1. Lecture 1 Big Data + Cloud Computing张奇
复旦大学
COMP Data Intensive Computing

2. Big Data
Cloud Computing
Data Intensive Computing
Grid Computing
Parallel Computing

3. MPI
HADOOP
MAP REDUCE
NO SQL
GFS
EC2
HDFS S3

5. How much data? Wayback Machine has 2 PB + 20 TB/month (2006)
Google processes 20 PB a day (2008)
“all words ever spoken by human beings” ~ 5 EB
NOAA has ~1 PB climate data (2007)
CERN’s LHC will generate 15 PB a year (2008)
640K ought to be enough for anybody.

6. Network traffic (spam)
Happening everywhere!
Molecular biology
(cancer)
microarray chips
fiber optics
Network traffic (spam)
300M/day
Simulations
(Millennium)
microprocessors
particle colliders
Particle events (LHC) 1B

7. Maximilien Brice, © CERN

8. Maximilien Brice, © CERN

9. Example: Wikipedia Anthropology
Kittur, Suh, Pendleton (UCLA, PARC), “He Says, She Says: Conflict and Coordination in Wikipedia” CHI, 2007
Increasing fraction of edits are for
work indirectly related to articles
Computation
Hadoop on 20-machine cluster
Experiment
Download entire revision history of Wikipedia
4.7 M pages, 58 M revisions, 800 GB
Analyze editing patterns & trends
3 slides from
“Data Intensive Super Computing”
Randal E. Bryant, Carnegie Mellon University

10. Example: Scene Completion
Hays, Efros (CMU), “Scene Completion Using Millions of Photographs” SIGGRAPH, 2007
Image Database Grouped by Semantic Content
30 different Flickr.com groups
2.3 M images total (396 GB).
Select Candidate Images Most Suitable for Filling Hole
Classify images with gist scene detector [Torralba]
Color similarity
Local context matching
Computation
Index images offline
50 min. scene matching, 20 min. local matching, 4 min. compositing
Reduces to 5 minutes total by using 5 machines
Extension
Flickr.com has over 500 million images …

11. Example: Web Page Analysis
Fetterly, Manasse, Najork, Wiener (Microsoft, HP), “A Large-Scale Study of the Evolution of Web Pages,” Software-Practice & Experience, 2004
Experiment
Use web crawler to gather 151M HTML pages weekly 11 times
Generated 1.2 TB log information
Analyze page statistics and change frequencies
Systems Challenge
“Moreover, we experienced a catastrophic disk failure during the third crawl, causing us to lose a quarter of the logs of that crawl.”

12. ? Reads Subject genome Sequencer GATGCTTACTATGCGGGCCCC
CGGTCTAATGCTTACTATGC
GCTTACTATGCGGGCCCCTT
AATGCTTACTATGCGGGCCCCTT ?
TAATGCTTACTATGC
AATGCTTAGCTATGCGGGC
AATGCTTACTATGCGGGCCCCTT
AATGCTTACTATGCGGGCCCCTT
CGGTCTAGATGCTTACTATGC
AATGCTTACTATGCGGGCCCCTT
CGGTCTAATGCTTAGCTATGC
ATGCTTACTATGCGGGCCCCTT
Reads
Subject genome
Sequencer

13. DNA Sequencing
Genome of an organism encodes genetic information in long sequence of 4 DNA nucleotides: ATCG
Bacteria: ~5 million bp(base pairs)
Humans: ~3 billion bp
Current DNA sequencing machines can generate 1-2 Gbp of sequence per day, in millions of short reads (25-300bp)
Shorter reads, but much higher throughput
Per-base error rate estimated at 1-2% (Simpson, et al, 2009)
Recent studies of entire human genomes have used 3.3 (Wang, et al., 2008) & 4.0 (Bentley, et al., 2008) billion 36bp reads
~144 GB of compressed sequence data
4 slides from Michael Schatz.
CloudBurst: Highly Sensitive Read Mapping with MapReduce. Bioinformatics, 2009, in press.
ATCTGATAAGTCCCAGGACTTCAGT
GCAAGGCAAACCCGAGCCCAGTTT
TCCAGTTCTAGAGTTTCACATGATC
GGAGTTAGTAAAAGTCCACATTGAG

14. Subject reads
CTATGCGGGC
CTAGATGCTT
A T CTATGCGG
TCTAGATGCT
GCTTA T CTAT
A T CTATGCGG
A T CTATGCGG
A T CTATGCGG
TTA T CTATGC
CTATGCGGGC
GCTTA T CTAT
Alignment
CGGTCTAGATGCTTAGCTATGCGGGCCCCTT
Reference sequence

