ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted
C03934969, January 2014
IntroductiontoRandHadoop
HP ES Korea
์ด์ฑ๋ณต
2.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted2
๋ด์ฉ
๏ง ๋น ๋ฐ์ดํฐ ๋ถ์์ ์ํด ํ์ํ ๊ธฐ์
๏ง R์ ์์๋ณด์
๏ง ๋น ๋ฐ์ดํฐ ๋ถ์์ ์ํด R๊ณผ Hadoop์ ์ด๋ป๊ฒโฆ
๏ง R + Hadoop = RHadoop?
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted9
R์ด๋?
โR is a language and environment for
statistical computing and graphics. It is a
GNU project which is similar to the S
language and environment which was
developed at Bell Laboratories (formerly
AT&T, now Lucent Technologies) by John
Chambers and colleagues. R can be considered
as a different implementation of S. There are
some important differences, but much code
written for S runs unaltered under R.โ
(http://www.r-project.org/about.html)
R์ ๋ฐ์ดํฐ ๋ถ์์ ์ํ ํต๊ณ์ ๊ทธ๋ํฝ์ค๋ฅผ ์ง์ํ๋
์คํ ์ํํธ์จ์ด ํ๊ฒฝ์ด๋ค.
๏ง ๋ฐ์ดํฐ ๋ถ์์ ์ํ ์ํํธ์จ์ด
๏ง ํ๋ก๊ทธ๋๋ฐ ์ธ์ด
๏ผ ํต๊ณํ์๋ค์ด ๋์์ธํ, ํต๊ณํ์๋ค์ ์ํ ๊ฐ๋ฐ
ํ๋ซํผ
๏ง ๊ฐ๋ฐ ํ๊ฒฝ(Environment) ์ ๊ณต
๏ผ ๋ฐ์ดํฐ์ ๊ด๋ จ๋ ์ ์ถ๋ ฅ, ์ฒ๋ฆฌ, ๊ด๋ฆฌ, ๋ถ์, ๊ทธ๋ํ ๋ฑ
์ต์ ์ ์๊ณ ๋ฆฌ์ฆ๊ณผ ๋ผ์ด๋ธ๋ฌ๋ฆฌ ์ ๊ณต
๏ง ์คํ์์ค ์ํํธ์จ์ด ํ๋ก์ ํธ
๏ผ Free, open, active
๏ง ์ปค๋ฎค๋ํฐ
๏ผ ์ ์ฒ๋ช ์ ๊ธฐ์ฌ์๋ค, 2๋ฐฑ๋ง์ด ๋๋ ์ฌ์ฉ์๋ค
๏ผ ๊ฐ ์ ๋ฌด ๋๋ฉ์ธ๊ณผ ๊ด๋ จ๋ ๋ฆฌ์์ค์ ๋์๋ง ์ ๊ณต
10.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted10
R์ ์ฃผ ์ฌ์ฉ์ฒ
Statistical Analysis & Modeling
โข Classification
โข Scoring
โข Ranking
โข Clustering
โข Finding relationships
โข Characterization
Common Uses
โข Interactive Data Analysis
โข General Purpose Statistics
โข Predictive Modeling
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted15
R์ ์ญ์ฌ
โข R์ 1993๋ ๋ด์ง๋๋
์คํด๋๋๋ํ์ ํต๊ณํ๊ณผ ๊ต์
2๋ช (Ross Ihaka, Robert
Gentleman)์ ์ํ์ฌ ๊ฐ๋ฐ
โข 1976๋ Bell Lab์ John Chambers,
Rick Becker, Allan Wilks์ ์ํ์ฌ
๊ฐ๋ฐ๋ S Language์ ๊ทธ ๋ฟ๋ฆฌ๋ฅผ
๋๊ณ ์์
1976 1980 1988 1998
Version 4
Java interface
Class/Method
Version 3
C-base
Class/Method
Version2
UNIX
Version1
Fortran-
based
1988 1993 2001 2008
StacSci Insightful
WithMathSoft
E-license
05/V.7/Bigdata
07/V.8/Rpackage
TIBCO
1993 1997.4.1 1997.4.23 2000.11997.12.5
CRANMailingList Version1.0GNUProjectRoss Ihaka
Robert Gentleman
Version3.2.1
2015.6.18
John Chamber
16.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted16
[์ฐธ๊ณ ] R์ ์ญ์ฌ โ OS์ System ๊ด์
O/S
Analysis
System
UNIX
The S system
ํ๋ ํ์ฐ ์ ํ
Bell Lab
BSD/System V
HP, IBM, SUN
S-PLUS
Commercial
LINUX
Application
R
Packages
GNU/OpenSource
1976๋ Bell Lab ํ์ 1988๋ ๋ผ์ด์ผ์ค ์๋ 1993๋ ๋ ์คํ์์คํ
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted22
Why Hadoop for Big Data Analysis?
