Frankie Pike. 2010: 1.2 zettabytes 1.2 trillion gigabytes DVDs past the moon 2-way = 6 newspapers everyday ~58% growth per year Why care?

MapReduce and HadoopFrankie Pike

2010: 1.2 zettabytes1.2 trillion gigabytesDVDs past the moon2-way = 6 newspapers everyday~58% growth per year

Why care?

Why care?

Why care?

Google’s capacity = 1 exabyte 24 hours of Youtube > Internet in 2000 4 years of video / day on Youtube 100 trillion words online

Common Architecture

http://www.adopenstatic.com/images/resources/blog/Kerberos6.jpg

Common Architecture

Single point of failureSpace-constraintsMulti-tenancy difficultiesRe-writing of programs or

changes to network config

MapReduce

The PromiseHigh reliability

any node can go downHigh scalability

easy to add nodesMulti-tenancyCost Reduction“Cloud-friendly”Java, C++, C#, Python, RTransparent Parallelization

The Kryptonite

Data set needs to be

“big enough”

Consistency mid-processing

Two Steps in MapReduce

MapReduce

Mapping

Input K/V pairs -> Intermediate K/V PairsInput and Intermediate can be different(Server Key, Blog Data) -> (Blog Key, Post Count)

Sorted and Partitioned for reduction

Number of maps depends on task and cluster10TB data with blocksize 128MB = 82,000 maps10-100 maps per node ideal

ReducingIntermediate K/V -> Intermediate K/V (smaller)

Matching keys consolidated(A, 15); (B, 6); (A, 3) -> (A, 18); (B, 6)

Number of Reductions >= 0Hopefully smaller dataset at each iterationReduce as much as needed

An Example{

"type": "post", "name": "Raven's Map/Reduce functionality", "blog_id": 1342, "post_id": 29293921, "tags": ["raven", "nosql"], "post_content": "<p>...</p>", "comments": [

{ "source_ip": '124.2.21.2', "author": "martin", "text": "..." }

] } Want count of comments for blog

http://ayende.com/blog/4435/map-reduce-a-visual-explanation

Step 1: Map to final format

http://ayende.com/blog/4435/map-reduce-a-visual-explanation

Step 2: Reduce (Partition)

http://ayende.com/blog/4435/map-reduce-a-visual-explanation

Step 3: Reduce (more)

http://ayende.com/blog/4435/map-reduce-a-visual-explanation

Step 4: Reduce (most)

http://ayende.com/blog/4435/map-reduce-a-visual-explanation

Single Node

http://bc.tech.coop/blog/070520.html

Dual Node

http://map-reduce.wikispaces.asu.edu/

N-Nodes

http://www.inventoland.net/img/blog/mapReduce.png

Dealing with Failure

WorkersOccasional check-in pings by masters

MastersData structures get periodic auto-saves and consistency

checks. Can restart from periodic savesBandwidth

Tasks attempt to pair with local storage

Has it worked?

PatentedRegenerated index

Apache Hadoop“open source software for reliable, scalable, distributed

computing”

Hadoop Distributed File System (HDFS)Hadoop MapReduceCassandra (multi-master database)HBase (scalable, distributed, structured database)Mahout (data mining and machine learning libs)ZooKeeper (coordination service)

Sources

Avankipu & Sdsalvi, Cloud Computing - An Overview. http://map-reduce.wikispaces.asu.edu

Ayende Rahien, Map/Reduce – A Visual Explanation. http://ayende.com/blog/4435/map-reduce-a-visual-explanation

http://hadoop.apache.org/http://en.wikipedia.org/wiki/MapReduce/