15. Subject reads
ATGCGGGCCC
CTAGATGCTT
CTATGCGGGC
TCTAGATGCT
ATCTATGCGG
CGGTCTAG
ATCTATGCGG
CTT
CGGTCT
TTATCTATGC
CCTT
CGGTC
GCTTATCTAT
GCCCCTT
CGG
GCTTATCTAT
GGCCCCTT
CGGTCTAGATGCTTATCTATGCGGGCCCCTT
Reference sequence

16. Example: Bioinformatics
Michael Schatz. CloudBurst: Highly Sensitive Read Mapping with MapReduce. Bioinformatics, 2009, in press.
Evaluate running time on local 24 core cluster
Running time increases linearly with the number of reads

17. Example: Data Mining
Haoyuan Li, Yi Wang, Dong Zhang, Ming Zhang, Edward Y. Chang: Pfp: parallel fp-growth for query recommendation. RecSys 2008:
del.icio.us crawl->a bipartite graph covering Webpages and tags.

18. There’s nothing like more data!
s/inspiration/data/g;
(Banko and Brill, ACL 2001)
(Brants et al., EMNLP 2007)

19. Big Data + Cloud Computing

20. What is Cloud Computing?
First write down your own opinion about “cloud computing” , whatever you thought about in your mind.
Question: What ? Who? Why? How? Pros and cons?
The most important question is: What is the relation with me?
From “Introduction to Cloud Computing and Technical Issues” Eom, Hyeonsang

21. What Is Cloud Computing?
Internet computing
Computation done through the Internet
No concern about any maintenance or management of actual resources
Shared computing resources
As opposed to local servers and devices
Comparable to Grid Infrastructure
Web applications
Specialized raw computing services
From “Introduction to Cloud Computing and Technical Issues” Eom, Hyeonsang

22. Cloud Computing Resources
Large pool of easily usable and accessible virtualized resources
Dynamic reconfiguration for adjusting to different loads (scales)
Optimization of resource utilization
Pay-per-use model
Guarantees offered by the Infrastructure Provider by means of customized SLAs(Service Level Agreements)
Set of features
Scalability, pay-per-use utility model and virtualization
From “Introduction to Cloud Computing and Technical Issues” Eom, Hyeonsang

23. Technical Key Points
User interaction interface: how users of cloud interface with the cloud
Services catalog: services a user can request
System management: management of available resources
Provisioning tool: carving out the systems from the cloud to deliver on the requested service
Monitoring and metering: tracking the usage of the cloud
Servers: virtual or physical servers managed by system administrators
From “Introduction to Cloud Computing and Technical Issues” Eom, Hyeonsang

24. Key Characteristics (1/2)
Cost savings for resources
Cost is greatly reduced as initial expense and recurring expenses are much lower than traditional computing
Maintenance cost is reduced as a third party maintains everything from running the cloud to storing data
Platform, Location and Device independency
Adoptable for all sizes of businesses, in particular small and mid-sized ones
From “Introduction to Cloud Computing and Technical Issues” Eom, Hyeonsang

25. Key Characteristics (2/2)
Scalable services and applications
Achieved through server virtualization technology
Redundancy and disaster recovery
From “Introduction to Cloud Computing and Technical Issues” Eom, Hyeonsang

26. Timeline
From “Introduction to Cloud Computing and Technical Issues” Eom, Hyeonsang

27. Cloud Computing Architecture
Front End
End user, client or any application (i.e., web browser, etc.)
Back End (Cloud services)
Network of servers with any computer program and data storage system
It is usually assumed that clouds have infinite storage capacity for any software available in market

29. Cloud Service Taxonomy
Layer
Software-as-a-Service (SaaS)
Platform-as-a-Service (PaaS)
Infrastructure-as-a-Service (IaaS)
Data Storage-as-a-Service (DaaS)
Communication-as-a-Service (CaaS)
Hardware-as-a-Service(HaaS)
Type
Public cloud
Private cloud
Inter-cloud

30. Infrastructure as a Service

31. What Is Infrastructure-as-a-Service (IaaS)
Characteristics
Utility computing and billing model
Automation of administrative tasks
Dynamic scaling
Desktop virtualization
Policy-based services
Internet connectivity

32. Use Scenario for IaaS

33. Business Example: Amazon EC2

34. Amazon EC2 CLI (Client Level Interface)

35. Amazon EC2 Automated Management

36. Data Storage as a Service (DaaS)
Definition
Delivery of data storage as a service, including database-like services, often billed on a utility computing basis
Database (Amazon SimpleDB & Google App Engine's BigTable datastore)
Network attached storage (MobileMe iDisk & Nirvanix CloudNAS)
Synchronization (Live Mesh Live Desktop component & MobileMe push functions)
Web service (Amazon Simple Storage Service & Nirvanix SDN)

