SlideShare una empresa de Scribd logo
1 de 10
Descargar para leer sin conexión
Getting started with Hadoop, Hive,
and Elastic MapReduce
Hunter Blanks, hblanks@monetate.com / @tildehblanks
github.com/hblanks/talks/




                                                      PhillyAWS. January 12, 2012
Overview
In the figure below, 4 inputs are mapped to 16 (key, value) tuples.




These tuples are sorted by those keys, and then reduced, into 4 new (key, value) tuples.
Finally, these new tuples are once again being sorted and reduced, and so on...
                                                   source: http://code.google.com/p/mapreduce-framework/wiki/MapReduce
Timeline

        2004                                2008         2009             2010             2011




    MapReduce paper Doug Cutting    Hadoop wins     Elastic MapReduce             mrjob
       published     (and other     terabyte sort        launched.                open
                    Yahoos) start    benchmark      Facebook presents            sourced
                   work on Hadoop                   first paper on Hive.
An overview of available (and not available) tools
1) MapReduce. Google's C++ implementation, which sits atop GFS, chubby,
BigTable, and who knows what else. See also sawzall.

2) Hadoop. Apache's/Facebook's/Yahoo's Java implementation, which
furthermore includes, or else integrates with: HDFS (a reimplementation of
GFS), Zookeeper (a reimplementation of chubby), and HBase (a
reimplementation of BigTable). See also Pig (not exactly sawzall, but another
take on it).

3) Hive. Apache's/Facebook's data warehousing system, which allows users to
query large datasets using SQL-like syntax, over Hadoop.

4) Elastic MapReduce. Amazon's API for spinning up temporary Hadoop
clusters (aka "Jobflows"), typically reading input and writing output to S3. It is
much, much easier than bringing up your own Hadoop cluster.

5) mrjob. Yelp's Python framework for running MapReduce jobs on Hadoop
clusters, with or without ElasticMapReduce.
Terms for Elastic MapReduce
1) job flow. A transient (or semi-transient) Hadoop cluster. Jobflows typically
terminate after all their steps are done, but they can also be flagged as
"permanent," which just means they don't shut down on their own.

2) master, core, and task nodes. A jobflow has one master node; when it
dies, your jobflow dies (or is at least supposed to). It has a fixed number of
core nodes (determined when the jobflow was started) -- these core nodes can
do map/reduce tasks, and also serve HDFS. It can also have a changeable
number of task nodes, which do tasks but don't serve HDFS.

3) step. A jobflow contains one or more "steps". These may or may not be
actual Hadoop "jobs" -- sometimes a step is just for initialization, other times it
is an actual map/reduce job.

4) spot instance vs on-demand. By default, any job flow you start will use
normal, "on-demand" EC2 instances, with a fixed hourly price. You may
alternatively specify a spot instances by naming a maximum price (a bid price)
you're willing to pay. If your bid price is below the current spot price, then you
will pay the spot price for your instances (typically 50-66% cheaper than on-
demand instances).
The easiest ways to get started with Elastic MapReduce, #1
If you know Python, start with mrjob.

Here's a sample MRJob class, which you might write to wordcounter.py:

    from mrjob.job import MRJob

    class MRWordCounter(MRJob):
        def mapper(self, key, line):
            for word in line.split():
                yield word, 1

        def reducer(self, word, occurrences):
            yield word, sum(occurrences)

    if __name__ == '__main__':
        MRWordCounter.run()



And how you run it:

    export AWS_ACCESS_KEY_ID=...; export AWS_SECRET_ACCESS_KEY=...
    python wordcounter.py -r emr < input > output

These samples, and much more documentation, is at http://
packages.python.org/mrjob/writing-and-running.html.
The easiest ways to get started with Elastic MapReduce, #2
If you know any other scripting language, start with the AWS elastic-mapreduce
command line tool, or the AWS web console. Both are outlined at:

  http://docs.amazonwebservices.com/ElasticMapReduce/latest/GettingStartedGuide/

An example of the command line, assuming a file wordSplitter.py, is:

   ./elastic-mapreduce --create --stream 
       --mapper s3://elasticmapreduce/samples/wordcount/wordSplitter.py 
       --input s3://elasticmapreduce/samples/wordcount/input 
       --output s3://mybucket.foo.com/wordcount 
       --reducer aggregate

A (not so) quick reference card for the command line tool is at:

  http://s3.amazonaws.com/awsdocs/ElasticMapReduce/latest/emr-qrc.pdf
Getting started with Hive is not so easy
1) You're going to need to put your data into some sort of warehouse, probably in S3.
(HDFS will not persist in the cloud). This may involve its own MapReduce step, but with
careful attention to how you output your files to S3. Partitioning matters, but you won't
get it with just the default Hadoop streaming jar! (We use oddjob's)

2) To start a jobflow with Hive programmatically (i.e., not using the command line elastic-
mapreduce client), you need to add several initial steps to the jobflow -- notably ones
that start Hive and possibly configure its heapsize. Fortunately, Amazon's elastic-
mapreduce client is easy enough to reverse engineer...

3) You're going to need to make a schema (surprise) for your data warehouse.

4) You may have warehoused data in JSON. Hive much prefers flat, tabular files.

5) To talk to Hive programmatically from other servers, you'll need to use Hive's thrift
binding, and also open up the ElasticMapReduce-master security group (yes, EMR
created this for you when you first ran a jobflow) so that your application box can talk to
the master node of your job flow.

6) All that said, it may beat writing your own custom MapReduce jobs.
Other pitfalls to beware
1) The JVM is not your friend. It especially loves memory.

- be prepared to add swap on your task nodes, at least until Hadoop's fork()/exec()
  behavior is fixed in AWS' version of Hadoop

- be prepared to profile your cluster with Ganglia (it's really, really easy to setup). Just
  add this bootstrap action to do so:
  s3://elasticmapreduce/bootstrap-actions/install-ganglia

2) Multiple output files can get you into trouble, especially if you don't sort your outputs
first.

3) Hadoop, and Hive as well, much prefer larger files to lots of small files.
Further reading

 - "MapReduce: Simplified Data Processing on Large Clusters"
   OSDI'04. http://research.google.com/archive/mapreduce.html

 - "Hive - A Warehousing Solution Over a Map-Reduce Framework"
   VLDB, 2009. http://www.vldb.org/pvldb/2/vldb09-938.pdf

 - Hadoop: The Definitive Guide. O'Reilly Associates, 2009.


Thank you!




                                        (P.S. We're hiring. hblanks@monetate.com)

Más contenido relacionado

La actualidad más candente

Introduction to the Hadoop Ecosystem (IT-Stammtisch Darmstadt Edition)
Introduction to the Hadoop Ecosystem (IT-Stammtisch Darmstadt Edition)Introduction to the Hadoop Ecosystem (IT-Stammtisch Darmstadt Edition)
Introduction to the Hadoop Ecosystem (IT-Stammtisch Darmstadt Edition)Uwe Printz
 
Integration of HIve and HBase
Integration of HIve and HBaseIntegration of HIve and HBase
Integration of HIve and HBaseHortonworks
 
The Hadoop Ecosystem
The Hadoop EcosystemThe Hadoop Ecosystem
The Hadoop EcosystemJ Singh
 
Hadoop demo ppt
Hadoop demo pptHadoop demo ppt
Hadoop demo pptPhil Young
 
Intro to Hadoop
Intro to HadoopIntro to Hadoop
Intro to Hadoopjeffturner
 
Introduction to Hadoop
Introduction to HadoopIntroduction to Hadoop
Introduction to Hadoopjoelcrabb
 
Hadoop online training
Hadoop online training Hadoop online training
Hadoop online training Keylabs
 
Hadoop Administration pdf
Hadoop Administration pdfHadoop Administration pdf
Hadoop Administration pdfEdureka!
 
