Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
Hadoop Operations
Search
Sponsored
·
SiteGround - Reliable hosting with speed, security, and support you can count on.
→
Marc Cluet
June 09, 2013
Technology
130
0
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
Hadoop Operations
Marc Cluet
June 09, 2013
More Decks by Marc Cluet
See All by Marc Cluet
FOSDEM'14 - Autoscaling Best Practices
lynxman
1
130
A metadata ocean in Chef and Puppet
lynxman
0
67
Rackspace Hack Night - Vagrant & Packer
lynxman
0
160
Innovation in the Cloud - Rackspace Zurich Event
lynxman
0
110
Introduction to DevOps - Rackspace Tech Night
lynxman
1
88
Introduction To Hadoop
lynxman
1
120
SSH That Wonderful Thing
lynxman
1
100
Networking & DNS 101
lynxman
0
110
Juju and Puppet - Rapid Harmonious Deployment
lynxman
0
110
Other Decks in Technology
See All in Technology
Kiro5兄弟のいまどきのセキュリティ基礎知識
kentapapa
0
390
Dylib Hijacking on macOS: Dead or Alive?
patrickwardle
0
130
OAuth SPIFFE Client Authentication(OAuth/OIDC Numa (Immersion) Workshop 2026)
oidfj
PRO
0
350
Introduction to Sansan for Engineers / エンジニア向け会社紹介
sansan33
PRO
6
77k
アクセスキーが漏れた日にやるべきこと- 無効化の先にある本当の対応
kazzpapa3
0
360
atproto spaces概要
yamarten
0
110
500名弱規模の組織のPythonプロジェクト(dbt) をどう管理するか?
hiracky16
0
360
Windows の互換機能 - 古いプログラムはなぜ動作できるのか
murachiakira
PRO
0
130
医療の現場を変革に挑戦した半年間の軌跡 - PythonとAIで現場を変える / From Code to Care
soudai
PRO
1
640
Distributed Transactions Under Fire: Building a Zero-Oversell Flash Sale Platform with Amazon Aurora DSQL
yama3133
1
110
Master Dataグループ紹介資料
sansan33
PRO
1
4.8k
LLMアプリ、 雰囲気で運用してませんか? 〜LLMOpsの現在地〜
taka_aki
1
560
Featured
See All Featured
Deep Space Network (abreviated)
tonyrice
0
270
コードの90%をAIが書く世界で何が待っているのか / What awaits us in a world where 90% of the code is written by AI
rkaga
63
45k
Game over? The fight for quality and originality in the time of robots
wayneb77
1
260
A designer walks into a library…
pauljervisheath
211
24k
New Earth Scene 8
popppiees
3
2.5k
Leveraging Curiosity to Care for An Aging Population
cassininazir
1
480
Large-scale JavaScript Application Architecture
addyosmani
515
110k
jQuery: Nuts, Bolts and Bling
dougneiner
66
8.6k
YesSQL, Process and Tooling at Scale
rocio
174
15k
16th Malabo Montpellier Forum Presentation
akademiya2063
PRO
0
360
Applied NLP in the Age of Generative AI
inesmontani
PRO
4
2.4k
Lightning Talk: Beautiful Slides for Beginners
inesmontani
PRO
2
650
Transcript
Marc Cluet – Lynx Consultants How Hadoop Works
What we’ll cover? ¡ Understand Hadoop in detail ¡
See how Hadoop works operationally ¡ Be able to start asking the right questions from your data Lynx Consultants © 2013
Hadoop Distributions ¡ Cloudera CDH ¡ Hortonworks ¡
MapR Lynx Consultants © 2013
Hadoop Components ¡ HDFS ¡ Hbase ¡ MapRed
¡ YARN Lynx Consultants © 2013
Hadoop Components ¡ HDFS § Hadoop Distributed File System
§ Everything sits on top of it § Has 3 copies by default of every block ¡ Hbase ¡ MapRed ¡ YARN Lynx Consultants © 2013
Hadoop Components ¡ HDFS ¡ Hbase § Hadoop
Schemaless Database § Key value Store § Sits on top of HDFS ¡ MapRed ¡ YARN Lynx Consultants © 2013
Hadoop Components ¡ HDFS ¡ Hbase ¡ MapRed
§ Hadoop Map/Reduce § Non-‐pluggable, archaic § Requires HDFS for temp storage ¡ YARN Lynx Consultants © 2013
Hadoop Components ¡ HDFS ¡ Hbase ¡ MapRed
¡ YARN § Hadoop Map/Reduce version 2.0 § Pluggable, you can add your own § Fast and not so much memory hungry Lynx Consultants © 2013
Hadoop Component Breakdown ¡ All these components divide themselves in
§ client/server § master/slave scenarios ¡ We will now check each individual component breakdown Lynx Consultants © 2013
Hadoop Components Breakdown ¡ HDFS § Master Namenode
▪ Keeps track of all file allocation on Datanodes ▪ Rebalances data if one of the namenodes goes down ▪ Is Rack aware § Secondary Namenode ▪ Does cleanup services for the namenode ▪ Not necessarily two different servers § Datanode ▪ Stores the data ▪ Good to have not RAID disks for extra I/O speed Lynx Consultants © 2013
Hadoop Components Breakdown ¡ HDFS § How to access
▪ Client can connect with hadoop client to hdfs://namenode:8020 ▪ Supports all basic Unix commands § Configuration files ▪ /etc/hadoop/conf/core-‐site.xml ▪ Defines major configuration as hdfs namenode and default parameters ▪ /etc/hadoop/conf/hdfs-‐site.xml ▪ Defines configuration specific to namenode or datanode on file locations ▪ /etc/hadoop/conf/slaves ▪ Defines the list of servers that are available in this cluster Lynx Consultants © 2013
