DeepOLA: Experiment Tracker

[nikhil96sher]

Experiment Setup

• Azure VM
• Data Generation: TBL
  • Scale = 50: TPC-H Original Data
  • Scale = 100: TPC-H Original Data
• Data Conversion: Parquet (Both Partition = 1 and Partition = 512M)
  • Scale = 100: TPC-H Original Data
  • Scale = 100: TPC-H Skewed Data
• Things to Remember
  • Important: Disable Cache
  • Mount /data/ in read-only mode if you only need to copy the data.

WAKE

• Implementation: 22 Queries Without Scaling
• Implementation: 22 Queries With Scaling
• Script to set up the repository and run experiments
• Latency Results
  • Scale = 100; Partition = 512M dataset.
  • Skewed Scale = 100; Partition = 512M dataset.
• Error Results
  • Scale = 100; Partition = 512M dataset.
  • Skewed Scale = 100; Partition = 512M dataset.
• Variation of Partitions Experiment
• Variation of Scaling Methods Experiment

Presto

• Experiment Script
• Latency Results on Scale = 100; No partitions dataset.

Polars

• Experiment Script
• Latency Results on Scale = 100; No partitions dataset.

Postgres

• Experiment Script
• Latency Results on Scale = 100; No partitions dataset.

ProgressiveDB

• Experiment Script
• Latency
  • Scale = 100; Partition = 512M
  • Skewed Scale = 100; Partition = 512M
• Error
  • Scale = 100; Partition = 512M
  • Skewed Scale = 100; Partition = 512M

WanderJoin

• Experiment Script
• Latency
  • Scale = 100; Partition = 512M
  • Skewed Scale = 100; Partition = 512M
• Error
  • Scale = 100; Partition = 512M
  • Skewed Scale = 100; Partition = 512M

[pyongjoo]

@mIXs222 Please leave a comment here if you finish adding placeholder figures. Overleaf link: https://www.overleaf.com/project/6213dd85e75ce371a781fed9