Pig, Making Hadoop Easy
Pig, Making Hadoop EasyPig, Making Hadoop Easy
Pig, Making Hadoop EasyNick Dimiduk
 
Hadoop hive presentation
Hadoop hive presentationHadoop hive presentation
Hadoop hive presentationArvind Kumar
 
Learning Apache HIVE - Data Warehouse and Query Language for Hadoop
Learning Apache HIVE - Data Warehouse and Query Language for HadoopLearning Apache HIVE - Data Warehouse and Query Language for Hadoop
Learning Apache HIVE - Data Warehouse and Query Language for HadoopSomeshwar Kale
 
Seminar Presentation Hadoop
Seminar Presentation HadoopSeminar Presentation Hadoop
Seminar Presentation HadoopVarun Narang
 

La actualidad más candente (20)

Introduction to the Hadoop Ecosystem (IT-Stammtisch Darmstadt Edition)
Introduction to the Hadoop Ecosystem (IT-Stammtisch Darmstadt Edition)Introduction to the Hadoop Ecosystem (IT-Stammtisch Darmstadt Edition)
Introduction to the Hadoop Ecosystem (IT-Stammtisch Darmstadt Edition)
 
Hadoop basics
Hadoop basicsHadoop basics
Hadoop basics
 
Integration of HIve and HBase
Integration of HIve and HBaseIntegration of HIve and HBase
Integration of HIve and HBase
 
The Hadoop Ecosystem
The Hadoop EcosystemThe Hadoop Ecosystem
The Hadoop Ecosystem
 
Hadoop demo ppt
Hadoop demo pptHadoop demo ppt
Hadoop demo ppt
 
Intro to Hadoop
Intro to HadoopIntro to Hadoop
Intro to Hadoop
 
Hadoop Family and Ecosystem
Hadoop Family and EcosystemHadoop Family and Ecosystem
Hadoop Family and Ecosystem
 
Introduction to Pig
Introduction to PigIntroduction to Pig
Introduction to Pig
 
Introduction to Hadoop
Introduction to HadoopIntroduction to Hadoop
Introduction to Hadoop
 
Hadoop online training
Hadoop online training Hadoop online training
Hadoop online training
 
מיכאל
מיכאלמיכאל
מיכאל
 
Hadoop overview
Hadoop overviewHadoop overview
Hadoop overview
 
Hadoop pig
Hadoop pigHadoop pig
Hadoop pig
 
Hadoop Administration pdf
Hadoop Administration pdfHadoop Administration pdf
Hadoop Administration pdf
 
Pig, Making Hadoop Easy
Pig, Making Hadoop EasyPig, Making Hadoop Easy
Pig, Making Hadoop Easy
 
Hadoop hive presentation
Hadoop hive presentationHadoop hive presentation
Hadoop hive presentation
 
An Introduction to Hadoop
An Introduction to HadoopAn Introduction to Hadoop
An Introduction to Hadoop
 
Learning Apache HIVE - Data Warehouse and Query Language for Hadoop
Learning Apache HIVE - Data Warehouse and Query Language for HadoopLearning Apache HIVE - Data Warehouse and Query Language for Hadoop
Learning Apache HIVE - Data Warehouse and Query Language for Hadoop
 
Hadoop
HadoopHadoop
Hadoop
 
Seminar Presentation Hadoop
Seminar Presentation HadoopSeminar Presentation Hadoop
Seminar Presentation Hadoop
 

Similar a Getting started with Hadoop, Hive, and Elastic MapReduce

Hadoop interview questions
Hadoop interview questionsHadoop interview questions
Hadoop interview questionsKalyan Hadoop
 
Hadoop MapReduce Fundamentals
Hadoop MapReduce FundamentalsHadoop MapReduce Fundamentals
Hadoop MapReduce FundamentalsLynn Langit
 
Spark,Hadoop,Presto Comparition
Spark,Hadoop,Presto ComparitionSpark,Hadoop,Presto Comparition
Spark,Hadoop,Presto ComparitionSandish Kumar H N
 
