1. CS246:Mining Massive Datasets Intro & MapReduce
Jure Leskovec, Stanford University

2. CS246 Course Staff TAs: Office hours: Bahman Bahmani Juthika Dabholkar
Add pictures of TAs
TAs:
Bahman Bahmani
Juthika Dabholkar
Pierre Kreitmann
Lu Li
Aditya Ramesh
Office hours:
Jure: Tuesdays 9-10am, Gates 418
See course website for TA office hours
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

3. Course Logistics Course website: http://cs246.stanford.edu
Lecture slides (at least 6h before the lecture)
Announcements, homeworks, solutions
Readings!
Readings: Book Mining of Massive Datasets by Anand Rajaraman and Jeffrey D. Ullman
Free online:
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

4. Work for the Course 4 longer homeworks: 40%
Theoretical and programming questions
All homeworks (even if empty) must be handed in
Assignments take time. Start early!
How to submit?
Paper: Box outside the class and in the Gates east wing
We will grade on paper!
You should also submit electronic copy:
1 PDF/ZIP file (writeups, experimental results, code)
Submission website:
SCPD: Only submit electronic copy & send us
7 late days for the quarter:
Max 5 late days per assignment
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

5. Work for the Course It’s going to be fun and hard work 
Short weekly quizzes: 20%
Short e-quizzes on Gradiance (see course website!)
First quiz is already online
You have 7 days to complete it. No late days!
Final exam: 40%
March 19 at 8:30am
It’s going to be fun and hard work 
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

6. Course Calendar Homework schedule:
No class: 1/16: Martin Luther King Jr /20: President’s day
Date
Out In
1/11
HW1
1/25
HW2
2/8
HW3
2/22
HW4
3/7
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

7. Prerequisites
Algorithms (CS161)
Dynamic programming, basic data structures
Basic probability (CS109 or Stat116)
Moments, typical distributions, MLE, …
Programming (CS107 or CS145)
Your choice, but C++/Java will be very useful
We provide some background, but the class will be fast paced
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

8. Recitation Sessions Recitation sessions:
Review of probability and statistics
Installing and working with Hadoop
We prepared a virtual machine with Hadoop preinstalled
HW0 helps you write your first Hadoop program
See course website!
We will announce the dates later
Sessions will be recorded
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

9. Relationship to CS345A CS345a: Data mining got split into 2 courses
Can skip CS345a and just say that there is a follow up class in Spring that is project oriented
CS345a: Data mining got split into 2 courses
CS246: Mining massive datasets:
Methods/algorithms oriented course
Homeworks (theory & programming)
No class project
CS341: Project in mining massive datasets:
Project oriented class
Lectures/readings related to the project
Unlimited access to Amazon EC2 cluster
We intend to keep the class small
Taking CS246 is basically prerequisite
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

10. Communication For questions/clarifications use Piazza!
If you don’t address us and we will register you
To communicate with the course staff use
We will post announcements to
If you are not registered or auditing send us and we will subscribe you!
You are welcome to sit-in & audit the class
Send us saying that you will be auditing
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

11. This Course: Web Mining
Chould skip!
Much of the course will be devoted to ways to data mining on the Web:
Mining to discover things about the Web
E.g., PageRank, finding spam sites
Mining data from the Web itself
E.g., analysis of click streams, similar products at Amazon, making recommendations
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

12. This Course: MapReduce
Much of the course will be devoted to large scale computing for data mining
Challenges:
How to distribute computation?
Distributed/parallel programming is hard
Map-reduce addresses all of the above
Google’s computational/data manipulation model
Elegant way to work with big data
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

13. Course Outline: Methods
High-dimensional data:
Locality Sensitive Hashing
Dimensionality reduction
Clustering
The data is a graph:
Link Analysis: PageRank, Hubs & Authorities
Machine Learning:
k-NN, Perceptron, SVM, Decision Trees
Data is infinite:
Mining data streams
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

14. Course Outline: Applications
Association Rules
Recommender systems
Advertising on the Web
Web spam detection
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

15. What is Data Mining? Knowledge discovery from data

16. What is Data Mining? Discovery of patterns and models that are:
Valid: hold on new data with some certainty
Useful: should be possible to act on the item
Unexpected: non-obvious to the system
Understandable: humans should be able to interpret the pattern
Subsidiary issues:
Data cleansing: detection of bogus data
Visualization: something better than MBs of output
Warehousing of data (for retrieval)
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

17. Data Mining Tasks Predictive Methods Descriptive Methods
Use some variables to predict unknown or future values of other variables
Descriptive Methods
Find human-interpretable patterns that describe the data
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

18. Why is it Hard? Scalability Dimensionality
Skip
Scalability
Dimensionality
Complex and Heterogeneous Data
Data Quality
Data Ownership and Distribution
Privacy Preservation
Streaming Data
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

19. Data Mining: Cultures Overlaps with: Different cultures:
Databases: Large-scale (non-main-memory) data
Machine learning: Complex methods, small data
Statistics: Models
Different cultures:
To a DB person, data mining is an extreme form of analytic processing – queries that examine large amounts of data
Result is the query answer
To a statistician, data-mining is the inference of models
Result is the parameters of the model
Statistics/ AI
Machine Learning/
Pattern Recognition
Data Mining
Database systems
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

20. Meaningfulness of Answers
A big data-mining risk is that you will “discover” patterns that are meaningless.
Bonferroni’s principle: (roughly) if you look in more places for interesting patterns than your amount of data will support, you are bound to find crap
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

21. Example: Rhine Paradox
Joseph Rhine was a parapsychologist in the 1950’s who hypothesized that some people had Extra-Sensory Perception
He devised an experiment where subjects were asked to guess 10 hidden cards – red or blue
He discovered that almost 1 in 1000 had ESP – they were able to get all 10 right!
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

22. Example: Rhine Paradox
The point is that the patterns should be real and significant
He told these people they had ESP and called them in for another test of the same type
Alas, he discovered that almost all of them had lost their ESP
What did he conclude?
He concluded that you shouldn’t tell people they have ESP; it causes them to lose it 
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

23. MapReduce

24. Single Node Architecture
CPU
Machine Learning, Statistics
Memory
“Classical” Data Mining
Disk
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

25. Motivation: Google Example
20+ billion web pages x 20KB = 400+ TB
1 computer reads MB/sec from disk
~4 months to read the web
~1,000 hard drives to store the web
Takes even more to do something useful with the data!
Standard architecture is emerging:
Cluster of commodity Linux nodes
Gigabit ethernet interconnect
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

26. Cluster Architecture … … 2-10 Gbps backbone between racks
Can skip this slide
2-10 Gbps backbone between racks
1 Gbps between
any pair of nodes
in a rack
Switch
Switch
Switch
Mem
Disk
CPU
Mem
Disk
CPU
Mem
Disk
CPU
Mem
Disk
CPU … …
Each rack contains nodes
In Aug 2006 Google had ~450,000 machines
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

27. Large-scale Computing
Large-scale computing for data mining problems on commodity hardware
Challenges:
How do you distribute computation?
How can we make it easy to write distributed programs?
Machines fail:
One server may stay up 3 years (1,000 days)
If you have 1,0000 servers, expect to loose 1/day
In Aug 2006 Google had ~450,000 machines
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

28. Idea and Solution Idea: Map-reduce addresses these problems
Bring computation close to the data
Store files multiple times for reliability
Map-reduce addresses these problems
Google’s computational/data manipulation model
Elegant way to work with big data
Storage Infrastructure – File system
Google: GFS
Hadoop: HDFS
Programming model
Map-Reduce
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

29. Storage Infrastructure
Problem
If nodes fail, how to store data persistently?
Answer
Distributed File System:
Provides global file namespace
Google GFS; Hadoop HDFS;
Typical usage pattern
Huge files (100s of GB to TB)
Data is rarely updated in place
Reads and appends are common
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

30. Distributed File System
Chunk Servers
File is split into contiguous chunks
Typically each chunk is 16-64MB
Each chunk replicated (usually 2x or 3x)
Try to keep replicas in different racks
Master node
a.k.a. Name Nodes in Hadoop’s HDFS
Stores metadata
Might be replicated
Client library for file access
Talks to master to find chunk servers
Connects directly to chunk servers to access data
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

31. Distributed File System
Reliable distributed file system
Data kept in “chunks” spread across machines
Each chunk replicated on different machines
Seamless recovery from disk or machine failure C0 C1 D0 C1 C2 C5 C0 C5 …
Chunk server 1
Chunk server 2 D0
Chunk server 3 D1 D0
Chunk server N C5 C2 C5 C3 C2
Bring computation directly to the data!
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets, 31

32. Programming Model: MapReduce
Warm-up task:
We have a huge text document
Count the number of times each distinct word appears in the file
Sample application:
Analyze web server logs to find popular URLs
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

33. Word Count Case 1: Case 2: Count occurrences of words:
File too large for memory, but all <word, count> pairs fit in memory
Case 2:
Count occurrences of words:
words(doc.txt) | sort | uniq -c
where words takes a file and outputs the words in it, one per a line
Captures the essence of MapReduce
Great thing is it is naturally parallelizable
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

34. Outline stays the same, map and reduce change to fit the problem
Map-Reduce: Overview
Sequentially read a lot of data
Map:
Extract something you care about
Group by key: Sort and Shuffle
Reduce:
Aggregate, summarize, filter or transform
Write the result
Outline stays the same, map and reduce change to fit the problem
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

35. MapReduce: The Map Step
Input
key-value pairs
Intermediate
key-value pairs k v
map v k
map v k k v … … k v v k
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

36. MapReduce: The Reduce Step
Output
key-value pairs k v …
Intermediate
key-value pairs k v …
Key-value groups
reduce k v
reduce k v
group … k v
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

37. More Specifically Input: a set of key/value pairs
Programmer specifies two methods:
Map(k, v)  <k’, v’>*
Takes a key value pair and outputs a set of key value pairs
E.g., key is the filename, value is a single line in the file
There is one Map call for every (k,v) pair
Reduce(k’, <v’>*)  <k’, v’’>*
All values v’ with same key k’ are reduced together and processed in v’ order
There is one Reduce function call per unique key k’
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

38. Map-Reduce: Word Counting
Bif\gger document text. So that people can read the example – know the answer.
Provided by the programmer
Provided by the programmer
MAP:
reads input and produces a set of key value pairs
Group by key:
Collect all pairs with same key
Reduce:
Collect all values belonging to the key and output
Sequentially read the data
Only sequential reads
The crew of the space shuttle Endeavor recently returned to Earth as ambassadors, harbingers of a new era of space exploration. Scientists at NASA are saying that the recent assembly of the Dextre bot is the first step in a long-term space-based man/machine partnership. '"The work we're doing now -- the robotics we're doing -- is what we're going to need to do to build any work station or habitat structure on the moon or Mars," said Allard Beutel.
(the, 1)
(crew, 1)
(of, 1)
(space, 1)
(shuttle, 1)
(Endeavor, 1)
(recently, 1) ….
(crew, 1)
(space, 1)
(the, 1)
(shuttle, 1)
(recently, 1) …
(crew, 2)
(space, 1)
(the, 3)
(shuttle, 1)
(recently, 1) …
Big document
(key, value)
(key, value)
(key, value)
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

39. Word Count Using MapReduce
map(key, value):
// key: document name; value: text of the document
for each word w in value:
emit(w, 1)
reduce(key, values):
// key: a word; value: an iterator over counts
result = 0
for each count v in values:
result += v
emit(key, result)
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

40. Map-Reduce: Environment
Map-Reduce environment takes care of:
Partitioning the input data
Scheduling the program’s execution across a set of machines
Performing the group by key step
Handling machine failures
Managing required inter-machine communication
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

41. Map-Reduce: A diagram Big document MAP: Group by key: Reduce:
The sort is the magical part. The map reduce does it by itself.
Big document
MAP:
reads input and produces a set of key value pairs
Group by key:
Collect all pairs with same key
Call it hash merge
Group by key
Sort
Call it Partition or
Hash Merge
Reduce:
Collect all values belonging to the key and output
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

42. Map-Reduce: In Parallel
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

43. Data Flow
Input and 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
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

44. Coordination Master data structures:
Skip
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
Master pushes this info to reducers
Master pings workers periodically to detect failures
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

45. Dealing with Failures Map worker failure Reduce worker failure
How exactly is reducer failure recovered?
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
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

46. How many Map and Reduce jobs?
Debugging have 1 mapper1 reducer
M and R are independent of chungs, system diced which mapper gets what part of the input file.
Similarly for reducers.
M map tasks, R reduce tasks
Rule of a thumb:
Make M and R much larger than the number of nodes in cluster
One DFS chunk per map is common
Improves dynamic load balancing and speeds recovery from worker failure
Usually R is smaller than M
because output is spread across R files
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

47. Task Granularity & Pipelining
Fine granularity tasks: map tasks >> machines
Minimizes time for fault recovery
Can pipeline shuffling with map execution
Better dynamic load balancing
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

48. Refinements: Backup Tasks
Picture with chunk servers
Problem
Slow workers significantly lengthen the job completion time:
Other jobs on the machine
Bad disks
Weird things
Solution
Near end of phase, spawn backup copies of tasks
Whichever one finishes first “wins”
Effect
Dramatically shortens job completion time
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

49. Refinements: Combiners
Example back to the word count
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 the Word Count example
Can save network time by pre-aggregating values at the mapper:
combine(k, list(v1))  v2
Combiner is usually same as the reduce function
Works only if reduce function is commutative and associative
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

50. Refinements: Partition Function
Inputs to map tasks are created by contiguous splits of input file
Reduce needs to ensure that records with the same intermediate key end up at the same worker
System uses a default partition function:
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
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

51. Problems Suited for Map-Reduce
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

52. Example: Host size Suppose we have a large web corpus
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
Other examples:
Link analysis and graph processing
Machine Learning algorithms
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

53. Example: Language Model
Statistical machine translation:
Need to count number of times every 5-word sequence occurs in a large corpus of documents
Very easy with MapReduce:
Map:
Extract (5-word sequence, count) from document
Reduce:
Combine counts
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

54. Example: Join By Map-Reduce
Join was bad
Compute the natural join R(A,B) ⋈ S(B,C)
R and S each are stored in files
Tuples are pairs (a,b) or (b,c) A B a1 b1 a2 a3 b2 a4 b3 B C b2 c1 c2 b3 c3 A C a3 c1 c2 a4 c3 ⋈ =
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

55. Map-Reduce Join Use a hash function h from B-values to 1...k
Explain better what’s going on
What are tuples?
Give animation, …
Use a hash function h from B-values to 1...k
A Map process turns:
Each input tuple R(a,b) into key-value pair (b,(a,R))
Each input tuple S(b,c) into (b,(c,S))
Map processes send each key-value pair with key b to Reduce process h(b).
Hadoop does this automatically; just tell it what k is.
Each Reduce process matches all the pairs (b,(a,R)) with all (b,(c,S)) and outputs (a,b,c).
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

56. END
-- Intro is too birocratic. Maybe it would be better to create a printout and hand it out to the students or tell them to read the class website
-- What is data mining is too abstract? Maybe say an application or two. Check how Manning and Ng do intro lectures
-- MapReduce was good
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

57. Cost Measures for Algorithms
Communication cost = total I/O of all processes.
Elapsed communication cost = max of I/O along any path.
(Elapsed ) computation costs analogous, but count only running time of processes.
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

58. Example: Cost Measures
For a map-reduce algorithm:
Communication cost = input file size + 2  (sum of the sizes of all files passed from Map processes to Reduce processes) + the sum of the output sizes of the Reduce processes.
Elapsed communication cost is the sum of the largest input + output for any map process, plus the same for any reduce process
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

59. What Cost Measures Mean
Either the I/O (communication) or processing (computation) cost dominates
Ignore one or the other
Total costs tell what you pay in rent from your friendly neighborhood cloud
Elapsed costs are wall-clock time using parallelism
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

60. Cost of Map-Reduce Join
Total communication cost = O(|R|+|S|+|R ⋈ S|)
Elapsed communication cost = O(s)
We’re going to pick k and the number of Map processes so I/O limit s is respected
We put a limit s on the amount of input or output that any one process can have. s could be:
What fits in main memory
What fits on local disk
With proper indexes, computation cost is linear in the input + output size
So computation costs are like comm. costs
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

61. Pointers and Further Reading
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

62. Implementations Google Hadoop Not available outside Google
An open-source implementation in Java
Uses HDFS for stable storage
Download:
Aster Data
Cluster-optimized SQL Database that also implements MapReduce
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

63. Cloud Computing Ability to rent computing by the hour
Additional services e.g., persistent storage
Amazon’s “Elastic Compute Cloud” (EC2)
Aster Data and Hadoop can both be run on EC2
For CS341 (offered next quarter) Amazon will provide free access for the class
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

64. Reading
Jeffrey Dean and Sanjay Ghemawat: MapReduce: Simplified Data Processing on Large Clusters
Sanjay Ghemawat, Howard Gobioff, and Shun-Tak Leung: The Google File System
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

65. Resources Hadoop Wiki Javadoc Introduction Getting Started
Getting Started
Map/Reduce Overview
Eclipse Environment
Javadoc
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

66. Resources Releases from Apache download mirrors
Nightly builds of source
Source code from subversion
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,

67. Further Reading
Programming model inspired by functional language primitives
Partitioning/shuffling similar to many large-scale sorting systems
NOW-Sort ['97]
Re-execution for fault tolerance
BAD-FS ['04] and TACC ['97]
Locality optimization has parallels with Active Disks/Diamond work
Active Disks ['01], Diamond ['04]
Backup tasks similar to Eager Scheduling in Charlotte system
Charlotte ['96]
Dynamic load balancing solves similar problem as River's distributed queues
River ['99]
9/22/2018
Jure Leskovec, Stanford CS246: Mining Massive Datasets,