splitql · planned execution

input query
SELECT count(*) AS n, sum(amount) AS s FROM sales WHERE d >= DATE '2026-05-01'
1 fragments6 files2.1 MB scannedpartials table partials

scan + partial

fragment 0 6 files · 2.1 MB · 100% of scan
SQL
SELECT COUNT(*) AS a0, SUM(amount) AS a1 FROM READ_PARQUET(['_demo_data/sales-2026-06.parquet', '_demo_data/sales-2026-05.parquet', '_demo_data/sales-2026-04.parquet', '_demo_data/sales-2026-03.parquet', '_demo_data/sales-2026-02.parquet', '_demo_data/sales-2026-01.parquet']) AS sales WHERE d >= CAST('2026-05-01' AS DATE)
files
  • _demo_data/sales-2026-06.parquet 463.8 KB
  • _demo_data/sales-2026-05.parquet 420.7 KB
  • _demo_data/sales-2026-04.parquet 380.3 KB
  • _demo_data/sales-2026-03.parquet 339.9 KB
  • _demo_data/sales-2026-02.parquet 299.3 KB
  • _demo_data/sales-2026-01.parquet 258.5 KB
→

gather

partials

concatenation of 1 fragment results
(any transport: Arrow, files, UNION ALL)

→

reduce

reduce
SQL
SELECT SUM(a0) AS "n", SUM(a1) AS "s" FROM partials
→

result

✓ identical to single-node