Parallel Computing for Econometricians with Amazon Web Services
Parallel Computing for Econometricians with Amazon Web ServicesParallel Computing for Econometricians with Amazon Web Services
Parallel Computing for Econometricians with Amazon Web Servicesstephenjbarr
 
May 29, 2014 Toronto Hadoop User Group - Micro ETL
May 29, 2014 Toronto Hadoop User Group - Micro ETLMay 29, 2014 Toronto Hadoop User Group - Micro ETL
May 29, 2014 Toronto Hadoop User Group - Micro ETLAdam Muise
 
Hadoop live online training
Hadoop live online trainingHadoop live online training
Hadoop live online trainingHarika583
 
Introduction to Apache Hadoop
Introduction to Apache HadoopIntroduction to Apache Hadoop
Introduction to Apache HadoopSteve Watt
 
Understanding hadoop
Understanding hadoopUnderstanding hadoop
Understanding hadoopRexRamos9
 
Mapreduce by examples
Mapreduce by examplesMapreduce by examples
Mapreduce by examplesAndrea Iacono
 
Hadoop interview question
Hadoop interview questionHadoop interview question
Hadoop interview questionpappupassindia
 
Learn what is Hadoop-and-BigData
Learn  what is Hadoop-and-BigDataLearn  what is Hadoop-and-BigData
Learn what is Hadoop-and-BigDataThanusha154
 
Finding the needles in the haystack. An Overview of Analyzing Big Data with H...
Finding the needles in the haystack. An Overview of Analyzing Big Data with H...Finding the needles in the haystack. An Overview of Analyzing Big Data with H...
Finding the needles in the haystack. An Overview of Analyzing Big Data with H...Chris Baglieri
 
NoSQL, Hadoop, Cascading June 2010
NoSQL, Hadoop, Cascading June 2010NoSQL, Hadoop, Cascading June 2010
NoSQL, Hadoop, Cascading June 2010Christopher Curtin
 
Taylor bosc2010
Taylor bosc2010Taylor bosc2010
Taylor bosc2010BOSC 2010
 
Cascading on starfish
Cascading on starfishCascading on starfish
Cascading on starfishFei Dong
 
Harnessing Hadoop: Understanding the Big Data Processing Options for Optimizi...
Harnessing Hadoop: Understanding the Big Data Processing Options for Optimizi...Harnessing Hadoop: Understanding the Big Data Processing Options for Optimizi...
Harnessing Hadoop: Understanding the Big Data Processing Options for Optimizi...Cognizant
 
Big data vahidamiri-tabriz-13960226-datastack.ir
Big data vahidamiri-tabriz-13960226-datastack.irBig data vahidamiri-tabriz-13960226-datastack.ir
Big data vahidamiri-tabriz-13960226-datastack.irdatastack
 

Similar a Getting started with Hadoop, Hive, and Elastic MapReduce (20)

Hadoop interview questions
Hadoop interview questionsHadoop interview questions
Hadoop interview questions
 
Hadoop MapReduce Fundamentals
Hadoop MapReduce FundamentalsHadoop MapReduce Fundamentals
Hadoop MapReduce Fundamentals
 
Spark,Hadoop,Presto Comparition
Spark,Hadoop,Presto ComparitionSpark,Hadoop,Presto Comparition
Spark,Hadoop,Presto Comparition
 
Parallel Computing for Econometricians with Amazon Web Services
Parallel Computing for Econometricians with Amazon Web ServicesParallel Computing for Econometricians with Amazon Web Services
Parallel Computing for Econometricians with Amazon Web Services
 
May 29, 2014 Toronto Hadoop User Group - Micro ETL
May 29, 2014 Toronto Hadoop User Group - Micro ETLMay 29, 2014 Toronto Hadoop User Group - Micro ETL
May 29, 2014 Toronto Hadoop User Group - Micro ETL
 
Hadoop live online training
Hadoop live online trainingHadoop live online training
Hadoop live online training
 
