A MapReduce job usually splits the input data-set into independent chunks which are processed by the map tasks in a completely parallel manner. The framework sorts the outputs of the maps, which are then input to the reduce tasks. Typically both the input and the output of the job are stored in a file-system.

1. What is Hadoop?
• The Apache Hadoop project develops open-
source software for reliable, scalable,
distributed computing.
In a nutshell
• Hadoop provides: a reliable shared storage and
analysis system.
• The storage is provided by HDFS
• The analysis by MapReduce.

2. Map Reduce
• HDFS handles the Distributed Filesystem layer
• MapReduce is a programming model for data processing.
• MapReduce
– Framework for parallel computing
– Programmers get simple API
– Don’t have to worry about handling
• parallelization
• data distribution
• load balancing
• fault tolerance
• Allows one to process huge amounts of data
(terabytes and petabytes) on thousands of processors

3. Map Reduce Concepts (Hadoop-
1.0)
Data Node
Task
Tracker
Data Node
Task
Tracker
Data Node
Task
Tracker
Data Node
Task
Tracker
MapReduce
Engine
HDFS
Cluster
Job Tracker
Admin Node
Name node

4. Map Reduce Concepts
 Job Tracker
The Job-Tracker is responsible for accepting jobs
from clients, dividing those jobs into tasks, and
assigning those tasks to be executed by worker
nodes.
 Task Tracker
Task-Tracker is a process that manages the
execution of the tasks currently assigned to that
node. Each Task Tracker has a fixed number of
slots for executing tasks (two maps and two
reduces by default).

5. Job Tracker
DFS
Job Tracker
1. Copy Input Files
User
2. Submit Job
3. Get Input Files’ Info
6. Submit Job
4. Create Splits
5. Upload Job
Information
Input Files
Client
Job.xml.
Job.jar.

6. Job Tracker
DFS
Client
Job.xml.
Job.jar.
6. Submit Job
8. Read Job Files
7. Initialize Job
Job Queue
As many maps
as splitsInput Spilts
Maps Reduces
9. Create
maps and
reduces
Job Tracker

7. Job Tracker
Job Tracker
Job QueueH3
H4
H5
H1
Task Tracker -
H2
Task Tracker -
H4
10. Heartbeat
12. Assign Tasks
10. Heartbeat
10. Heartbeat
10. Heartbeat
11. Picks Tasks
(Data Local if possible)
Task Tracker -
H3
Task Tracker -
H1

8. Understanding Data Transformations
• In order to write MapReduce applications you need to have an understanding of
how data is transformed as it executes in the MapReduce framework.
From start to finish, there are four fundamental transformations. Data is:
• Transformed from the input files and fed into the mappers
• Transformed by the mappers
• Sorted, merged, and presented to the reducer
• Transform by reducers and written to output files

9. Solving a Programming Problem using
MapReduce
There are a total of 10 fields of information in each line. Our
programming objective uses only the first and fourth fields, which
are arbitrarily called "year" and "delta" respectively. We will ignore
all the other fields of data.

10. Designing and Implementing the
Mapper Class

13. Designing and Implementing the
Reducer Class

16. Design and Implement The Driver

18. Introduction to MapReduce
Framework
 A programming model for parallel data processing.
 Hadoop can run map reduce programs in multiple
languages like Java, Python, Ruby etc.
 Map function:
 Operate on set of key, value pairs
 Map is applied in parallel on input data set
 This produces output keys and list of values for each key
depending upon the functionality
 Mapper output are partitioned per reducer
 Reduce function:
 Operate on set of key, value pairs
 Reduce is then applied in parallel to each group, again
producing a collection of key, values.
 Total number of reducers can be set by the user.

19. Skeleton of a MapReduce Program
public class WordCount {
public static class TokenizerMapper
extends Mapper<Object, Text, Text, IntWritable>{
private final static IntWritable one = new IntWritable(1);
private Text word = new Text();
public void map(Object key, Text value, Context context)
throws IOException, InterruptedException {
StringTokenizer itr = new StringTokenizer(value.toString());
while (itr.hasMoreTokens()) {
word.set(itr.nextToken());
context.write(word, one);
}
}
}

20. Skeleton of a MapReduce Program
public static class IntSumReducer
extends Reducer<Text,IntWritable,Text,IntWritable> {
private IntWritable result = new IntWritable();
public void reduce(Text key, Iterable<IntWritable> values,Context context)
throws IOException, InterruptedException {
int sum = 0;
for (IntWritable val : values) {
sum += val.get();
}
result.set(sum);
context.write(key, result);
}
}

21. Skeleton of a MapReduce program
public static void main(String[] args) throws Exception {
Configuration conf = new Configuration();
Job job = Job.getInstance(conf, "word count");
job.setJarByClass(WordCount.class);
job.setMapperClass(TokenizerMapper.class);
job.setReducerClass(IntSumReducer.class);
job.setOutputKeyClass(Text.class);
job.setOutputValueClass(IntWritable.class);
FileInputFormat.addInputPath(job, new Path(args[0]));
FileOutputFormat.setOutputPath(job, new Path(args[1]));
FileSystem.get(conf).delete(new Path(args[1]), true);
job.waitForCompletion(true);
}

22. Executing MR Job in Java
1)Compile all the 3 java files which will create 3
.class files
2)Add all 3 .class files into 1 single jar file by
writing this command
jar –cvf file_name.jar *.class
3)Now you just need to execute single jar file
by writing this command
bin/hadoop jar file_name.jar Basic
input_file_name output_file_name

23. Overall MR Word Count Process

24. Understanding processing in a
MapReduce framework
User runs a program on the client computer
Program submits a job to HDFS.
Job contains:
 Input data
 Map / Reduce program
 Configuration information
Two types of daemons that control job
execution:
 Job Tracker (master node)
 Task Trackers (slave nodes)

25. Understanding processing in a
MapReduce framework
Job sent to JobTracker.
JobTracker communicates with NameNode
and assigns parts of job to TaskTrackers
 Task is a single MAP or REDUCE
operation over piece of data.
 The JobTracker knows (from NameNode)
which node contains the data, and which
other machines are nearby.
Task processes send heartbeats to
TaskTracker
 TaskTracker sends heartbeats to the
JobTracker.

26. Understanding processing in a
MapReduce framework
Any tasks that did not report in certain time
(default is 10 min) assumed to be failed and
it’s JVM will be killed by TaskTracker and
reported to the JobTracker
The JobTracker will reschedule any failed
tasks (with different TaskTracker)
If same task failed 4 times all job fails
Any TaskTracker reporting high number of
failed jobs on particular node will be blacklist
the node (remove metadata from NameNode)
JobTracker maintains and manages the status
of each job. Results from failed tasks will be
ignored

27. MapReduce Job Submission Flow
Input data is distributed to nodes
Node 1 Node 2
INPUT DATA
Node 1 Node 2

28. MapReduce Job Submission Flow
Input data is distributed to nodes
Each map task works on a “split” of data
Map
Node 1
Map
Node 2
INPUT DATA
Node 1 Node 2

29. MapReduce Job Submission Flow
Input data is distributed to nodes
Each map task works on a “split” of data
Mapper outputs intermediate data
Map
Node 1
Map
Node 2
INPUT DATA
Node 1 Node 2

30. MapReduce Job Submission Flow
Input data is distributed to nodes
Each map task works on a “split” of data
Mapper outputs intermediate data
Data exchange between nodes in a “shuffle” process
Map
Node 1
Map
Node 2
INPUT DATA
Node 1 Node 2

31. MapReduce Job Submission Flow
Input data is distributed to nodes
Each map task works on a “split” of data
Mapper outputs intermediate data
Data exchange between nodes in a “shuffle” process
Intermediate data of the same key goes to the same reducer
Map
Node 1
Map
Node 2
Reduce Reduce
INPUT DATA
Node 1 Node 2

32. MapReduce Job Submission Flow
Input data is distributed to nodes
Each map task works on a “split” of data
Mapper outputs intermediate data
Data exchange between nodes in a “shuffle” process
Intermediate data of the same key goes to the same reducer
Reducer output is stored
Map
Node 1
Map
Node 2
Reduce Reduce
INPUT DATA
Node 1 Node 2

33. MapReduce Flow - Mapper
InputSplit InputSplit InputSplit
Input File Input File
InputSplit InputSplit
RecordReader RecordReader RecordReader RecordReader RecordReader
Mapper
Intermediates
Mapper
Intermediates
Mapper
Intermediates
Mapper
Intermediates
Mapper
Intermediates
InputFormat

34. MapReduce Flow – Shuffle and Sort
Mapper Mapper Mapper Mapper Mapper
Partitioner Partitioner Partitioner Partitioner Partitioner
Intermediates Intermediates Intermediates Intermediates Intermediates
Reducer Reducer Reducer
Intermediates Intermediates Intermediates

35. MapReduce Flow - Reducer
Reducer Reducer Reduce
Output File
RecordWriter
OutputFormat
Output File
RecordWriter
Output File
RecordWriter

36. Map Reduce – Again Closure look
file
file
InputFormat
Split Split Split
RR RR RR
Map Map Map
Input (K, V) pairs
Partitioner
Intermediate (K, V) pairs
Sort
Reduce
OutputFormat
Files loaded from local HDFS store
RecordReaders
Final (K, V) pairs
Writeback to local
HDFS store
file
file
InputFormat
Split Split Split
RR RR RR
Map Map Map
Input (K, V) pairs
Partitioner
Intermediate (K, V) pairs
Sort
Reduce
OutputFormat
Files loaded from local HDFS store
RecordReaders
Final (K, V) pairs
Writeback to local
HDFS store
Node 1 Node 2
Shuffling
Process
Intermediate
(K,V) pairs
exchanged by
all nodes

37. MapReduce API - Overview

38. MapReduce API Data Types: Writable

39. MapReduce - Input Format
 How the input files are split up and read is defined by
the InputFormat
 InputFormat is a class that does the following:
 Selects the files that should be used for input
 Defines the InputSplits that break a file
 Provides a factory for RecordReader objects that
read the file

40. MapReduce API - InputFormats

41. Input Splits
 An input split describes a unit of work that comprises
a single map task in a MapReduce program
 By default, the InputFormat breaks a file upto 64MB
splits
 By dividing the file into splits, we allow several map
tasks to operate on a single file in parallel
 If the file is very large, this can improve performance
significantly through parallelism
 Each map task corresponds to a single input split

42. RecordReader
 The input split defines a slice of work but does
not describe how to access it
 The RecordReader class actually loads data from
its source and converts it into (K, V) pairs suitable
for reading by Mappers
 The RecordReader is invoked repeatedly on the
input until the entire split is consumed
 Each invocation of the RecordReader leads to
another call of the map function defined by the
programmer

43. Mapper and Reducer
 The Mapper performs the user-defined work of
the first phase of the MapReduce program.
 A new instance of Mapper is created for each
split.
 The Reducer performs the user-defined work of
the second phase of the MapReduce program.
 A new instance of Reducer is created for each
partition.
 For each key in the partition assigned to a
Reducer, the Reducer is called once.

44. Combiner
• Apply reduce function to map output before it is sent to reducer
• Reduces number of records outputted by mapper!

45. Partitioner
 Each mapper may produce (K, V) pairs to any
partition.
 Therefore, the map nodes must all agree on
where to send different pieces of intermediate
data.
 The partitioner class determines which
partition a given (K,V) pair will go to.
The default partitioner computes a hash value for a
given key and assigns it to a partition based on
this result.

46. MapReduce Execution – Single
Reduced Task

47. MapReduce Execution – Multiple
Reduce tasks

48. MapReduce Execution – With No
Reduce Tasks

49. Shuffle and Sort
Mapper
Reducer
other mappers
other reducers
circular buffer
(in memory)
spills (on disk)
merged spills
(on disk)
intermediate files
(on disk)
Combiner
Combiner

50. Shuffle and Sort
• Probably the most complex aspect of MapReduce and heart of
the map reduce!
• Map side
 Map outputs are buffered in memory in a circular buffer.
 When buffer reaches threshold, contents are “spilled” to
disk.
 Spills merged in a single, partitioned file (sorted within each
partition): combiner runs here first.
• Reduce side
 First, map outputs are copied over to reducer machine.
 “Sort” is a multi-pass merge of map outputs (happens in
memory and on disk): combiner runs here again.
 Final merge pass goes directly into reducer.

51. Output Format
 The OutputFormat class defines the way (K,V) pairs
produced by Reducers are written to output files
 The instances of OutputFormat provided by Hadoop
write to files on the local disk or in HDFS
 Several OutputFormats are provided by Hadoop:
TextOutputFormat - Default; writes lines in "key t
value" format
SequenceFileOutputFormat - Writes binary files
suitable for reading into subsequent MR jobs
NullOutputFormat - Generates no output files

52. Job Scheduling in MapReduce
Job Queue

53. Job Scheduling in MapReduce
Job Queue

54. Job Scheduling in MapReduce
Job Queue

55. Job Scheduling in MapReduce
Job Queue

56. Job Scheduling in MapReduce
Job Queue

57. Fault Tolerance
 MapReduce can guide jobs toward a successful completion even
when jobs are run on a large cluster where probability of failures
increases
 The primary way that MapReduce achieves fault tolerance is
through restarting tasks
 If a TT fails to communicate with Application Manager for a
period of time (by default, 1 minute in Hadoop), JT will assume
that TT in question has crashed
 If the job is still in the map phase, JT asks another TT to re-
execute all Mappers that previously ran at the failed TT
 If the job is in the reduce phase, Application Manager asks
another TT to re-execute all Reducers that were in progress
on the failed TT