37. Amazon S3

38. Platform as a Service

39. PssS Example: Google App Engine
Service that allows user to deploy user’s Web applications on Google's very scalable architecture
Providing user with a sandbox for user’s Java and Python application that can be referenced over the Internet
Providing Java and Python APIs for persistently storing and managing data (using the Google Query Language or GQL)

40. Google App Engine

41. Google App Engine (Python Example)

42. Software-as-a-Service (SaaS)
Definition
Software deployed as a hosted service and accessed over the Internet
Features
Open, Flexible
Easy to Use
Easy to Upgrade
Easy to Deploy

43. Software as a Service

44. Human as a Service

45. Administration/Business Support

46. Cloud Architecture -> Cloud Players

47. Players

48. Players: Providers

49. Players: Cloud Intermediaires

50. Players: Application Providers

51. Programming Model

52. Web-Scale Problems? Don’t hold your breath: It all boils down to…
Biocomputing
Nanocomputing
Quantum computing …
It all boils down to…
Divide-and-conquer
Throwing more hardware at the problem
Simple to understand… a lifetime to master…

53. Divide and Conquer Partition Combine “Work” w1 w2 w3 r1 r2 r3 “Result”
“worker”
“worker”
“worker” r1 r2 r3
Combine
“Result”

54. Different Workers Different threads in the same core
Different cores in the same CPU
Different CPUs in a multi-processor system
Different machines in a distributed system

55. Choices, Choices, Choices
Commodity vs. “exotic” hardware
Number of machines vs. processor vs. cores
Bandwidth of memory vs. disk vs. network
Different programming models

56. Flynn’s Taxonomy Instructions Single (SI) Multiple (MI) Single (SD)
SISD
Single-threaded process
MISD
Pipeline architecture
SIMD
Vector Processing
MIMD
Multi-threaded Programming
Data
Multiple (MD)

57. SISD
Processor D D D D D D D
Instructions

58. SIMD D0 D0 D0 D0 D0 D0 D0 D1 D1 D1 D1 D1 D1 D1 D2 D2 D2 D2 D2 D2 D2 D3
Processor D0 D0 D0 D0 D0 D0 D0 D1 D1 D1 D1 D1 D1 D1 D2 D2 D2 D2 D2 D2 D2 D3 D3 D3 D3 D3 D3 D3 D4 D4 D4 D4 D4 D4 D4 … … … … … … … Dn Dn Dn Dn Dn Dn Dn
Instructions

59. MIMD D
Processor
Instructions D
Processor
Instructions

60. Memory Typology: Shared
Processor
Processor
Memory
Processor
Processor

61. Memory Typology: Distributed
Processor
Memory
Processor
Memory
Network
Processor
Memory
Processor
Memory

62. Memory Typology: Hybrid
Processor
Memory
Processor
Memory
Processor
Processor
Network
Processor
Memory
Processor
Memory
Processor
Processor

63. Parallelization Problems
How do we assign work units to workers?
What if we have more work units than workers?
What if workers need to share partial results?
How do we aggregate partial results?
How do we know all the workers have finished?
What if workers die?
What is the common theme of all of these problems?

64. General Theme? Parallelization problems arise from:
Communication between workers
Access to shared resources (e.g., data)
Thus, we need a synchronization system!
This is tricky:
Finding bugs is hard
Solving bugs is even harder

65. Managing Multiple Workers
Difficult because
(Often) don’t know the order in which workers run
(Often) don’t know where the workers are running
(Often) don’t know when workers interrupt each other
Thus, we need:
Semaphores (lock, unlock)
Conditional variables (wait, notify, broadcast)
Barriers
Still, lots of problems:
Deadlock, livelock, race conditions, ...
Moral of the story: be careful!
Even trickier if the workers are on different machines

66. Patterns for Parallelism
Parallel computing has been around for decades
Here are some “design patterns” …

67. Master/Slaves
master
slaves

68. Producer/Consumer Flow

69. Work Queues P C
shared queue P W W W W W C P C

70. Rubber Meets Road From patterns to implementation: The reality:
pthreads, OpenMP for multi-threaded programming
MPI for clustering computing …
The reality:
Lots of one-off solutions, custom code
Write you own dedicated library, then program with it
Burden on the programmer to explicitly manage everything
MapReduce to the rescue!
(for next time)

71. Questions?