Head First Hadoop

Head First Hadoop
http://www.jco.or.kr/conference/data/10th/Merlin-1.zip
제 10회 한국자바개발자 컨퍼런스에서 심탁길님이 발표하신 자료입니다.

발표 자료를 보니 심탁길은 NHN에 근무하시는 듯 합니다.
그리고 NHN에서는 NHN Hadoop Pool을 운영중이고, 120대 장비에서 50TB의 데이터를 하루에 약 10만개정도의 테스크를 처리하는 것 같습니다.

Web Scale Data Processing, How?
Money Talks?
Hadoop is ...
Hadoop Architecture
Hadoop Distributed File System
HDFS Architecture
Pipelining Write
Check Sum Read
IO Performance and Scalability
DataNode Failure
NameNode Failure
Enhancement
LAB1: HDFS
MapReduce is
MapReduce Concept
Hadoop MapReduce Flow
Class Implementation
Optional Class Implementation
Mapper for Word Count
Reducer for Word Count
InputFormat for Word Count
OutputFormat for Word Count
Driver and Run
Task Procedures
LAB2: MapReduce
Contributions: Hadoop Streaming
Contributions: eclipse Plug-In
LAB3: eclipse plug-in
MapReduce is a Piece of Cake?
For Better Productivity
Hive Architecture
Hive
Hive CLI
DDL & DML Operation
Join
Running Custom MapReduce Scripts
LAB4: Hive
Case Study 1: PageRank & Hadoop
Create Link Graph
PageRank Distribution
Cleanup and Sorting
Mapper & Reducer
Page Rank with Hive
Case Study 2: NHN Hadoop Pool
Motivations of Hadoop Pool
Some Stats
Flow of Execution
Learning from Experience