58. Speculative Execution
 A MapReduce job is dominated by the slowest task
 MapReduce attempts to locate slow tasks (stragglers)
and run redundant (speculative) tasks that will
optimistically commit before the corresponding
stragglers
 This process is known as speculative execution
 Only one copy of a straggler is allowed to be
speculated
 Whichever copy (among the two copies) of a task
commits first, it becomes the definitive copy, and the
other copy is killed by JT

59. Locating Stragglers
 How does Hadoop locate stragglers?
 Hadoop monitors each task progress using a
progress score between 0 and 1
 If a task’s progress score is less than (average –
0.2), and the task has run for at least 1 minute, it
is marked as a straggler
PS= 2/3
PS= 1/12
 Not a stragglerT1
T2
Time
A straggler

60. MapReduce Execution - One Picture

61. Data Flow in a MapReduce Program
• InputFormat
• Map function
• Partitioner
• Sorting & Merging
• Combiner
• Shuffling
• Merging
• Reduce function
• OutputFormat
 1:many

62. Counters
There are often things that you would like to know about the data you are analyzing
but that are peripheral to the analysis you are performing. Counters are a useful
channel for gathering statistics about the job: for quality control or for application-
level statistics.
Built-in Counters
Hadoop maintains some built-in counters for every job, and these report various
metrics. For example, there are counters for the number of bytes and records
processed, which allow you to confirm that the expected amount of input was
consumed and the expected amount of output was produced.
Counters are divided into groups, and there are several groups for the built-in
counters,
• MapReduce task counters
• Filesystem counters
• FileInputFormat counters
• FileOutputFormat counters
Each group either contains task counters (which are updated as a task progresses) or
job counters (which are updated as a job progresses).

63. Counters
Task counters
Task counters gather information about tasks over the course of their execution, and
the results are aggregated over all the tasks in a job.
The MAP_INPUT_RECORDS counter, for example, counts the input records read by
each map task and aggregates over all map tasks in a job, so that the final figure is the
total number of input records for the whole job.
Task counters are maintained by each task attempt, and periodically sent to the
application master so they can be globally aggregated.
Job counters
Job counters are maintained by the application master, so they don’t need to be sent
across the network, unlike all other counters, including user-defined ones. They
measure job-level statistics, not values that change while a task is running. For
example, TOTAL_LAUNCHED_MAPS counts the number of map tasks that were
launched over the course of a job (including tasks that failed).
User-Defined Java Counters
MapReduce allows user code to define a set of counters, which are then incremented
as desired in the mapper or reducer. Counters are defined by a Java enum.

64. JOINS
MAP-JOIN
• A map-side join between large inputs works by performing the join
before the data reaches the map function. For this to work, though,
the inputs to each map must be partitioned and sorted in a
particular way. Each input dataset must be divided into the same
number of partitions, and it must be sorted by the same key (the
join key) in each source. All the records for a particular key must
reside in the same partition. This may sound like a strict
requirement (and it is), but it actually fits the description of the
output of a MapReduce job.
Distributed Cache
• It is preferable to distribute datasets using Hadoop’s distributed
cache mechanism which provides a service for copying files to the
task nodes for the tasks to use them when they run. To save
network bandwidth, files are normally copied to any particular
node once per job.

65. JOINS
Reducer side –JOIN
• Reduce-side join is more general than a map-side
join, in that the input datasets don’t have to be
structured in any particular way, but it is less
efficient because both datasets have to go
through the MapReduce shuffle. The basic idea is
that the mapper tags each record with its source
and uses the join key as the map output key, so
that the records with the same key are brought
together in the reducer.

66. Secondary Sorting
• The MapReduce framework sorts the records by key
before they reach the reducers. For any particular key,
however, the values are not sorted. The order in which
the values appear is not even stable from one run to
the next, because they come from different map tasks,
which may finish at different times from run to run.
Generally, most MapReduce programs are written so as
not to depend on the order in which the values appear
to the reduce function. However, it is possible to
impose an order on the values by sorting and grouping
the keys in a particular way.

67. ToolRunner
public int run(String[] args) throws Exception
{
Configuration conf = new Configuration();
Job job = new Job(conf);
job.setJarByClass(multiInputFile.class);
…..
……
……
FileOutputFormat.setOutputPath(job, new Path(args[2]));
FileSystem.get(conf).delete(new Path(args[2]), true);
return (job.waitForCompletion(true) ? 0 : 1);
}
public static void main(String[] args) throws Exception
{
int ecode = ToolRunner.run(new multiInputFile(), args);
System.exit(ecode);
}

68. Practice Session

69. Thank You
• Question?
• Feedback?
explorehadoop@gmail.com