Data pipeline Last load succeeded Last run 2026.09.16 19:48 KST Stores 2026.06 · Licences 2026.09

Load history

Last 5 runs
DatasetAs of RowsTime Status
District borders #6통계청 SGIS 행정동 경계 (vuski/admdongkor) HangJeongDong_ver20260701.geojson 2026.07 427 0m 05s Succeeded
Business licences #5서울 열린데이터광장 LOCALDATA_072405 · LOCALDATA_072404 2026.09 684,712 3m 39s Succeeded
Business licences #4서울 열린데이터광장 LOCALDATA_072405 · LOCALDATA_072404 2026.09 685,035 2m 36s Failed · Duplicate key in the source data
Reloaded in #5 and live again Why did it fail?
Cause
The source contained the same licence number (REST, 3100000-104-2026-00079) twice, so the primary key couldn't be built.
Fix
Now only the most recently updated row is kept per licence number, and the dropped count is reported as "Duplicate licence number" under data quality.
Show the original error
ERROR: could not create unique index "license_load_pkey"
  세부 정보: Key (service, mgt_no)=(REST, 3100000-104-2026-00079) is duplicated.
  위치: parallel worker
Store data #3소상공인시장진흥공단_상가(상권)정보_서울_202606.csv 2026.06 554,092 0m 40s Succeeded
Store data #2소상공인시장진흥공단_상가(상권)정보_서울_202606.csv 2026.06 554,092 0m 53s Succeeded

Store data #3 step timings ·Prepare 21 ms·Parse · COPY 17.9 s·Primary key · dedupe 1.6 s·Indexes 11.9 s·Quality checks 3.6 s·Pre-aggregate cells 5.4 s·Swap in 84 ms

Business licences #5 step timings ·Prepare 11 ms·Fetch · COPY 152.7 s·Reproject · match addresses 5.8 s·Dedupe 7.6 s·Indexes 13.4 s·Verify coordinates 39.5 s·Quality checks 84 ms·Swap in 49 ms

District borders #6 step timings ·Prepare 16 ms·Load dongs 694 ms·Merge district borders 114 ms·Indexes 14 ms·Verify borders 4.6 s·Swap in 29 ms

Performance, measured

p95 response time of the same query run repeatedly

Same-type stores within 500 m

15× faster
Full scan 56.6 ms
Spatial index 3.7 ms

GiST (geography) + ST_DWithin · the full scan uses 2 parallel workers · 554,092 stores · p95 · Full scan 20 runs / Spatial index 100 runs

Seoul-wide heatmap

51× faster
Raw GROUP BY 40.8 ms
Pre-aggregated 0.8 ms

Food stores · resolution 7 · pre-aggregated table vs GROUP BY over 554,092 rows · p95 · Raw GROUP BY 20 runs / Pre-aggregated 100 runs