• Apache Hadoop is an open source software framework offering reusable components for Big Data work.  It is written in Java and I will go into some more detail about it later, but I'm bringing it up now so you won't be confused about what it is when I reference it.
• Fun fact:  It is named for a toy elephant!

• "Big" data sizes are always a moving target
• In the Hadoop documentation they cite 200-300GB as being on the small end of the scale
• Facebook users generate ~ 500 TB of data per day! (http://gigaom.com/2012/08/22/facebook-is-collecting-your-data-500-terabytes-a-day/)
• That's around 4,369,066 times the size of my MacBook Air hard drive (120 GB)
• 
  

• An individual computer usually only holds about 2GB - 16GB in memory
• Therefore, we have to keep the data on disk and read it into memory in much smaller chunks
• Even a large computer hard drive rarely exceeds 1TB
• Therefore, we have to be able to synchronize multiple hard drives and computers working on this data
• This is a hard problem

• Networks experience total or partial failure as switches and routers go down.
• Data doesn't arrive at expected time due to network congestion.
• Individual nodes crash, run out of disk space, overheat.
• And so on...

• Moore's Law: transistor density will double ~ every 2 years
• But nowadays instead of speeding up single core processors we're cramming more cores on one chip.  So, we're living in a parallel world anyway.
• Besides, hard drive read/write speeds are the bottleneck.

• Hadoop provides a distributed filesystem to address some of these concerns
• It allows for redundancy as well as splitting up a huge data workload over a cluster of many machines

• The data is broken up into blocks over the machines in the network (referred to as DataNodes).  Each block is 64MB by default.
• Each file on the filesystem is made up of several blocks, which aren't necessarily on the same machine.  Access to a single file may require access to multiple machines the size of storable files is much larger than what is capable on single machines.
• Copies of the same are block stored on multiple machines (3 by default) to account for possibility of DataNode failure
• Metadata about these files (which blocks are where, for instance) is stored in a master node called the NameNode (it's very bad if the NameNode crashes, but it's less likely)

• Say we have a huge work of literature and we want to count how many times each word is used, using Map/Reduce so we can easily parallelize the task.
• We'll do the operation manually to get the hang of it.
• Let's use this example quote:

"Some are born great, some achieve greatness, and some have greatness thrust upon them."

- William Shakespeare

("some", 1)
("are", 1)
("born", 1)
("great", 1)
("some", 1)
("achieve", 1)
("greatness", 1)
("and", 1)
("some", 1)
("have", 1)
("greatness", 1)("thrust", 1)
("upon", 1)
("them", 1)

("achieve", 1)("and", 1)
("are", 1)
("born", 1)
("great", 1)
("greatness", 1)
("greatness", 1)
("have", 1)
("some", 1)
("some", 1)
("some", 1)
("them", 1)
("thrust", 1)
("upon", 1)

("achieve", 1)("and", 1)
("are", 1)
("born", 1)
("great", 1)
("greatness", 2)
("have", 1)
("some", 3)
("them", 1)
("thrust", 1)
("upon", 1)

import java.io.IOException;
import java.util.StringTokenizer;

import org.apache.hadoop.io.IntWritable;
import org.apache.hadoop.io.LongWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.io.Writable;
import org.apache.hadoop.io.WritableComparable;
import org.apache.hadoop.mapred.MapReduceBase;
import org.apache.hadoop.mapred.Mapper;
import org.apache.hadoop.mapred.OutputCollector;
import org.apache.hadoop.mapred.Reporter;

public class WordCountMapper extends MapReduceBase
    implements Mapper<LongWritable, Text, Text, IntWritable> {

  private final IntWritable one = new IntWritable(1);
  private Text word = new Text();

  public void map(WritableComparable key, Writable value,
      OutputCollector output, Reporter reporter) throws IOException {

    String line = value.toString();
    StringTokenizer itr = new StringTokenizer(line.toLowerCase());
    while(itr.hasMoreTokens()) {
      word.set(itr.nextToken());
      output.collect(word, one);
    }
  }
}

import java.io.IOException;
import java.util.Iterator;

import org.apache.hadoop.io.IntWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.io.WritableComparable;
import org.apache.hadoop.mapred.MapReduceBase;
import org.apache.hadoop.mapred.OutputCollector;
import org.apache.hadoop.mapred.Reducer;
import org.apache.hadoop.mapred.Reporter;

public class WordCountReducer extends MapReduceBase
    implements Reducer<Text, IntWritable, Text, IntWritable> {

  public void reduce(Text key, Iterator values,
      OutputCollector output, Reporter reporter) throws IOException {

    int sum = 0;
    while (values.hasNext()) {
      IntWritable value = (IntWritable) values.next();
      sum += value.get(); // process value
    }

    output.collect(key, new IntWritable(sum));
  }
}

import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.IntWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapred.FileInputFormat;
import org.apache.hadoop.mapred.FileOutputFormat;
import org.apache.hadoop.mapred.JobClient;
import org.apache.hadoop.mapred.JobConf;

public class WordCount {

  public static void main(String[] args) {
    JobClient client = new JobClient();
    JobConf conf = new JobConf(WordCount.class);

    // specify output types
    conf.setOutputKeyClass(Text.class);
    conf.setOutputValueClass(IntWritable.class);

    // specify input and output dirs
    FileInputPath.addInputPath(conf, new Path("input"));
    FileOutputPath.addOutputPath(conf, new Path("output"));

    // specify a mapper
    conf.setMapperClass(WordCountMapper.class);

    // specify a reducer
    conf.setReducerClass(WordCountReducer.class);
    conf.setCombinerClass(WordCountReducer.class);

    client.setConf(conf);
    try {
      JobClient.runJob(conf);
    } catch (Exception e) {
      e.printStackTrace();
    }
  }
}

• Criticism:  Map/Reduce is not novel and is too low-level
  
• Also:  Is your data truly "big" ?
• Didn't even cover HBase, Cassandra, Apache Pig, Storm, Hive etc. which are cool things to look into if you are interested in Big Data.
  
• Any questions?