Introduction to Apache Hadoop
Introduction to Apache HadoopIntroduction to Apache Hadoop
Introduction to Apache Hadoop
 
Understanding hadoop
Understanding hadoopUnderstanding hadoop
Understanding hadoop
 
Mapreduce by examples
Mapreduce by examplesMapreduce by examples
Mapreduce by examples
 
Hadoop interview question
Hadoop interview questionHadoop interview question
Hadoop interview question
 
Learn what is Hadoop-and-BigData
Learn  what is Hadoop-and-BigDataLearn  what is Hadoop-and-BigData
Learn what is Hadoop-and-BigData
 
Hadoop Interview Questions and Answers
Hadoop Interview Questions and AnswersHadoop Interview Questions and Answers
Hadoop Interview Questions and Answers
 
Finding the needles in the haystack. An Overview of Analyzing Big Data with H...
Finding the needles in the haystack. An Overview of Analyzing Big Data with H...Finding the needles in the haystack. An Overview of Analyzing Big Data with H...
Finding the needles in the haystack. An Overview of Analyzing Big Data with H...
 
NoSQL, Hadoop, Cascading June 2010
NoSQL, Hadoop, Cascading June 2010NoSQL, Hadoop, Cascading June 2010
NoSQL, Hadoop, Cascading June 2010
 
Hadoop Seminar Report
Hadoop Seminar ReportHadoop Seminar Report
Hadoop Seminar Report
 
Unit 4 lecture2
Unit 4 lecture2Unit 4 lecture2
Unit 4 lecture2
 
Taylor bosc2010
Taylor bosc2010Taylor bosc2010
Taylor bosc2010
 
Cascading on starfish
Cascading on starfishCascading on starfish
Cascading on starfish
 
Harnessing Hadoop: Understanding the Big Data Processing Options for Optimizi...
Harnessing Hadoop: Understanding the Big Data Processing Options for Optimizi...Harnessing Hadoop: Understanding the Big Data Processing Options for Optimizi...
Harnessing Hadoop: Understanding the Big Data Processing Options for Optimizi...
 
Big data vahidamiri-tabriz-13960226-datastack.ir
Big data vahidamiri-tabriz-13960226-datastack.irBig data vahidamiri-tabriz-13960226-datastack.ir
Big data vahidamiri-tabriz-13960226-datastack.ir
 

Último

Finding Java's Hidden Performance Traps @ DevoxxUK 2024
Finding Java's Hidden Performance Traps @ DevoxxUK 2024Finding Java's Hidden Performance Traps @ DevoxxUK 2024
Finding Java's Hidden Performance Traps @ DevoxxUK 2024Victor Rentea
 
Spring Boot vs Quarkus the ultimate battle - DevoxxUK
Spring Boot vs Quarkus the ultimate battle - DevoxxUKSpring Boot vs Quarkus the ultimate battle - DevoxxUK
Spring Boot vs Quarkus the ultimate battle - DevoxxUKJago de Vreede
 
Ransomware_Q4_2023. The report. [EN].pdf
Ransomware_Q4_2023. The report. [EN].pdfRansomware_Q4_2023. The report. [EN].pdf
Ransomware_Q4_2023. The report. [EN].pdfOverkill Security
 
AWS Community Day CPH - Three problems of Terraform
AWS Community Day CPH - Three problems of TerraformAWS Community Day CPH - Three problems of Terraform
AWS Community Day CPH - Three problems of TerraformAndrey Devyatkin
 
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, AdobeApidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobeapidays
 
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemkeProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemkeProduct Anonymous
 
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...apidays
 
2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...Martijn de Jong
 
Architecting Cloud Native Applications
Architecting Cloud Native ApplicationsArchitecting Cloud Native Applications
Architecting Cloud Native ApplicationsWSO2
 
Axa Assurance Maroc - Insurer Innovation Award 2024
Axa Assurance Maroc - Insurer Innovation Award 2024Axa Assurance Maroc - Insurer Innovation Award 2024
Axa Assurance Maroc - Insurer Innovation Award 2024The Digital Insurer
 
