Skip to content

Conversation

@gabotechs
Copy link
Collaborator

@gabotechs gabotechs commented Jan 2, 2026

This PR:

  • Refactors the code in order to better hold new query engines in the remote benchmarks
  • Adds ballista to benchmarks

TPCH SF1 Ballista (prev) VS DataFusion Distributed (new)

  q1.sql: prev=1058 ms, new= 534 ms, 1.98x faster ✅
  q2.sql: prev=2410 ms, new= 561 ms, 4.30x faster ✅
  q3.sql: prev=2144 ms, new= 809 ms, 2.65x faster ✅
  q4.sql: prev=1539 ms, new= 379 ms, 4.06x faster ✅
  q5.sql: prev=2750 ms, new= 621 ms, 4.43x faster ✅
  q6.sql: prev= 773 ms, new= 395 ms, 1.96x faster ✅
  q7.sql: prev=2630 ms, new= 747 ms, 3.52x faster ✅
  q8.sql: prev=3425 ms, new=1012 ms, 3.38x faster ✅
  q9.sql: prev=3054 ms, new=1028 ms, 2.97x faster ✅
 q10.sql: prev=2367 ms, new= 513 ms, 4.61x faster ✅
 q11.sql: prev=1678 ms, new= 417 ms, 4.02x faster ✅
 q12.sql: prev=1455 ms, new= 432 ms, 3.37x faster ✅
 q13.sql: prev=1280 ms, new= 452 ms, 2.83x faster ✅
 q14.sql: prev=1140 ms, new= 449 ms, 2.54x faster ✅
 q15.sql: prev=1754 ms, new= 675 ms, 2.60x faster ✅
 q16.sql: prev=1930 ms, new= 328 ms, 5.88x faster ✅
 q17.sql: prev=1614 ms, new= 864 ms, 1.87x faster ✅
 q18.sql: prev=2073 ms, new= 591 ms, 3.51x faster ✅
 q19.sql: prev=1256 ms, new= 528 ms, 2.38x faster ✅
 q20.sql: prev=1901 ms, new= 709 ms, 2.68x faster ✅
 q21.sql: prev=2508 ms, new=1263 ms, 1.99x faster ✅
 q22.sql: prev=1282 ms, new= 327 ms, 3.92x faster ✅

TPCH SF10 Ballista (prev) VS DataFusion Distributed (new)

  q1.sql: prev=2857 ms, new=2896 ms, 1.01x slower ✖
  q2.sql: prev=4327 ms, new=1298 ms, 3.33x faster ✅
  q3.sql: prev=4306 ms, new=2302 ms, 1.87x faster ✅
  q4.sql: prev=2389 ms, new=1998 ms, 1.20x faster ✔
  q5.sql: prev=6682 ms, new=3707 ms, 1.80x faster ✅
  q6.sql: prev=1369 ms, new=1332 ms, 1.03x faster ✔
  q7.sql: prev=9022 ms, new=4351 ms, 2.07x faster ✅
  q8.sql: prev=8147 ms, new=3720 ms, 2.19x faster ✅
  q9.sql: prev=9715 ms, new=5664 ms, 1.72x faster ✅
 q10.sql: prev=5563 ms, new=2649 ms, 2.10x faster ✅
 q11.sql: prev=2945 ms, new=1331 ms, 2.21x faster ✅
 q12.sql: prev=3007 ms, new=1395 ms, 2.16x faster ✅
 q13.sql: prev=2538 ms, new=1881 ms, 1.35x faster ✅
 q14.sql: prev=2164 ms, new=1410 ms, 1.53x faster ✅
 q15.sql: prev=3398 ms, new=1966 ms, 1.73x faster ✅
 q16.sql: prev=2330 ms, new= 785 ms, 2.97x faster ✅
 q17.sql: prev=7066 ms, new=4540 ms, 1.56x faster ✅
 q18.sql: prev=7958 ms, new=4136 ms, 1.92x faster ✅
 q19.sql: prev=2389 ms, new=1687 ms, 1.42x faster ✅
 q20.sql: prev=3142 ms, new=1942 ms, 1.62x faster ✅
 q21.sql: prev=9751 ms, new=7852 ms, 1.24x faster ✅
 q22.sql: prev=1958 ms, new= 697 ms, 2.81x faster ✅

