Slide 1

Slide 1 text

Apache™ Pig An Introduction Moty Michaely June, 2015

Slide 2

Slide 2 text

Outline What is Apache™ Pig Motivation Hands On Use Cases Questions

Slide 3

Slide 3 text

Outline What is Apache™ Pig Motivation Hands On Use Cases Questions

Slide 4

Slide 4 text

Outline What is Apache™ Pig Motivation Hands On Use Cases Questions

Slide 5

Slide 5 text

Outline What is Apache™ Pig Motivation Hands On Use Cases Questions

Slide 6

Slide 6 text

Outline What is Apache™ Pig Motivation Hands On Use Cases Questions

Slide 7

Slide 7 text

What Is Apache™ Pig

Slide 8

Slide 8 text

“High-level platform for creating MapReduce programs used with Hadoop.” (Wikipedia)

Slide 9

Slide 9 text

Apache™ Pig is High Level Scripting Language Pig Latin Hadoop MapReduce/Tez Compiler Commonly Used Open Source

Slide 10

Slide 10 text

Apache™ Pig is High Level Scripting Language Hadoop MapReduce/Tez Compiler Supports MapReduce and Tez Commonly Used Open Source

Slide 11

Slide 11 text

Apache™ Pig is High Level Scripting Language Hadoop MapReduce/Tez Compiler Commonly Used Netflix, Xplenty, eBay, Yahoo, Wix... Open Source

Slide 12

Slide 12 text

Apache™ Pig is High Level Scripting Language Hadoop MapReduce/Tez Compiler Commonly Used Open Source Backed by the community

Slide 13

Slide 13 text

Why Pig (the name)? Pigs Eat Anything Relational, Nested, Unstructured Files, Key/Value stores, Databases Pigs Live Anywhere Pigs Are Domestic Animals Pigs Fly

Slide 14

Slide 14 text

Why Pig (the name)? Pigs Eat Anything Pigs Live Anywhere Not tied to particular framework Pigs Are Domestic Animals Pigs Fly

Slide 15

Slide 15 text

Why Pig (the name)? Pigs Eat Anything Pigs Live Anywhere Pigs Are Domestic Animals Easily controlled Integration of user code Pigs Fly

Slide 16

Slide 16 text

Why Pig (the name)? Pigs Eat Anything Pigs Live Anywhere Pigs Are Domestic Animals Pigs Fly Faster development Improved performance

Slide 17

Slide 17 text

Apache™ Pig architecture

Slide 18

Slide 18 text

APACHE PIG - Pig Latin scripting language - MR/Tez Compiler = RECAP

Slide 19

Slide 19 text

Motivation for Apache™ Pig

Slide 20

Slide 20 text

Motivation for Pig Increase productivity 10 lines of Pig Latin ≈ 200 lines of Java 4 hours of Java ≈ 15 minutes of Pig Latin Open to non-java developers Optimization opportunities Extensibility

Slide 21

Slide 21 text

Motivation for Pig Increase productivity Open to non-java developers It’s like SQL Optimization opportunities Extensibility

Slide 22

Slide 22 text

Motivation for Pig Increase productivity Open to non-java developers Optimization opportunities No need to tune Hadoop for your needs Execution plan, optimizer Extensibility

Slide 23

Slide 23 text

Motivation for Pig Increase productivity Open to non-java developers Optimization opportunities Extensibility User defined functions Integration with Python, Ruby and JS

Slide 24

Slide 24 text

MOTIVATION - Productivity - Non-java developers - Optimization - Extensibility = RECAP

Slide 25

Slide 25 text

Hands On Apache™ Pig

Slide 26

Slide 26 text

Pig Latin (For Apache™ Pig) “A high-level language that allows you to write data processing and analysis programs.”

Slide 27

Slide 27 text

Pig Latin For Apache™ Pig Relations - A relation (table) is a bag - A bag is a collection of tuples - A tuple (row) is an ordered set of fields - A field is a piece of data

Slide 28

Slide 28 text

Running Pig - Two execution modes Interactive Mode $ cd /path/to/pig/bin/ $ pig grunt> a = LOAD ‘/path/to/file’; grunt> DUMP a; Batch Mode

Slide 29

Slide 29 text

Running Pig Pig supports two execution modes Interactive Mode Batch Mode $ cd /path/to/pig/bin/ $ pig -f /path/to/pig/file.pig

Slide 30

Slide 30 text

Word Count Problem “Given text files, return how often words occur”

Slide 31

Slide 31 text

Word Count in MR

Slide 32

Slide 32 text

Word Count in MR (Mapper)

Slide 33

Slide 33 text

Word Count in MR (Reducer)

Slide 34

Slide 34 text

Word Count in MR

Slide 35

Slide 35 text

Word Count in Pig -- Word Count Script (wordcount.pig) text = LOAD 'word_count_text.txt'; words = FOREACH text GENERATE FLATTEN(TOKENIZE((chararray)$0)) AS word; grouped_words = GROUP words BY word; counts = FOREACH grouped_words GENERATE group AS word, COUNT(words) AS count; STORE counts INTO 'wordcount';