๏ง Hadoop has become the kernel of the distributed operating system for Big Data
๏ง ๋๋ถ๋ถ์ ๋ฐ์ดํฐ ์ฒ๋ฆฌ๋ ๋ฐ์ดํฐ ๋ถ์์ ์ํ ๊ธฐ๋ฐ ์์
๏ง ์ด๋ฏธ ๋ง์ ์ ์ฒด์์ Hadoop์ ๋ฐ์ดํฐ ๋ถ์ ๋ฐ ์ฒ๋ฆฌ ์ฉ๋๋ก ํ์ฉ์ค
๏ง ์ญ์ Hadoop ์ด์ธ์ ๋ค๋ฅธ ๋์์ ๊ฑฐ์ ์์
23.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted23
๋น ๋ฐ์ดํฐ ๋ถ์์์ R์ ๋ฌธ์ ์
๋ฉ๋ชจ๋ฆฌ ํ๊ณ ์ด์
1) ๋ชจ๋ ๋ฐ์ดํฐ๋ฅผ ๋ฉ๋ชจ๋ฆฌ์ ๋ก๋ฉ ํ ์ฒ๋ฆฌํ๋ ์์ ๋ฐฉ์
โข ff, bigmemory, RevoScaleR
โข GB๊ธ ๋ฐ์ดํฐ ์ฒ๋ฆฌ ๊ฐ๋ฅ. 10GB ์ด์ ๋ฐ์ดํฐ๋ ์ฒ๋ฆฌ
๊ฐ๋ฅํ๋ ๋๋ฌด ๋๋ฆฌ๋ค๋ ๋จ์
2) ๋ถํ์ํ ๋ฐ์ดํฐ ์ ์ฅ์ผ๋ก ์ธํ ๋ฉ๋ชจ๋ฆฌ ๋ถ์กฑ ํ์
โข gc(), rm()
3) 32๋นํธ์์ ํํ ๊ฐ๋ฅํ ์ซ์๋ง์ด ์ฌ์ฉ, 2^31-1
โข R 2.15๋ถํฐ 2^51 ์ด์์ ๋ฒกํฐ ๊ธธ์ด ์ฌ์ฉ ๊ฐ๋ฅ
4) No int64*
โข int64 package from Google
5) ๋ฉ๋ชจ๋ฆฌ ๋จํธํ
โข 64bit ๋จธ์ ์ฌ์ฉ
โข ๋ ๋ง์ ๋ฉ๋ชจ๋ฆฌ
Single Core ์ด์
1) ๋ฉํฐ์ฝ์ด CPU์์ 1์ฝ์ด๋ง ์ฌ์ฉํ๋ค.
2) R 2.14 ๋ถํฐ parallel ํจํค์ง ๊ธฐ๋ณธ ํ์ฌ
โป int64 = 64-bit integers
24.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted24
Why R and Hadoop?