Clickbench partitioned 100 files Ballista (prev) VS DataFusion Distributed (new)

  q0.sql: prev= 398 ms, new= 140 ms, 2.84x faster ✅
  q1.sql: prev=1398 ms, new= 824 ms, 1.70x faster ✅
  q2.sql: prev=1619 ms, new=1120 ms, 1.45x faster ✅
  q3.sql: prev=1612 ms, new=1178 ms, 1.37x faster ✅
  q4.sql: prev=2581 ms, new=1818 ms, 1.42x faster ✅
  q5.sql: prev=3735 ms, new=2601 ms, 1.44x faster ✅
  q6.sql: prev= 442 ms, new= 143 ms, 3.09x faster ✅
  q7.sql: prev=1454 ms, new= 849 ms, 1.71x faster ✅
  q8.sql: prev=3864 ms, new=2338 ms, 1.65x faster ✅
  q9.sql: prev=3424 ms, new=2628 ms, 1.30x faster ✅
 q10.sql: prev=2370 ms, new=1373 ms, 1.73x faster ✅
 q11.sql: prev=2289 ms, new=1378 ms, 1.66x faster ✅
 q12.sql: prev=3836 ms, new=2815 ms, 1.36x faster ✅
 q13.sql: prev=5125 ms, new=5314 ms, 1.04x slower ✖
 q14.sql: prev=4131 ms, new=2525 ms, 1.64x faster ✅
 q15.sql: prev=3264 ms, new=2288 ms, 1.43x faster ✅
 q16.sql: prev=5331 ms, new=4085 ms, 1.31x faster ✅
 q17.sql: prev=5866 ms, new=3753 ms, 1.56x faster ✅
 q18.sql: prev=8507 ms, new=7133 ms, 1.19x faster ✔
 q19.sql: prev=1251 ms, new=1066 ms, 1.17x faster ✔
 q20.sql: prev=5839 ms, new=5865 ms, 1.00x slower ✖
 q21.sql: prev=6133 ms, new=5967 ms, 1.03x faster ✔
 q22.sql: prev=10154 ms, new=9672 ms, 1.05x faster ✔
 q23.sql: prev=23440 ms, new=24130 ms, 1.01x slower ✖
 q24.sql: prev=2134 ms, new=1990 ms, 1.07x faster ✔
 q25.sql: prev=1824 ms, new=1593 ms, 1.15x faster ✔
 q26.sql: prev=2248 ms, new=2030 ms, 1.11x faster ✔
 q27.sql: prev=6449 ms, new=6503 ms, 1.01x slower ✖
 q28.sql: prev=22974 ms, new=26314 ms, 1.15x slower ✖
 q29.sql: prev=1960 ms, new=1792 ms, 1.09x faster ✔
 q30.sql: prev=3358 ms, new=2627 ms, 1.28x faster ✅
 q31.sql: prev=3334 ms, new=3134 ms, 1.06x faster ✔
 q32.sql: prev=7403 ms, new=8373 ms, 1.13x slower ✖
 q33.sql: prev=12352 ms, new=9336 ms, 1.32x faster ✅
 q34.sql: prev=13255 ms, new=9108 ms, 1.46x faster ✅
 q35.sql: prev=3846 ms, new=2599 ms, 1.48x faster ✅
 q36.sql: prev=1284 ms, new= 791 ms, 1.62x faster ✅
 q37.sql: prev= 919 ms, new= 496 ms, 1.85x faster ✅
 q38.sql: prev=1291 ms, new= 742 ms, 1.74x faster ✅
 q39.sql: prev=1659 ms, new=1186 ms, 1.40x faster ✅
 q40.sql: prev= 955 ms, new= 451 ms, 2.12x faster ✅
 q41.sql: prev= 972 ms, new= 444 ms, 2.19x faster ✅
 q42.sql: prev= 958 ms, new= 417 ms, 2.30x faster ✅

@gabotechs gabotechs force-pushed the gabrielmusat/ballista-benchmarks branch from 3de9721 to b7da3f8 Compare January 2, 2026 16:12
@gabotechs gabotechs force-pushed the gabrielmusat/ballista-benchmarks branch from b7da3f8 to a42d4a2 Compare January 2, 2026 16:38
@gabotechs gabotechs changed the title Draft: add ballista to benchmarks Rework AWS CDK code for the benchmarking cluster and add ballista to benchmarks Jan 2, 2026
@gabotechs gabotechs marked this pull request as ready for review January 2, 2026 16:48
Copy link
Collaborator

@NGA-TRAN NGA-TRAN left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Wow, great numbers. Congrats again!

@gabotechs gabotechs merged commit 086b89f into main Jan 6, 2026
7 checks passed
@gabotechs gabotechs deleted the gabrielmusat/ballista-benchmarks branch January 6, 2026 19:32
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants