0% found this document useful (0 votes)
11 views35 pages

Map Reduce Algorithm

The document discusses the principles of Big Data and MapReduce, highlighting its role in large-scale data processing through automatic parallelization, fault-tolerance, and I/O scheduling. It explains the architecture of commodity clusters, the Google File System, and provides examples of using MapReduce for tasks like word counting and graph reversal. Additionally, it introduces Hadoop as an open-source implementation of MapReduce, emphasizing its use of HDFS for stable storage.

Uploaded by

mahima78t
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
11 views35 pages

Map Reduce Algorithm

The document discusses the principles of Big Data and MapReduce, highlighting its role in large-scale data processing through automatic parallelization, fault-tolerance, and I/O scheduling. It explains the architecture of commodity clusters, the Google File System, and provides examples of using MapReduce for tasks like word counting and graph reversal. Additionally, it introduces Hadoop as an open-source implementation of MapReduce, emphasizing its use of HDFS for stable storage.

Uploaded by

mahima78t
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

CS 789 ADVANCED BIG DATA

ANALYTICS

BIG DATA AND MAP REDUCE

* The contents are adapted from Dr. Chengkai Li at UT Arlington, Jeff Dean, Sanjay
Ghemawat, Anand Rajaraman

Mingon Kang, Ph.D.


Department of Computer Science, University of Nevada, Las Vegas
Map? Reduce?
Higher-order function in functional programming
languages.

Example: Scheme (variant of LISP)

 (map square ‘(1 2 3))


 (1 4 9)
 (reduce + (map square ‘(1 2 3)))
 14
Motivation: Large Scale Data Processing

 Many tasks:
Process lots of data to produce other data
 Want to use hundreds or thousands of CPUs
... but this needs to be easy
 MapReduce provides:
 Automatic parallelization and distribution
 Fault-tolerance

 I/O scheduling

 Status and monitoring


Single-node architecture

CPU
Machine Learning, Statistics

Memory

“Classical” Data Mining

Disk
Commodity Clusters
 Web data sets can be very large
 Tens to hundreds of terabytes
 Cannot mine on a single server (why?)
 Standard architecture emerging:
 Cluster
of commodity Linux nodes
 Gigabit ethernet interconnect

 How to organize computations on this architecture?


 Mask issues such as hardware failure
Cluster Architecture
2-10 Gbps backbone between racks
1 Gbps between Switch
any pair of nodes
in a rack
Switch Switch

CPU CPU CPU CPU

Mem … Mem Mem … Mem

Disk Disk Disk Disk

Each rack contains 16-64 nodes


Stable storage
 First order problem: if nodes can fail, how can we
store data persistently?
 Answer: Distributed File System
 Provides
global file namespace
 Google GFS; Hadoop HDFS; Kosmix KFS

 Typical usage pattern


 Huge files (100s of GB to TB)
 Data is rarely updated in place

 Reads and appends are common


Google File System
 Distribute File System
 Master: control tower that monitors GFS’s status and manages
 Chunk Server: physical I/O operations
 Client: request I/O operations
Google File System
 A client requests I/O operations
 Master replies the information of the chunk server which is nearest to the
client
 Client communicates with the chunk server directly for I/O operations
Google File System
 Fault-tolerance
 If a chunk server fails
◼ Master uses other available chunk server
 If master server fails
◼ Thereis another device that monitors master server
◼ Master will be replaced with others
Warm up: Word Count
 We have a large file of words, one word to a line
 Count the number of times each distinct word
appears in the file
 sort datafile | uniq –c

 Sample application: analyze web server logs to


find popular URLs
Word Count (2)
 Case 1: Entire file fits in memory
 Case 2: File too large for mem, but all <word,
count> pairs fit in mem
 Case 3: File on disk, too many distinct words to fit in
memory
Word Count (3)
 To make it slightly harder, suppose we have a large
corpus of documents
 Count the number of times each distinct word occurs
in the corpus
 cat datafile | sed -r
's/[[:space:]]+/\n/g' | sed '/^$/d'
| sort | uniq -c
 The above captures the essence of MapReduce
 Great thing is it is naturally parallelizable
Example: Spam Collection
 Count words in spam
 [Link]
dataset#[Link]

14
Programming model
 Input & Output: each a set of key/value pairs

 Programmer specifies two functions:


map (in_key, in_value) -> list(out_key, intermediate_value)
 Processes input key/value pair
 Produces set of intermediate pairs
reduce (out_key, list(intermediate_value)) -> list(out_value)
 Combines all intermediate values for a particular key
 Produces a set of merged output values (usually just one)

 Inspired by similar primitives in LISP and other languages


MapReduce
 Input: a set of key/value pairs
 User supplies two functions:
 map(k1,v1) → list(k2,v2)
 reduce(k2, list(v2)) → list(v3)

 (k2,v2) is an intermediate key/value pair


 Output: for each k2, the output is a list of (k2, v3)
pairs.
 usually just one value or empty.
 k2 is omitted since it is pre-determined based on the
input
Word Count using MapReduce
map(key, value):
// key: document name; value: text of document
for each word w in value:
emit(w, 1)

reduce(key, values):
// key: a word; values: an iterator over counts
result = 0
for each count v in values:
result += v
emit(result)
Two blocks of the input
file

#iblock 1 #iblock 2
1 Algorithm design with MapReduce 1 MapReduce Algorithm implementattion
2 MapReduce Algorithm 2 Hadoop implmentation of MapReduce

Computing node 1: Invoke map Computing node 2: Invoke map


function on each key value pair function on each key value pair

(algorithm, 1), (design, 1), (with, 1), (MapReduce, 1) (MapReduce, 1), (algorithm, 1),
(implementation, 1)
(MapReduce, 1), (algorithm, 1) (Hadoop, 1), (implementation, 1), (of, 1),
(MapReduce, 1)
Shuffle and Sort

(algorithm, [1, 1, 1]), (desgin, [1]), (with, [1]), (MapReduce, [1, 1, 1, 1]), (implementation, [1, 1]),
(Hadoop, [1], (of, [1])

(algorithm, [1, 1, 1]), (desgin, [1]), (Hadoop, [1]) (implementation, [1, 1]), (MapReduce,
[1, 1, 1, 1]), (of, [1]), (with, [1])
Computing node 3 – Reducer 1: Invoke reduce
function on each pair Computing node 4 – Reducer 2: :
Invoke reduce function on each pair
(algorithm, 3), (design, 1), (Hadoop, 1) (implementation, 2), (MapReduce, 4), (of, 1), (with, 1)
Distributed Execution Overview
User
Program

fork fork fork

assign Master
assign
map reduce
Input Data Worker
write Output
local Worker File 0
Split 0 read
write
Split 1 Worker
Split 2 Output
Worker File 1
Worker remote
read,
sort
Data flow
 Input, final output are stored on a distributed file
system
 Scheduler tries to schedule map tasks “close” to
physical storage location of input data

 Intermediate results are stored on local FS of map


and reduce workers

 Output is often input to another map reduce task


Coordination
 Master data structures
 Task status: (idle, in-progress, completed)
 Idle tasks get scheduled as workers become available

 When a map task completes, it sends the master the location


and sizes of its R intermediate files, one for each reducer

◼ R: the number of reducers.

 Master pushes this info to reducers

 Master pings workers periodically to detect failures


Failures
 Map worker failure
 Map tasks completed or in-progress at worker are
reset to idle
 Reduce workers are notified when task is rescheduled
on another worker
 Reduce worker failure
 Only in-progress tasks are reset to idle
 Master failure
 MapReduce task is aborted and client is notified
Combiners
 Often a map task will produce many pairs of the
form (k,v1), (k,v2), … for the same key k
 E.g., popular words in Word Count

 Can save network time by pre-aggregating at


mapper
 combine(k1, list(v1)) → v2
 Usually same as reduce function

 Works only if reduce function is commutative and


associative
Partition Function
 Inputs to map tasks are created by contiguous splits of input
file

 For reduce, we need to ensure that records with the same


intermediate key end up at the same worker

 System uses a default partition function e.g., hash(key) mod R

 Sometimes useful to override


 E.g., hash(hostname(URL)) mod R ensures URLs from a host
end up in the same output file
Execution
Parallel Execution
Model is Widely Applicable
 MapReduce Programs In Google Source Tree

Example uses:
distributed grep distributed sort web link-graph reversal
term-vector / host web access log stats inverted index construction
document clustering machine learning statistical machine translation
... ... ...
Exercise 1: Host size
 Suppose we have a large web corpus
 Let’s look at the metadata file
 Lines of the form (URL, size, date, …)
 For each host, find the total number of bytes
 i.e., the sum of the page sizes for all URLs from that host
 Map (key= position, value = “URL, size, data, …”)
foreach hostname URL
emit(hostname, size)

 Reduce( key = hostname, value = size)


totalsize = 0
for each size v in sizes:
totalsize += v
emit(hostname, totalsize)

29
Exercise 2: Graph reversal
 Given a directed graph as an adjacency list:
src1: dest11, dest12, …
src2: dest21, dest22, …

 Construct the graph in which all the links are


reversed
 Map (key= filename, value = file content)
foreach line <src : destination list>
foreach dest in destination list
emit(dest, src)

 Reduce( key = node, value = rev_src )


String concat = node + “ : ”
foreach n in rev_src
concat += n + “ ”
emit (concat)
Exercise 4: Frequent Pairs
 Given a large set of market baskets, find all
frequent pairs
 Data: Basket1, Item11, Item12, …

 A lot of transaction files

 Each line of a transaction file is a list of items

 Threshold = t
 Map(key= marketbasket file, value=content)
foreach line=item_1, …., item_n in content
for i=1; i<n; i++
for j=i+1; j<=n; j++
emit(<item_i, item_j>, 1)

 Reduce(key= <item_i, item_j>, value = counts)


total = 0
foreach count in counts
total += count
if (total >= t) emit(total)
Exercise 5: Incoming Links
Given a set of HTML pages, compute the number of
incoming hyperlinks for each URL. For example,
suppose the a HTML file appears in 3 pages: 3
times in page A, 3 times in page B, and 4 times in
page C. Then its number of incoming hyper-links is
10.
Hadoop
 An open-source implementation of Map Reduce in
Java
 Uses HDFS for stable storage
 Download from:
[Link]

You might also like