R์ ๋น ๋ฐ์ดํฐ ์ฒ๋ฆฌ ๋ฅ๋ ฅ์ด ํ์ํ๋ฉฐ, Hadoop์ ๊ณ ๊ธ ๋ถ์ ๋ฅ๋ ฅ์ด ํ์
Hadoop - a scalable infrastructure for processing massive amounts of data
โข Storage โ HDFS, HBASE
โข Distributed Computing - MapReduce
R - a statistical programming language
Need for more than counts and averages
Analyze all of the data
๏จ Make it easy for the R programmer to interact with the Hadoop data stores and writeMapReduce
programs
๏จ Run R on a massively distributed system without having to understand the underlying infrastructure
๏จ Statisticians stay focused on the analysis
๏จ Open source
25.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted25
๋น ๋ฐ์ดํฐ ๋ถ์์ ์ํด ํ์ํ ์์?
โข SAS, SPSS์ ๋์
โข ์คํ ์์ค, ์ต์ ๊ธฐ์ ์ ์ฉ
Hadoop
R
R
+
Hadoop
โข ๋์ฉ๋ ๋ฐ์ดํฐ๋ฅผ ์ํ ํ์ผ ์์คํ
โข ๋ถ์ฐ ์ปดํจํ ํ๋ ์์ํฌ
โข ๊ฒ์ฆ๋ ๊ธฐ์
โข R์ ๋ถ์๋ฅ๋ ฅ๊ณผ Visualization ๋ฅ๋ ฅ
โข Hadoop์ ๋์ฉ๋ ๋ฐ์ดํฐ ์ฒ๋ฆฌ ๋ฅ๋ ฅ
26.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted26
R๊ณผ Hadoop์ ์ฎ๋ 5๊ฐ์ง ๋ฐฉ๋ฒ
1. RHadoop โ RHadoop is a great open source solution for R and Hadoop provided by Revolution
Analytics. RHadoop is bundled with four main R packages to manage and analyze the data with
Hadoop framework.
2. RHIPE โ RHIPE is the R and Hadoop Integrated Programming Environment specially designed with
Divide and Recombine (D&R) techniques to analyze the large datasets.
3. ORCH โ ORCH is Oracle R connector for Hadoop. ORCH can be used on the Oracle Big Data Appliance or
on non-Oracle Hadoop clusters.
4. HadoopStreaming โ Hadoopstreaming utilities as R scripts which is R packages available at CRAN. This
R package is developed by David S. Rosenberg with the consideration of making this Hadoop
Streaming more easy as possible for R users.
5. Hadoop Streaming โ Hadoop Streamingis Hadoop utility which allows users to develop and run
MapReduce program in language other than java. Hadoop Streaming is a utility which allows users to
create and run jobs with any executables (e.g. shell utilities) as the mapper and/or the reducer.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted30
R + Hadoop = Data Analytics Heaven
RHadoop is a small, open-source package developed by Revolution Analytics that binds R to Hadoop and
allows for the representation of MapReduce algorithms using R - allowing data scientists access to Hadoopโs
scalability from their favorite language, R. It allows users to write general MapReduce programs, offering the
full power and ecosystem of an existing, established programming language.
์ ๋ ดํ ๋น์ฉ์ผ๋ก
โ๋น ๋ฐ์ดํฐโ๋ฅผ ๋ค๋ฃจ๋ ๋ฐ ๊ฐ์ฅ
๋ง์ด ์ฌ์ฉ๋๋ ๊ธฐ์
ํต๊ณ ์ฐ์ฐ๊ณผ ๊ทธ๋ํฝ์
์ํ ํ๋ก๊ทธ๋จ
์ธ์ด์ด์ ์ํํธ์จ์ด
๊ฐ๋ฐ ํ๊ฒฝ
Put the two together to
provide easy to use R
interfaces for the
distributed
computing Hadoop
environment and you
have one king-hell data
crunching tool for
serious data analytics.
