• Software Installation
  • Set up PyMySQL
• Datasets
• Run experiments
  • Scale data set
  • Generate range queries
  • Run DeepMapping-R
• Run baselines
  • VerdictDB
  • DBEst++
  • Synopsis
  • DeepMapping-R variants

Software Installation

Note that the software requirement for DeepMapping-R is different from VerdictDB and DBEst++. For VerdictDB and DBEst++, please refer to their original repositories for installation instructions. For VerdictDB, we used python API and pymyql. According to their document, python 3.7 is required in conda environment.

For DeepMapping-R, install

python -m pip install -r requirements.txt

Set up PyMySQL

To set up default password for root user, follow here:

sudo mysql -u root # I had to use "sudo" since it was a new installation

mysql> USE mysql;
mysql> UPDATE user SET plugin='mysql_native_password' WHERE User='root';
mysql> FLUSH PRIVILEGES;
mysql> exit;

sudo service mysql restart

Datasets

Datasets are publicly available from the web. You can download them via the following links.

PM25, CCPP, Flights, TPC-DS

We generated the Store Sales table with the TPC-DS tools with SF=1.0.

dsqgen --SCALE 1 --OUTPUT_DIR tpcds_data_SF1

Run experiments

Scale data set

The dataset is first scale up using IDEBench.

bash bash_scipts/run_datagen.sh

Generate range queries

After data is prepared, generated the range queries:

bash bash_scripts/deepmapping/run_query_gen.sh

Run DeepMapping-R

Run DeepMapping-R:

bash bash_scripts/deepmapping/pm25.sh
bash bash_scripts/deepmapping/ccpp.sh
bash bash_scripts/deepmapping/flights.sh
bash bash_scripts/deepmapping/store_sales.sh

Run baselines

There are five baselines: VerdictDB, DBEst++, Synopsis, two DeepMapping Variants

VerdictDB

First, insert data into MySQL database

bash bash_scripts/verdictdb/add_data.sh

Then you can run the queries with different sample ratios

bash bash_scripts/verdictdb/diff_sample_ratio.sh

You can run the insertion and then query.

bash bash_scripts/verdictdb/insert.sh

DBEst++

You can run DBEst++ by downloading their repository, and copy all files in "DBEst++" in our repository to "experiments" folder. You can run the queries:

bash run_train.sh

To run the insertion experiment:

bash run_insert.sh

If you want to know the model size, you can get a model size chart by running

python get_model_sizes.py

Synopsis

"Synopsis" is directly using the training data of DeepMapping-R to answer range queries.

bash bash_scripts/deepmapping/run_synopsis.py

DeepMapping-R variants

Run two DeepMapping-R variants NHP and NHR:

bash bash_scripts/deepmapping/run_deepmapping_variants.py