Slide 36

Slide 36 text

Word Count in Pig -- Word Count Script (wordcount.pig) text = LOAD 'word_count_text.txt'; words = FOREACH text GENERATE FLATTEN(TOKENIZE((chararray)$0)) AS word; grouped_words = GROUP words BY word; counts = FOREACH grouped_words GENERATE group AS word, COUNT(words) AS count; STORE counts INTO 'wordcount';

Slide 37

Slide 37 text

-- Word Count Script (wordcount.pig) text = LOAD 'word_count_text.txt'; words = FOREACH text GENERATE FLATTEN(TOKENIZE((chararray)$0)) AS word; grouped_words = GROUP words BY word; counts = FOREACH grouped_words GENERATE group AS word, COUNT(words) AS count; STORE counts INTO 'wordcount'; Word Count in Pig

Slide 38

Slide 38 text

Word Count in Pig -- Word Count Script (wordcount.pig) text = LOAD 'word_count_text.txt'; words = FOREACH text GENERATE FLATTEN(TOKENIZE((chararray)$0)) AS word; grouped_words = GROUP words BY word; counts = FOREACH grouped_words GENERATE group AS word, COUNT(words) AS count; STORE counts INTO 'wordcount';

Slide 39

Slide 39 text

Word Count in Pig -- Word Count Script (wordcount.pig) text = LOAD 'word_count_text.txt'; words = FOREACH text GENERATE FLATTEN(TOKENIZE((chararray)$0)) AS word; grouped_words = GROUP words BY word; counts = FOREACH grouped_words GENERATE group AS word, COUNT(words) AS count; STORE counts INTO 'wordcount';

Slide 40

Slide 40 text

Word Count in Pig -- Word Count Script (wordcount.pig) text = LOAD 'word_count_text.txt'; words = FOREACH text GENERATE FLATTEN(TOKENIZE((chararray)$0)) AS word; grouped_words = GROUP words BY word; counts = FOREACH grouped_words GENERATE group AS word, COUNT(words) AS count; STORE counts INTO 'wordcount';

Slide 41

Slide 41 text

Word Count in Pig -- Word Count Script (wordcount.pig) text = LOAD 'word_count_text.txt'; words = FOREACH text GENERATE FLATTEN(TOKENIZE((chararray)$0)) AS word; grouped_words = GROUP words BY word; counts = FOREACH grouped_words GENERATE group AS word, COUNT(words) AS count; STORE counts INTO 'wordcount';

Slide 42

Slide 42 text

Word Count in Pig - Sorted -- Word Count Script (wordcount.pig) text = LOAD 'word_count_text.txt'; words = FOREACH text GENERATE FLATTEN(TOKENIZE((chararray)$0)) AS word; grouped_words = GROUP words BY word; counts = FOREACH grouped_words GENERATE group AS word, COUNT(words) AS count; sorted_counts = ORDER counts BY count DESC, word ASC; STORE counts INTO 'wordcount';

Slide 43

Slide 43 text

Word Count in Pig - Sorted -- Word Count Script (wordcount.pig) text = LOAD 'word_count_text.txt'; words = FOREACH text GENERATE FLATTEN(TOKENIZE((chararray)$0)) AS word; grouped_words = GROUP words BY word; counts = FOREACH grouped_words GENERATE group AS word, COUNT(words) AS count; sorted_counts = ORDER counts BY count DESC, word ASC; STORE sorted_counts INTO 'wordcount_sorted';

Slide 44

Slide 44 text

Word Count - MR vs. Pig 63 Lines of code 5 lines of code -- Word Count Script (wordcount.pig) text = LOAD 'word_count_text.txt'; words = FOREACH text GENERATE FLATTEN(TOKENIZE((chararray)$0)) AS word; grouped_words = GROUP words BY word; counts = FOREACH grouped_words GENERATE group AS word, COUNT(grouped_words) AS count; STORE counts INTO 'wordcount';

Slide 45

Slide 45 text

HANDS ON - Word Count - Execution Plan - Optimization = RECAP

Slide 46

Slide 46 text

Use Cases Of Apache™ Pig

Slide 47

Slide 47 text

“80% of the work in any data project is in cleaning the data.” (D.J Patel, Data Jujitsu)

Slide 48

Slide 48 text

Pig is great for Web log processing Data processing for web search platforms Ad hoc queries across large data sets Rapid prototyping of algorithms for processing large data sets

Slide 49

Slide 49 text

Pig is great for Web log processing Data processing for web search platforms Ad hoc queries across large data sets Rapid prototyping of algorithms for processing large data sets

Slide 50

Slide 50 text

Pig is great for Web log processing Data processing for web search platforms Ad hoc queries across large data sets Rapid prototyping of algorithms for processing large data sets

Slide 51

Slide 51 text

Pig is great for Web log processing Data processing for web search platforms Ad hoc queries across large data sets Rapid prototyping of algorithms for processing large data sets

Slide 52

Slide 52 text

Questions

Slide 53

Slide 53 text

Resources Apache Pig Philosophy Apache Pig 0.14 Documentation Word Count Example Introduction to Apache Tez Pig (Language) Pig for dummies Pig Latin (Language Game) Xplenty Data Jujitsu: The art of turning data into product Pig Cheat Sheet

Slide 54

Slide 54 text

THANK YOU [email protected]