BASE: Bridging the Gap between Cost and Latency for Query Optimization

BASE is a two-stage RL-based framework that bridges the gap between cost and latency for training a learned query optimizer efficiently. For more details, check out the VLDB 2023 paper: BASE: Bridging the Gap between Cost and Latency for Query Optimization.

Dataset

The pre-training queries for IMDB are available at this link.

Usage

BASE is a two stage training framework for learned query optimizer:

• First stage: policy pretraining with cost from PostgreSQL cost model.

python -u RLGOOTest_NEW.py --path ./Models/now.pth --epoch_start 1 --epoch_end 100 --epsilon_decay 0.95 --epsilon_end 0.02 --capacity 60000 --batch_size 512 --sync_batch_size 50 --steps_per_epoch 1000 --max_lr 0.0008 --learning_rate 0.0003 > log_NEW.txt 2>&1

• Second stage (starting from the first stage): latency finetuning with latency from actual execution.

python Transfer_Active.py

Contact

[email protected]