31.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted31
RHadoop์ ๊ตฌ์ฑ
Hadoop์์ ๋ฐ์ดํฐ๋ฅผ ๊ด๋ฆฌํ๊ณ ๋ถ์ํ๋ RHadoop์ ์๋์ ๊ฐ์ ์ธ ๊ฐ์ ํจํค์ง๋ก ๊ตฌ์ฑ๋จ:
HDFS์ ํ์ผ๊ด๋ฆฌ
This is an R package for providing
all Hadoop HDFS access to R. All
distributed files can be managed
with R functions.
MR ์ธํฐํ์ด์ค
This is an R package for providing
Hadoop MapReduce interfaces to R.
With the help of this package, the
Mapper and Reducer can easily be
developed.
Hbase ๋ฐ์ดํฐ๋ฒ ์ด์ค ๊ด๋ฆฌ
This is an R package for handling
data at HBase distributed
database through R.
32.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted32
R Client
R
Map
or
Reduce
Job Tracker
Task Node
HDFS
HBASE
Avro(Thrift)
RHadoop์ ์ํคํ ์ฒ
rhdfs = R + HDFS
functions providing file management of
the HDFS from within R
rmr = R + MapReduce
functions providing Hadoop MapReduce
functionality in R
rhbase = R + Hbase
functions providing database management
for the HBase distributed database from
within R
avro
read and write files in avro format
plyrmr
higher level plyr-like data processing for
structured data, powered by rmr
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted34
rhdfs
๏ง Manipulate HDFS directly from R
๏ผ Access HDFS from R
๏ผ Read from HDFS to R dataframe
๏ผ Write from R dataframe to HDFS
๏ง Mimic as much of the HDFS Java API as
possible
Examples:
โข Read a HDFS text file into a data frame.
โข Serialize/Deserialize a model to HDFS
โข Write an HDFS file to local storage
โข rhdfs/pkg/inst/unitTests
rhdfs/pkg/inst/examples
์ฃผ์ ๊ธฐ๋ฅ
File Manipulations
hdfs.copy, hdfs.move, hdfs.rename, hdfs.delete,
hdfs.rm, hdfs.del, hdfs.chown, hdfs.put, hdfs.get
File Read/Write
hdfs.file, hdfs.write, hdfs.close, hdfs.flush,
hdfs.read, hdfs.seek, hdfs.tell, hdfs.line.reader,
hdfs.read.text.file
Directory
hdfs.dircreate, hdfs.mkdir
Utility
hdfs.ls, hdfs.list.files, hdfs.file.info, hdfs.exists
Initialization
hdfs.init, hdfs.defaults
35.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted35
rhbase
๏ง Manipulate HBASE tables and their
content : Access and change data within
HBase
๏ง Uses Thrift C++ API as the mechanism to
communicate to HBASE
Examples
โข Create a data frame from a collection of rows
and columns in an HBASE table
โข Update an HBASE table with values from a data
frame
โข rhbase/pkg/inst/unitTests
์ฃผ์ ๊ธฐ๋ฅ
Table Manipulation
hb.new.table, hb.delete.table, hb.describe.table,
hb.set.table.mode, hb.regions.table
Row Read/Write
hb.insert, hb.get, hb.delete, hb.insert.data.frame,
hb.get.data.frame, hb.scan
Utility
hb.list.tables
Initialization
hb.defaults, hb.init
36.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted36
rmr2
๏ง Enables writing MapReduce jobs using R
(Writing MapReduce programs in R )
๏ง Ability to parallelize algorithms
๏ง Ability to use big data sets without
needing to sample data
๏ง Mapreduce(input, output, map, reduce, โฆ )
๏ง Reduces takes a key and a collection of
values which could be vector, list, data
frame or matrix
rmr์ MapReduce ๊ธฐ๋ฅ
mapreduce (input, output, map, reduce, โฆ)
<<์ํ ์ฝ๋ โ Word Count>>
wc.map =
function(., lines) {
keyval(
unlist(!
