manual export · tracks
manual 232 tracks
LAKE
LAKE
GCS Raw Lake
gs://takerun-lake/raw/
nightly 23:00 JST immutable
WAREHOUSE · BigQuery
RAW
raw_sheets.*
orbital-age-442710-b1
nightly mirror
MASTER
dim.* / ref.*
5 dim + 4 ref · MERGE on PK
nightly MERGE ~520 rows
FACT
fact.*
fact_runs · fact_sessions
nightly (partial) ~1,200 rows
MART
mart.*
4 aggregation views
on query views
SERVING
CDN
GCS CDN
export_json · *.json
5 min cache 6 endpoints
LIVE
djtakerun.jp
Cloud Run · nginx
always on fetch /data/*.json

Click any node to see asset metadata, lineage, and schedule details.

Column-level lineage

— loading from mart.mart_governance_column_lineage …
Source column → Target column Transformation
loading…

② Catalog — every table, every column.

Tier Dataset Table Type Created
loading from /data/governance_catalog.json…

③ Framework — who owns what, and the rules.

Ownership, quality contracts, access tiers, and retention — the four pillars that make the lineage and catalog below actionable. Pipeline topology (what connects to what) lives in the Lineage section.

◈ Ownership
TakeRun (Editorial)
Google Sheets · Mixcloud · Rekordbox
manual update
GCP Service Account
Cloud Functions · Strava OAuth
automated
System (BigQuery)
All schemas — raw · dim · fact · mart
IAM managed
◎ Quality SLA
Freshness
Sheets / Mixcloud ≤ 24h
Strava: not deployed
Completeness (%C/A)
raw.*: 100% · dim.*: ≥ 99%
mart.*: ≥ 98%
Alert Threshold
≥ 48h delay → critical
◐ Access Control
Public
mart.* exports CDN JSON
djtakerun.jp — anonymous read
Internal Only
dim.* fact.* raw.*
GCP IAM — TakeRun only
PII Policy
No PII stored
All metrics aggregated / anonymised
◷ Lifecycle
GCS Raw Lake
30-day partitions
Auto-delete after window
BigQuery
raw.*: 90d · fact.*: 2yr
dim.*: indefinite mart.*: indefinite
Backup
Weekly snapshot
GCS Coldline · 90-day retention

④ Tools & Architecture — how it's all wired.

Every tool in the stack, and how data flows between them. Click any node for details.

LOCAL · DEV
Windows 11
dev machine
Claude Code
AI CLI · code agent
◢
Rekordbox
DJ library → CSV
DATA SOURCES · SaaS
Google Sheets
master data editor
Strava
running activities
Mixcloud
DJ mixes & stats
Spotify
listening history
Notion
PKB · operating notes
Gmail
subscriptions & receipts
B978
biblia978.com
reading log
GCP · PIPELINE
Cloud Functions
ingest_* · export_json
BigQuery
warehouse · 6 datasets
Cloud Storage
raw lake + CDN JSON
Data Lineage
column-level graph
SERVING
Cloud Run
nginx · containerised
DJ
djtakerun.jp
public website

Click any node to see details.

⑤ Sync Status — last touched.

Source Rows Last sync Platforms
loading…

⑥ Data Library — what's actually in BigQuery.

Live snapshot of every dataset, table, and view in orbital-age-442710-b1. Numbers come from __TABLES__ + INFORMATION_SCHEMA via the mart_governance_data_library view. Click any table to expand its columns.

Dataset Table / View Type Cols Rows Size (KB)
loading…

⑦ Decisions — why this, not that.