Big Data
Hadoop Stack
Lecture #1
Hadoop Beginnings
What is Hadoop?
Apache Hadoop is an open source
software framework for storage and
large scale processing of data-sets on
clusters of commodity hardware
Hadoop was created by Doug Cutting
and Mike Cafarella in 2005
Named the project after son's toy
elephant
Moving Computation to Data
Data
Computation
Scalability at Hadoop’s core!
Reliability!
Reliability!
Reliability!
Reliability!
Reliability!
Reliability!
Reliability!
Reliability!
Reliability!
Google File System
Once
a year
365 Computers
Once
a day
Hourly
New Approach to Data
Keep all data
New Kinds of Analysis
Schema-on read style
New Kinds of Analysis
Schema-on read style
New analysis
Small Data & Complex
Algorithm
Vs.
Large Data &
Simple Algorithm
Lecture #2
Apache Framework
Hadoop Modules
Apache Framework
Basic Modules
Hadoop Common
Hadoop Distributed File System
(HDFS)
Hadoop YARN
Hadoop MapReduce
Apache Framework Basic
Modules
Hadoop Common
Hadoop Distributed File System
(HDFS)
Hadoop YARN
Hadoop MapReduce
Apache Framework Basic
Modules
Hadoop Common
Hadoop Distributed File System
(HDFS)
Hadoop YARN
Hadoop MapReduce
Apache Framework Basic
Modules
Hadoop Common
Hadoop Distributed File System
(HDFS)
Hadoop YARN
Hadoop MapReduce
Lecture #3
Hadoop Distributed File System
(HDFS)
HDFS
Hadoop Distributed File System
Distributed, scalable, and portable file-
system written in Java for the Hadoop
framework
HDFS
MapReduce Engine
Job Tracker
MapReduce Engine
Job Tracker
Task Tracker
Apache Hadoop NextGen
MapReduce (YARN)
What is Yarn?
• YARN enhances the power of a
Hadoop compute cluster
Scalability
What is Yarn?
• YARN enhances the power of a
Hadoop compute cluster
Scalability
Improved cluster utilization
What is Yarn?
• YARN enhances the power of a
Hadoop compute cluster
Scalability Improved cluster utilization
MapReduce Compatibility
What is Yarn?
• YARN enhances the power of a
Hadoop compute cluster
Scalability Improved cluster utilization
Supports Other
MapReduce Compatibility
Workloads
Lecture #4
The Hadoop “Zoo”
How to figure out the Zoo??
Original Google Stack
Original Google Stack
Original Google Stack
Original Google Stack
Data Integration
Original Google Stack
Original Google Stack
Original Google Stack
Facebook’s Version of the Stack
Data Integration Languages
Compilers
Data Store
Coordination
Yahoo’s Version of the Stack
Data Integration Languages
Compilers
Data Store
Coordination
LinkedIn’s Version of the Stack
Data Integration Languages
Compilers
Data Store
Coordination
Cloudera’s Version of the Stack
Data Integration Languages
Compilers
Data Store
Coordination
Lecture #5
Hadoop Ecosystem
Major Components
Apache Sqoop
• Tool designed for
efficiently transferring
bulk data between
Apache Hadoop and
structured datastores
such as relational
databases
HBASE
• Column-oriented database
management system
• Key-value store
• Based on Google Big Table
• Can hold extremely large data
• Dynamic data model
• Not a Relational DBMS
PIG
High level programming on top of
Hadoop MapReduce
PIG
The language: Pig Latin
PIG
Data analysis problems as data flows
PIG
Originally developed at Yahoo 2006
Pig for ETL
Pig for ETL
Pig for ETL
Apache Hive
• Data warehouse software facilitates
querying and managing large
datasets residing in distributed
storage
Apache Hive
SQL-like language!
Apache Hive
Facilitates querying and
managing large datasets in
HDFS
Apache Hive
Mechanism to project
structure onto this data and
query the data using a SQL-
like language called HiveQL
Oozie
Workflow scheduler system to manage
Apache Hadoop jobs
Oozie
Oozie Coordinator jobs!
Oozie
Supports
MapReduce, Pig, Apache Hive,
and Sqoop, etc.
Zookeeper
Provides operational services for a
Hadoop cluster group services
Zookeeper
Centralized service for:
maintaining configuration information
naming services
providing distributed synchronization
and providing group services
Zookeeper
Centralized service for:
maintaining configuration information
Zookeeper
Centralized service for:
maintaining configuration information
naming services
Zookeeper
Centralized service for:
maintaining configuration information
naming services
providing distributed synchronization
and providing group services
Flume
Distributed, reliable, and available
service for efficiently collecting,
aggregating, and moving large
amounts of log data
Additional Cloudera Hadoop
Components
Impala
Impala
• Cloudera's open source massively
parallel processing (MPP) SQL query
engine Apache Hadoop
Additional Cloudera Hadoop
Components
Spark
The New Paradigm
Spark
Apache Spark™ is a fast and
general engine for large-scale
data processing
Spark Benefits
Multi-stage in-memory primitives
provides performance up to 100 times
faster for certain applications
Spark Benefits
Allows user programs to load data into
a cluster's memory and query it
repeatedly
Well-suited to machine learning!!!
Up Next
Tour of the Cloudera’s Quick Start VM