Hadoop Components Breakdown ¡ Hbase § Master ▪
Controls the Hbase cluster, knows where the data is allocated and provides a client listening socket using Thrift and/or a RESTful API § Regionserver ▪ Hbase node, stores some of the information in one of the regions, it’d be equivalent to sharding § Thrift / REST ▪ Interface to connect to HBase Lynx Consultants © 2013
Hadoop Components Breakdown ¡ Hbase § How to access
▪ Through the Hbase client (using Thrift) ▪ Through the RESTful API § Configuration files ▪ /etc/hbase/conf/hbase-‐site.xml ▪ Defines all the basic configuration for accessing hbase ▪ /etc/hbase/conf/hbase-‐policy.xml ▪ Defines all the security (ACL) and all the hbase memory tweaks ▪ /etc/hbase/conf/regionservers ▪ List all the regionservers available to this cluster Lynx Consultants © 2013
Hadoop Components Breakdown ¡ MapRed § JobTracker ▪
Creates the Map/Reduce jobs ▪ Stores all the intermediate data ▪ Keeps track of all the previous results through the HistoryServer § TaskTracker ▪ Executed Tasks related to the Map/Reduce job ▪ Very CPU and memory intensive ▪ Stores intermediate results which then are pushed to JobTracker Lynx Consultants © 2013
Hadoop Components Breakdown ¡ MapRed § How to access
▪ Through the Hadoop Client ▪ Through any MapRed client like Pig or Hive ▪ Own Java code § Configuration files ▪ /etc/hadoop/conf/mapred-‐site.xml ▪ Defines how to contact this MapRed Cluster ▪ /etc/hadoop/conf/mapred-‐queue-‐acls.xml ▪ Defines ACL structure for accessing MapRed, normally not necessary ▪ /etc/hadoop/conf/slaves ▪ Defines the list of TaskTrackers in this cluster Lynx Consultants © 2013
Hadoop Components Breakdown ¡ YARN § Same structure as
MapRed (lives on top of it) § Configuration files ▪ /etc/hadoop/conf/yarn-‐site.xml ▪ All required configuration for YARN Lynx Consultants © 2013
Hadoop Cluster Breakdown ¡ Namenode Server § HDFS Namenode
§ Hbase Master ¡ Secondary Namenode Server § HDFS Secondary Namenode ¡ JobTracker Server § MapRed JobTracker § MapRed History Server Lynx Consultants © 2013
Hadoop Cluster Breakdown ¡ Datanode Server § HDFS Datanode
§ Hbase RegionServer § MapRed TaskTracker Lynx Consultants © 2013
Hadoop Hardware Requirements ¡ Namenode Server § Redundant power
supplies § RAID1 Drives § Enough memory (16Gb) ¡ Secondary Namenode Server § Almost none Lynx Consultants © 2013
Hadoop Hardware Requirements ¡ Jobtracker Server § Redundant power
supplies § RAID1 Drives § Enough memory (16Gb) ¡ Datanode Server § Lots of cheap disk (no RAID) § Lots of memory (32Gb) § Lots of CPU Lynx Consultants © 2013
Hadoop Default Ports ¡ HDFS § 8020: HDFS Namenode
§ 50010: HDFS Datanode FS transfer ¡ MapRed § No defaults ¡ Hbase § 60010: Master § 60020: Regionserver Lynx Consultants © 2013
Hadoop HDFS Workflow Lynx Consultants © 2013
Hadoop MapRed Workflow Lynx Consultants © 2013
Hadoop MapRed Workflow Lynx Consultants © 2013
Flume ¡ Transports streams of data from point A to
point B ¡ Source § Where the data is read from ¡ Channel § How the data is buffered ¡ Sink § Where the data is written Lynx Consultants © 2013
Flume ¡ Flume is fault tolerant ¡ Sources are
pointer kept § With some exceptions, but most sources are in a known state ¡ Channels can be fault tolerant § Channel written to disk can recover from where it left ¡ Sinks can be redundant § More than one sink for the same data § Data is serialised and deduplicated using AVRO Lynx Consultants © 2013
Flume Lynx Consultants © 2013
Flume ¡ Configuration files § /etc/flume-‐ng/conf/flume.conf ▪ Defines
the agent configuration with source, channel, sink Lynx Consultants © 2013
Flume Lynx Consultants © 2013
Hadoop Recommended Reads Lynx Consultants © 2013
Hadoop References ¡ Hadoop § http://hadoop.apache.org/docs/stable/cluster_setup.html § http://rc.cloudera.com/cdh/4/hadoop/hadoop-‐yarn/hadoop-‐yarn-‐site/
ClusterSetup.html § http://pig.apache.org/docs/r0.7.0/setup.html § http://wiki.apache.org/hadoop/NameNodeFailover ¡ Hbase § http://hbase.apache.org/book/book.html ¡ Flume § http://archive.cloudera.com/cdh4/cdh/4/flume-‐ng/ FlumeUserGuide.html Lynx Consultants © 2013
Questions? Lynx Consultants © 2013