Biography Of Angeliki Cooney | Senior Vice President Life Sciences | Albany, ...
Biography Of Angeliki Cooney | Senior Vice President Life Sciences | Albany, ...Biography Of Angeliki Cooney | Senior Vice President Life Sciences | Albany, ...
Biography Of Angeliki Cooney | Senior Vice President Life Sciences | Albany, ...Angeliki Cooney
 
Navigating the Deluge_ Dubai Floods and the Resilience of Dubai International...
Navigating the Deluge_ Dubai Floods and the Resilience of Dubai International...Navigating the Deluge_ Dubai Floods and the Resilience of Dubai International...
Navigating the Deluge_ Dubai Floods and the Resilience of Dubai International...Orbitshub
 
ICT role in 21st century education and its challenges
ICT role in 21st century education and its challengesICT role in 21st century education and its challenges
ICT role in 21st century education and its challengesrafiqahmad00786416
 
Boost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdfBoost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdfsudhanshuwaghmare1
 
MS Copilot expands with MS Graph connectors
MS Copilot expands with MS Graph connectorsMS Copilot expands with MS Graph connectors
MS Copilot expands with MS Graph connectorsNanddeep Nachan
 
Exploring Multimodal Embeddings with Milvus
Exploring Multimodal Embeddings with MilvusExploring Multimodal Embeddings with Milvus
Exploring Multimodal Embeddings with MilvusZilliz
 
How to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerHow to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerThousandEyes
 
[BuildWithAI] Introduction to Gemini.pdf
[BuildWithAI] Introduction to Gemini.pdf[BuildWithAI] Introduction to Gemini.pdf
[BuildWithAI] Introduction to Gemini.pdfSandro Moreira
 
AXA XL - Insurer Innovation Award Americas 2024
AXA XL - Insurer Innovation Award Americas 2024AXA XL - Insurer Innovation Award Americas 2024
AXA XL - Insurer Innovation Award Americas 2024The Digital Insurer
 
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers:  A Deep Dive into Serverless Spatial Data and FMECloud Frontiers:  A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FMESafe Software
 

Último (20)

Finding Java's Hidden Performance Traps @ DevoxxUK 2024
Finding Java's Hidden Performance Traps @ DevoxxUK 2024Finding Java's Hidden Performance Traps @ DevoxxUK 2024
Finding Java's Hidden Performance Traps @ DevoxxUK 2024
 
Spring Boot vs Quarkus the ultimate battle - DevoxxUK
Spring Boot vs Quarkus the ultimate battle - DevoxxUKSpring Boot vs Quarkus the ultimate battle - DevoxxUK
Spring Boot vs Quarkus the ultimate battle - DevoxxUK
 
Ransomware_Q4_2023. The report. [EN].pdf
Ransomware_Q4_2023. The report. [EN].pdfRansomware_Q4_2023. The report. [EN].pdf
Ransomware_Q4_2023. The report. [EN].pdf
 
AWS Community Day CPH - Three problems of Terraform
AWS Community Day CPH - Three problems of TerraformAWS Community Day CPH - Three problems of Terraform
AWS Community Day CPH - Three problems of Terraform
 
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, AdobeApidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
 
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemkeProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
 
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
 
2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...
 
Architecting Cloud Native Applications
Architecting Cloud Native ApplicationsArchitecting Cloud Native Applications
Architecting Cloud Native Applications
 
Axa Assurance Maroc - Insurer Innovation Award 2024
Axa Assurance Maroc - Insurer Innovation Award 2024Axa Assurance Maroc - Insurer Innovation Award 2024
Axa Assurance Maroc - Insurer Innovation Award 2024
 
Biography Of Angeliki Cooney | Senior Vice President Life Sciences | Albany, ...
Biography Of Angeliki Cooney | Senior Vice President Life Sciences | Albany, ...Biography Of Angeliki Cooney | Senior Vice President Life Sciences | Albany, ...
Biography Of Angeliki Cooney | Senior Vice President Life Sciences | Albany, ...
 