strsplit(
x = lines,
split = pattern)),
1)}
wc.reduce =
function(word, counts ) {!
keyval(word, sum(counts))}!
mapreduce(
Input = input ,
output = output,
input.format = "text",
map = wc.map,
reduce = wc.reduce,
combine = T)}
37.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted37
37
๏ง A way to access big data sets
๏ง A simple way to write parallel
programs โ everyone will have to
๏ง Very R-like, building on the functional
characteristics of R
๏ง Just a library
For Programmers,
๏ง Much simpler than writing Java
๏ง Not as simple as Hive, Pig at what they
do, but more general
๏ง Great for prototyping, can transition to
production -- optimize instead of
rewriting! Lower risk, always
executable.
For MapReduce Developers
rmr for,
38.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted38
Rhadoop๊ณผ Graph engine
There are two basic types of graph engines:
1) Graph databases providing real-time, traversal-based algorithms over linked-list graphs represented on
a single-server (vendors include Neo4j, OrientDB, DEX, and InfiniteGraph).
2) Batch-processing using vertex-centric message passing within a graph represented across a cluster of
machines. (Hama, Golden Orb, Giraph, and Pregel).
With Hadoop, the results presented are via Hadoop (HDFS + MapReduce). Moreover, instead of developing
the MapReduce algorithms in Java, the R programming language is used.
A multi-machine graph engine is required. While Hadoop is not a graph engine, a graph can be represented in
its distributed HDFS file system and processed using its distributed processing MapReduce framework.
The graph generated previously is loaded up in R and a count of its vertices and edges is conducted. Next,
the graph is represented as an edge list. An edge list (for a single-relational graph) is a list of pairs, where
each pair is ordered and denotes the tail vertex id and the head vertex id of the edge. The edge list can be
pushed to HDFS using RHadoop.
39.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted39
RHadoop์ ์ฅ์ ๊ณผ ํ๊ณ
๏ง ๊ธฐ์กด์ ๋ฐ์ดํฐ ๋ถ์๊ฐ๋ Big Data
์ฒ๋ฆฌ ๊ฐ๋ฅ
๏ง ๋ถ์๊ฒฐ๊ณผ ํฅ์
-๊ฐ๋จํ ๋ชจ๋ธ + ๋์ฉ๋ ๋ฐ์ดํฐ
-๋ณต์กํ ๋ชจ๋ธ + ์ ์ ๋ฐ์ดํฐ
๏ง ๋ฒค๋ ์ข ์์ฑ ํํผ
์ฅ์ ํ๊ณ
๏ง Requires installation of R on all
TaskTracker nodes
๏ง Does not automatically parallelize
algorithms
๏ง Different slot/memory configuration
recommended to leave memory and
CPU resources for R.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted41
์ค์นํ๊ธฐ
R ์ค์น
http://cran.nexr.com/(ํ๊ตญ CRAN)
R Studio ์ค์น
https://www.rstudio.com/
โข R์ ์ํ ํตํฉ๊ฐ๋ฐํ๊ฒฝ(IDE)
โข R์ ์ฌ์ฉํ๊ธฐ ์ํ ๋ค์ํ ๊ธฐ๋ฅ๊ณผ ํธ์์ฑ ์ ๊ณต : ์ฝ๋ ์ง์ ์คํ,
๊ตฌ๋ฌธ๊ฐ์กฐ, ๊ดํธ ์๋์ ๋ ฅ์ง์, ๋ช ๋ น์ด ์์ฑ, ๋ค์ํ ๋จ์ถํค,
