Dask Tutorial

Dask provides multi-core execution on larger-than-memory datasets.

We can think of dask at a high and a low level

• High level collections: Dask provides high-level Array, Bag, and DataFrame collections that mimic NumPy, lists, and Pandas but can operate in parallel on datasets that don't fit into main memory. Dask's high-level collections are alternatives to NumPy and Pandas for large datasets.
• Low Level schedulers: Dask provides dynamic task schedulers that execute task graphs in parallel. These execution engines power the high-level collections mentioned above but can also power custom, user-defined workloads. These schedulers are low-latency (around 1ms) and work hard to run computations in a small memory footprint. Dask's schedulers are an alternative to direct use of threading or multiprocessing libraries in complex cases or other task scheduling systems like Luigi or IPython parallel.

Different users operate at different levels but it is useful to understand both. This tutorial will interleave between high-level use of dask.array and dask.dataframe (even sections) and low-level use of dask graphs and schedulers (odd sections.)

Prepare

You will need the following core libraries

conda install numpy pandas h5py Pillow matplotlib scipy toolz pytables snakeviz

And a recently updated version of dask

conda/pip install dask
conda/pip install distributed

You may find the following libraries helpful for some exercises

pip install graphviz cachey

Windows users can install graphviz as follows

1. Install Graphviz from http://www.graphviz.org/Download_windows.php
2. Add C:\Program Files (x86)\Graphviz2.38\bin to the PATH
3. Run "pip install graphviz" on the command line

You should clone this repository

git clone http://github.com/dask/dask-tutorial

and then run this script to prepare artificial data.

cd dask-tutorial
python prep.py

Links

• Reference
  • Docs
  • Code
  • Blog
• Ask for help
  • dask tag on Stack Overflow
  • github issues for bug reports and feature requests
  • blaze-dev mailing list for community discussion
  • Please ask questions during a live tutorial

Outline

1. Overview - dask's place in the universe

2. Foundations - low-level Dask and how it does what it does

3. Bag - the first high-level collection: a generalized iterator for use with a functional programming style and o clean messy data.

4. Distributed - Dask's scheduler for clusters, with details of how to view the UI.

5. Array - blocked numpy-like functionality with a collection of numpy arrays spread across your cluster.

6. Advanced Distributed - further details on distributed computing, including how to bebug.

7. Dataframe - parallelized operations on many pandas dataframes spread across your cluster.

8. Dataframe Storage - efficient ways to read and write dataframes to disc.