Fulfilling Apache Arrow's Promises: Pandas on JVM memory without a copy

Apache Arrow's promise was to reduce the (serialization & copy) overhead of working with columnar data between different systems. Using the latest Pandas release and Arrow's ability to share memory between the JVM and Python as ingredients, we demonstrate that Arrow can fulfill this bold statement. The performance benefits of this will be shown using a typical data engineering use-case that produces data in the JVM and then passes it on to a Python-based machine learning model.

Transcript

1. 1
   Fulfilling Apache Arrow's Promises:
   Pandas on JVM memory without a copy
   PyCon.DE Karlsruhe 2018
   Uwe L. Korn
   

   View Slide

2. 2
   • Senior Data Scientist at Blue Yonder
   (@BlueYonderTech)
   • Apache {Arrow, Parquet} PMC
   • Data Engineer and Architect with heavy
   focus around Pandas
   About me
   xhochy
   mail@uwekorn.com
   

   View Slide

3. 3
   What’s Apache Arrow?
   • Published in February 2016
   • Specification for in-memory columnar data layout
   • No overhead for cross-system communication
   • Designed for efficiency (exploit SIMD, cache locality, ..)
   • Exchange data without conversion between Python, C++, C(glib), Ruby,
   Lua, R, JavaScript, Go, Rust, Matlab and the JVM
   • Brought Parquet to Pandas and made PySpark fast (@pandas_udf)
   

   View Slide

4. 4
   February 2016: Birth of Apache Arrow
   Just a goal…
   

   View Slide

5. 5
   Data Science Workflow in 2018
   Python
   machine
   learning
   model
   pre-processing
   with pandas
   probability density
   function (PDF)
   SQL
   Engine
   

   View Slide

6. 6
   Looks simple?
   • It isn’t.
   • „Data“ is very heterogeneous landscape
   • Most common setup:
   • Java/Scala, i.e. JVM, for data processing
   • Python for machine learning
   

   View Slide

7. 7
   Data Science Workflow in 2018
   Python
   machine
   learning
   model
   pre-processing
   with pandas
   SQL
   Engine
   JDBC Driver JayDeBeApi
   P
   Y
   T
   H
   O
   N
   R
   O
   W
   S
   J
   D
   B
   C
   R
   O
   W
   S
   

   View Slide

8. 8
   org.apache.arrow.adapter.jdbc
   • Retrieve JDBC results as Arrow RecordBatch / VectorSchemaRoot
   • Do conversion of rows to columns in the JVM
   • Data is stored „off-heap“, i.e:
   • not managed by the JVM
   • native memorly layout, same as in pyarrow
   

   View Slide

9. 9
   Workflow in 2018 with Arrow
   Python
   machine
   learning
   model
   pre-processing
   with pandas
   SQL
   Engine
   JDBC Driver
   org.apache.
   arrow.adapter.
   jdbc
   A
   R
   R
   O
   W
   J
   D
   B
   C
   R
   O
   W
   S
   ?
   

   View Slide

10. 10
    So we’re done? No.
    • We still only have Arrow data in the JVM
    • Arrow and Pandas have a slightly different memory layout
    • We have this today in PySpark
    • It’s fast
    • Still involves a copy over the network
    • Arrow → pandas conversion is tuned but still a copy
    

    View Slide

11. 11
    pyarrow.jvm
    • Access Arrow data created in the JVM from Python
    • Involves no copy of the data
    • Translation of the helper objects
    • Actually passes memory addresses around
    No copy between the JVM and Python!
    

    View Slide

12. NumPy & the BlockManager
    Photo by Susan Holt Simpson on Unsplash
    

    View Slide

13. 13
    Pandas Shortcomings
    • Limited to NumPy data types, otherwise object
    • Columns are not separate, grouped by type
    • Nullability is not type-safe (yet)
    —> Arrow memory does not match Pandas memory
    —> Copy
    

    View Slide

14. 14
    Pandas ExtensionArrays
    • Introduced new interfaces in 0.23
    • ExtensionDtype
    • What type of scalars?
    • ExtensionArray
    • Implement basic array ops
    • Pandas provides algorithms on top
    • Still, experimental, wait for 0.24
    

    View Slide

15. 15 Photo by Niklas Tidbury on Unsplash
    

    View Slide

16. 16
    fletcher
    • https://github.com/xhochy/fletcher
    • Implements Extension{Array,Dtype} with Apache Arrow as storage
    • Uses Numba to implement the necessary analytic on top
    • Needs {pandas, Arrow, …} master
    No copy between Apache Arrow and pandas!
    

    View Slide

17. 17
    Workflow in 2018 with Arrow
    Python
    machine
    learning
    model
    pre-processing
    with pandas
    SQL
    Engine
    JDBC Driver
    org.apache.
    arrow.adapter.
    jdbc
    A
    R
    R
    O
    W
    J
    D
    B
    C
    R
    O
    W
    S
    pyarrow.jvm

    /
    fletcher
    

    View Slide

18. 18
    ???
    Does it work?
    

    View Slide

19. 19
    Does it work?
    

    View Slide

20. 20
    Does it work?
    

    View Slide

21. Make your
    best decision
    today.
    blueyonder.ai/en/careers
    Blue Yonder Analytics, Inc.
    5048 Tennyson Parkway
    Suite 250
    Plano, Texas 75024
    USA
    21
    

    View Slide

22. Cross language DataFrame library
    • Website: https://arrow.apache.org/
    • ML: dev@arrow.apache.org
    • Issues & Tasks: https://issues.apache.org/jira/
    browse/ARROW
    • Slack: https://
    apachearrowslackin.herokuapp.com/
    • Github mirror: https://github.com/apache/
    arrow
    Apache Arrow Apache Parquet
    Famous columnar file format
    • Website: https://parquet.apache.org/
    • ML: dev@parquet.apache.org
    • Issues & Tasks: https://issues.apache.org/jira/
    browse/PARQUET
    • Slack: https://parquet-slack-
    invite.herokuapp.com/
    • C++ Github mirror: https://github.com/
    apache/parquet-cpp
    22
    Get Involved!
    

    View Slide