Navigating the Deluge_ Dubai Floods and the Resilience of Dubai International...
Navigating the Deluge_ Dubai Floods and the Resilience of Dubai International...Navigating the Deluge_ Dubai Floods and the Resilience of Dubai International...
Navigating the Deluge_ Dubai Floods and the Resilience of Dubai International...
 
ICT role in 21st century education and its challenges
ICT role in 21st century education and its challengesICT role in 21st century education and its challenges
ICT role in 21st century education and its challenges
 
Boost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdfBoost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdf
 
MS Copilot expands with MS Graph connectors
MS Copilot expands with MS Graph connectorsMS Copilot expands with MS Graph connectors
MS Copilot expands with MS Graph connectors
 
Exploring Multimodal Embeddings with Milvus
Exploring Multimodal Embeddings with MilvusExploring Multimodal Embeddings with Milvus
Exploring Multimodal Embeddings with Milvus
 
How to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerHow to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected Worker
 
[BuildWithAI] Introduction to Gemini.pdf
[BuildWithAI] Introduction to Gemini.pdf[BuildWithAI] Introduction to Gemini.pdf
[BuildWithAI] Introduction to Gemini.pdf
 
AXA XL - Insurer Innovation Award Americas 2024
AXA XL - Insurer Innovation Award Americas 2024AXA XL - Insurer Innovation Award Americas 2024
AXA XL - Insurer Innovation Award Americas 2024
 
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers:  A Deep Dive into Serverless Spatial Data and FMECloud Frontiers:  A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
 