๋ฐ์ดํฐ ๋ณด๊ธฐ ๋ฐ ๊ฐ์ ธ์ค๊ธฐ, ๊ทธ๋ํฝ ์กฐ์, ํ๋ก์ ํธ ๊ด๋ฆฌ, ๋ฒ์ ๊ด๋ฆฌ
๋ฑ
42.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted42
์ค์ต
43.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted43
[์ฐธ๊ณ ] KNUG ์๊ฐ
KRUG (Korean R Users Group)
๏ผ GNU์ ์ฒ ํ์ ์ ๊ฐํ์ฌ,
๏ผ R์ ํ๊ตญ์ด ์ฌ์ฉ์๊ฐ ์ฌ๋ฐ๋ฅด๊ณ ์ฝ๊ฒ ์ฌ์ฉ ๋ ์ ์๋๋ก ๋ฌธ์๋ฅผ ๋ฒ์ญํ๊ณ ์ง์๊ณผ ๊ธฐ์ ์ ๊ณต์
ํ๋ ์ฌ์ฉ์ ๋ชจ์
๏ผ 2007๋ 1์๋ถํฐ ๊ณต์์ ์ผ๋ก ํ๋ํ ๋น์๋ฆฌ ๋ชจ์
Offline ํ๋ : Meetup์ ํตํ ๊ธฐ์ ๊ต๋ฅ
http://www.openstatistics.net
http://www.r-project.kr/
๋์ธํ๋ ฅ : ๋ฌธ์/White paper/Blog
์ ๋ฒ์ญ/๋ฐฐํฌ ๊ถ๋ฆฌ
Online ํ๋ : ๋ฌธ์๋ฒ์ญ,๊ธฐ์ ๊ณต์ , Q&A
R User Conference ๊ฐ์ต
44.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted44
R ํ๋ก์ ํธ ๊ณต์ ์ฌ์ดํธ
- http://www.r-project.org
ํ๊ตญ R ์ฌ์ฉ์ ๊ทธ๋ฃน ์ฌ์ดํธ(KRUG)
- http://www.r-project.kr (facebook: KRUG)
- http://ihelp.r-forge.r-project.org/
ํ๊ตญ์ CRAN ์ ๊ณต๊ธฐ๊ด
- NexR: http://cran.nexr.com/
- ์ค์๋ํ๊ต: http://biostat.cau.ac.kr/CRAN/
- ๋ค์(Daum): http://ftp.daum.net/CRAN/ โ
RTechCenter
RHadoop Open source project:
- https://github.com/RevolutionAnalytics/RHadoop
/wiki
Resources
R ์ฐธ๊ณ ์ฌ์ดํธ
- http://www.r-bloggers.com
- http://stackoverflow.com
- http://stats.stackexchange.com
- http://www.inside-r.org/
- http://www.r-statistics.com/
- http://support.rstudio.org/
- http://quora.com
Revolution R Enterprise:
- bit.ly/Enterprise-R
Cloudera CDH:
- http://www.cloudera.com/hadoop/
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted46
Options for R on Hadoop
47.
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted47
Options for R on Hadoop
๏ง SQL Access from R
RODBC/RJDBC
๏ง Broad access to Hive and HDFS
RHive
๏ง Tight integration with core
Hadoop components
RHadoop
Focus
Integration
Ease
Benefits
Limitations
๏ง Low impact on existing R scripts
leveraging other DB packages
๏ง Not required to install Hadoop
configuration/binaries on client
machines
๏ง Install Hortonworks Hive ODBC
driver
๏ง Install Hive Libraries
๏ง Parallelism limited to Hive
๏ง Result set size
๏ง Wide range of features expressed
through HQL
- rhive-apply R Distributed apply
function using HQL
๏ง Requires Hadoop binaries, libraries,
and configuration files on client
machines
๏ง Uses Java DFS Client and
HiveServer
๏ง Requires heavy client deployment
๏ง Dependent on HiveServer, and canโt
be used with HiveServer2
๏ง Ability to run R on a massively
distributed system
๏ง Ability to work with full data sets
instead of sample sets
Additional Information
https://github.com/RevolutionAnalytic
s/RHadoop/wiki
ยฉ Copyright 2014Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. HP Restricted
C03934969, January 2014
Thankyou