Tuplex: Blazing Fast Python Data Science
Tuplex is a parallel big data processing framework that runs data science pipelines written in Python at the speed of compiled code. Tuplex has similar Python APIs to Apache Spark or Dask, but rather than invoking the Python interpreter, Tuplex generates optimized LLVM bytecode for the given pipeline and input data set. Under the hood, Tuplex is based on data-driven compilation and dual-mode processing, two key techniques that make it possible for Tuplex to provide speed comparable to a pipeline written in hand-optimized C++.
You can join the discussion on Tuplex on our Gitter community or read up more on the background of Tuplex in our SIGMOD'21 paper.
Contributions welcome!
Contents
Installation
To install Tuplex, you can use a PyPi package for Linux, or a Docker container for MacOS which will launch a jupyter notebook with Tuplex preinstalled.
Docker
docker run -p 8888:8888 tuplex/tuplex
PyPI
pip install tuplex
Building
Tuplex is available for MacOS and Linux. The current version has been tested under MacOS 10.13-10.15 and Ubuntu 18.04 and 20.04 LTS. To install Tuplex, simply install the dependencies first and then build the package.
MacOS build from source
To build Tuplex, you need several other packages first which can be easily installed via brew.
brew install [email protected] boost boost-python3 aws-sdk-cpp pcre2 antlr4-cpp-runtime googletest gflags yaml-cpp celero
python3 -m pip cloudpickle numpy
python3 setup.py install
Ubuntu build from source
To faciliate installing the dependencies for Ubuntu, we do provide two scripts (scripts/ubuntu1804/install_reqs.sh
for Ubuntu 18.04, or scripts/ubuntu2004/install_reqs.sh
for Ubuntu 20.04). To create an up to date version of Tuplex, simply run
./scripts/ubuntu1804/install_reqs.sh
python3 -m pip cloudpickle numpy
python3 setup.py install
Customizing the build
Besides building a pip package, cmake can be also directly invoked. To compile the package via cmake
mkdir build
cd build
cmake ..
make -j$(nproc)
The python package corresponding to Tuplex can be then found in build/dist/python
with C++ test executables based on googletest in build/dist/bin
.
To customize the cmake build, the following options are available to be passed via -D
:
option | values | description |
---|---|---|
CMAKE_BUILD_TYPE |
Release (default), Debug , RelWithDebInfo , tsan , asan , ubsan |
select compile mode. Tsan/Asan/Ubsan correspond to Google Sanitizers. |
BUILD_WITH_AWS |
ON (default), OFF |
build with AWS SDK or not. On Ubuntu this will build the Lambda executor. |
GENERATE_PDFS |
ON , OFF (default) |
output in Debug mode PDF files if graphviz is installed (e.g., brew install graphviz ) for ASTs of UDFs, query plans, ... |
PYTHON3_VERSION |
3.6 , ... |
when trying to select a python3 version to build against, use this by specifying major.minor . To specify the python executable, use the options provided by cmake. |
LLVM_ROOT_DIR |
e.g. /usr/lib/llvm-9 |
specify which LLVM version to use |
BOOST_DIR |
e.g. /opt/boost |
specify which Boost version to use. Note that the python component of boost has to be built against the python version used to build Tuplex |
For example, to create a debug build which outputs PDFs use the following snippet:
cmake -DCMAKE_BUILD_TYPE=Debug -DGENERATE_PDFS=ON ..
Example
Tuplex can be used in python interactive mode, a jupyter notebook or by copying the below code to a file. To try it out, run the following example:
from tuplex import *
c = Context()
res = c.parallelize([1, 2, None, 4]).map(lambda x: (x, x * x)).collect()
# this prints [(1, 1), (2, 4), (4, 16)]
print(res)
More examples can be found here.
License
Tuplex is available under Apache 2.0 License, to cite the paper use:
@inproceedings{10.1145/3448016.3457244,
author = {Spiegelberg, Leonhard and Yesantharao, Rahul and Schwarzkopf, Malte and Kraska, Tim},
title = {Tuplex: Data Science in Python at Native Code Speed},
year = {2021},
isbn = {9781450383431},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3448016.3457244},
doi = {10.1145/3448016.3457244},
booktitle = {Proceedings of the 2021 International Conference on Management of Data},
pages = {1718–1731},
numpages = {14},
location = {Virtual Event, China},
series = {SIGMOD/PODS '21}
}
(c) 2017-2021 Tuplex contributors