Getting started with Hadoop, Hive, and Elastic MapReduce

  • 1. Getting started with Hadoop, Hive, and Elastic MapReduce Hunter Blanks, hblanks@monetate.com / @tildehblanks github.com/hblanks/talks/ PhillyAWS. January 12, 2012
  • 2. Overview In the figure below, 4 inputs are mapped to 16 (key, value) tuples. These tuples are sorted by those keys, and then reduced, into 4 new (key, value) tuples. Finally, these new tuples are once again being sorted and reduced, and so on... source: http://code.google.com/p/mapreduce-framework/wiki/MapReduce
  • 3. Timeline 2004 2008 2009 2010 2011 MapReduce paper Doug Cutting Hadoop wins Elastic MapReduce mrjob published (and other terabyte sort launched. open Yahoos) start benchmark Facebook presents sourced work on Hadoop first paper on Hive.
  • 4. An overview of available (and not available) tools 1) MapReduce. Google's C++ implementation, which sits atop GFS, chubby, BigTable, and who knows what else. See also sawzall. 2) Hadoop. Apache's/Facebook's/Yahoo's Java implementation, which furthermore includes, or else integrates with: HDFS (a reimplementation of GFS), Zookeeper (a reimplementation of chubby), and HBase (a reimplementation of BigTable). See also Pig (not exactly sawzall, but another take on it). 3) Hive. Apache's/Facebook's data warehousing system, which allows users to query large datasets using SQL-like syntax, over Hadoop. 4) Elastic MapReduce. Amazon's API for spinning up temporary Hadoop clusters (aka "Jobflows"), typically reading input and writing output to S3. It is much, much easier than bringing up your own Hadoop cluster. 5) mrjob. Yelp's Python framework for running MapReduce jobs on Hadoop clusters, with or without ElasticMapReduce.
  • 5. Terms for Elastic MapReduce 1) job flow. A transient (or semi-transient) Hadoop cluster. Jobflows typically terminate after all their steps are done, but they can also be flagged as "permanent," which just means they don't shut down on their own. 2) master, core, and task nodes. A jobflow has one master node; when it dies, your jobflow dies (or is at least supposed to). It has a fixed number of core nodes (determined when the jobflow was started) -- these core nodes can do map/reduce tasks, and also serve HDFS. It can also have a changeable number of task nodes, which do tasks but don't serve HDFS. 3) step. A jobflow contains one or more "steps". These may or may not be actual Hadoop "jobs" -- sometimes a step is just for initialization, other times it is an actual map/reduce job. 4) spot instance vs on-demand. By default, any job flow you start will use normal, "on-demand" EC2 instances, with a fixed hourly price. You may alternatively specify a spot instances by naming a maximum price (a bid price) you're willing to pay. If your bid price is below the current spot price, then you will pay the spot price for your instances (typically 50-66% cheaper than on- demand instances).
  • 6. The easiest ways to get started with Elastic MapReduce, #1 If you know Python, start with mrjob. Here's a sample MRJob class, which you might write to wordcounter.py: from mrjob.job import MRJob class MRWordCounter(MRJob): def mapper(self, key, line): for word in line.split(): yield word, 1 def reducer(self, word, occurrences): yield word, sum(occurrences) if __name__ == '__main__': MRWordCounter.run() And how you run it: export AWS_ACCESS_KEY_ID=...; export AWS_SECRET_ACCESS_KEY=... python wordcounter.py -r emr < input > output These samples, and much more documentation, is at http:// packages.python.org/mrjob/writing-and-running.html.
  • 7. The easiest ways to get started with Elastic MapReduce, #2 If you know any other scripting language, start with the AWS elastic-mapreduce command line tool, or the AWS web console. Both are outlined at: http://docs.amazonwebservices.com/ElasticMapReduce/latest/GettingStartedGuide/ An example of the command line, assuming a file wordSplitter.py, is: ./elastic-mapreduce --create --stream --mapper s3://elasticmapreduce/samples/wordcount/wordSplitter.py --input s3://elasticmapreduce/samples/wordcount/input --output s3://mybucket.foo.com/wordcount --reducer aggregate A (not so) quick reference card for the command line tool is at: http://s3.amazonaws.com/awsdocs/ElasticMapReduce/latest/emr-qrc.pdf
  • 8. Getting started with Hive is not so easy 1) You're going to need to put your data into some sort of warehouse, probably in S3. (HDFS will not persist in the cloud). This may involve its own MapReduce step, but with careful attention to how you output your files to S3. Partitioning matters, but you won't get it with just the default Hadoop streaming jar! (We use oddjob's) 2) To start a jobflow with Hive programmatically (i.e., not using the command line elastic- mapreduce client), you need to add several initial steps to the jobflow -- notably ones that start Hive and possibly configure its heapsize. Fortunately, Amazon's elastic- mapreduce client is easy enough to reverse engineer... 3) You're going to need to make a schema (surprise) for your data warehouse. 4) You may have warehoused data in JSON. Hive much prefers flat, tabular files. 5) To talk to Hive programmatically from other servers, you'll need to use Hive's thrift binding, and also open up the ElasticMapReduce-master security group (yes, EMR created this for you when you first ran a jobflow) so that your application box can talk to the master node of your job flow. 6) All that said, it may beat writing your own custom MapReduce jobs.
  • 9. Other pitfalls to beware 1) The JVM is not your friend. It especially loves memory. - be prepared to add swap on your task nodes, at least until Hadoop's fork()/exec() behavior is fixed in AWS' version of Hadoop - be prepared to profile your cluster with Ganglia (it's really, really easy to setup). Just add this bootstrap action to do so: s3://elasticmapreduce/bootstrap-actions/install-ganglia 2) Multiple output files can get you into trouble, especially if you don't sort your outputs first. 3) Hadoop, and Hive as well, much prefer larger files to lots of small files.
  • 10. Further reading - "MapReduce: Simplified Data Processing on Large Clusters" OSDI'04. http://research.google.com/archive/mapreduce.html - "Hive - A Warehousing Solution Over a Map-Reduce Framework" VLDB, 2009. http://www.vldb.org/pvldb/2/vldb09-938.pdf - Hadoop: The Definitive Guide. O'Reilly Associates, 2009. Thank you! (P.S. We're hiring. hblanks@monetate.com)