Scaling your data infrastructure

Slide 1

Slide 1 text

Scaling your data infrastructure C H R I S T I A N B A R R A @ P Y C O N N O V E

Slide 2

Slide 2 text

THE AGENDA 2 3 START THE DATA SCIENCE WORKFLOW SCALING IS NOT JUST A MATTER OF MACHINE WHEN THE SIZE OF YOUR DATA MATTERS 1

Slide 3

Slide 3 text

THE AGENDA 4 5 CONTAINERIZED DATA SCIENCE CASSINY: PUT ALL THE THINGS TOGETHER END

Slide 4

Slide 4 text

THE DATA SCIENCE WORKFLOW

Slide 5

Slide 5 text

HEXAGON PRESENTATION TEMPLATE

Slide 6

Slide 6 text

HOW YOU BUILD, ITERATE AND SHARE DEPENDS ON MANY THINGS Your Users Your Product Your Team Your Company Your Tech Stack Your Domain

Slide 7

Slide 7 text

SCIKIT-LEARN DOCKER DATA SCIENCE TOOLBELT PANDAS JUPYTER RAY

Slide 8

Slide 8 text

SCALING IS NOT JUST A MATTER OF MACHINES

Slide 9

Slide 9 text

We all use it.

Slide 10

Slide 10 text

We really care about versioning. We have Untitled_1.ipynb, Untitled_2.ipynb and Untitled_3.ipynb. HOMER SIMPSON C H I E F D A T A S C I E N T I S T D A T A B E E R I N C

Slide 11

Slide 11 text

Since JSON is a plain text format, they can be version-controlled and shared with colleagues. E X I P Y T H O N N O T E B O O K D O C U M E N T A T I O N

Slide 12

Slide 12 text

THEY GOT IT RIGHT

Slide 13

Slide 13 text

BUT WE KEEP IMPROVING

Slide 14

Slide 14 text

90% OF JUPITER IS MADE BY HYDROGEN

Slide 15

Slide 15 text

THE HARD THING ABOUT STORAGE

Slide 16

Slide 16 text

PARQUET P A R Q U E T + O B J E C T S T O R A G E = YO U C A N Q U E R Y I T U S I N G S Q L PA N DA S H A S N AT I V E S U P P O R T F O R G E T A B O U T C S V

Slide 17

Slide 17 text

WHEN THE SIZE OF YOUR DATA MATTERS

Slide 18

Slide 18 text

IT’S TOO SLOW DOESN’T FIT IN YOUR RAM

Slide 19

Slide 19 text

CODE OPTIMIZATION APPROACH SCALING FROM DIFFERENT SIDES A BIGGER MACHINE USE MULTIPLE CORES MORE MACHINES FRAMEWORKS: DASK RAY SPARK PANDAS: READ BY CHUNKS SCIKIT-LEARN: PARTIAL FIT

Slide 20

Slide 20 text

chunks & partial_fit 1 M A C H I N E

Slide 21

Slide 21 text

Multiple machines. n M A C H I N E S

Slide 22

Slide 22 text

I don’t want to use Spark/JVM, what do you have for me? H A P P Y P Y T H O N U S E R

Slide 23

Slide 23 text

WHAT IS RAY?

Slide 24

Slide 24 text

A high-performance distributed execution engine REDIS SCHEDULER WORKER ARROW & PLASMA

Slide 25

Slide 25 text

Use pandas through ray to query parquet files in an object storage. W O R K I N P R O G R E S S

Slide 26

Slide 26 text

CONTAINERIZED DATA SCIENCE

Slide 27

Slide 27 text

If you trained a model with scikit-learn 0.18.1, will the same model work with 0.19.1? P R O B L E M # 1

Slide 28

Slide 28 text

How do you share your models? P R O B L E M # 2

Slide 29

Slide 29 text

How do you put your models in production? P R O B L E M # 3

Slide 30

Slide 30 text

Containerize everything. T H E A N S W E R

Slide 31

Slide 31 text

1. It’s damn easy to move things around 2. You get versioning for free 3. Stack agnostic 4. Move Docker images around T O R E C A P

Slide 32

Slide 32 text

CASSINY: PUT ALL THE THINGS TOGETHER

Slide 33

Slide 33 text

CLEAR REQUIREMENTS CONTAINERIZED EASY OBJECT STORAGE JUPYTER + IPYTHON PLATFORM AGNOSTIC

Slide 34

Slide 34 text

OPEN SOURCE

Slide 35

Slide 35 text

DEMO

Slide 36

Slide 36 text

TAKEAWAYS UNIFIED DATA WAREHOUSE KEEP YOUR CODE RUNNING ON ONE MACHINE USE DOCKER TRY RAY BRING CI/CD TO YOUR DATASCIENCE WORKFLOW OBJECT STORAGE IS COOL DISTRIBUTED COMPUTING IS HARD I DIDN’T HAVE ANOTHER POINT

Slide 37

Slide 37 text

No content