apache hadoop crash course

88
Apache Hadoop Crash Course Rafael Coss Data Evangelist @racoss #FutureOfData

Upload: dataworks-summithadoop-summit

Post on 16-Apr-2017

397 views

Category:

Technology


0 download

TRANSCRIPT

ApacheHadoopCrashCourseRafaelCossDataEvangelist@racoss#FutureOfData

2 ©HortonworksInc.2011– 2016.AllRightsReserved

AgendaFutureofData

TraditionalDataArchitectures

What’sApacheHadoop?

DataAccesswithHadoop

LabIntro

3 ©HortonworksInc.2011– 2016.AllRightsReserved

CustomersarebuildingModernDataApplicationstotransformtheirindustries–renovatingtheirITarchitecturesandinnovatingwiththeirDatainMotionorDataatResttopoweractionableintelligence.

SocialMapping

PaymentTracking

FactoryYields

DefectDetection

CallAnalysis MachineData

ProductDesign M&A

DueDiligence

NextProductRecs

CyberSecurity

RiskModeling

AdPlacement

ProactiveRepair

DisasterMitigation

InvestmentPlanning

InventoryPredictions

CustomerSupport

SentimentAnalysis

SupplyChain

AdPlacement

BasketAnalysis Segments

Cross-Sell

CustomerRetention

VendorScorecards

OptimizeInventories

OPEXReduction

MainframeOffloads

HistoricalRecords

DataasaService

PublicData

Capture

FraudPrevention

DeviceDataIngest

RapidReporting

DigitalProtection

3 © HortonworksInc.2011– 2016.AllRightsReserved

FutureofData

5 ©HortonworksInc.2011– 2016.AllRightsReserved

INTERNETOF

ANYTHING

TheFutureofDataisaboutactionableintelligencederivedfromaconstantlyconnectedsociety,withinteractivesecureaccesstorichdatasetscomingfromtheInternetofAnything

6 ©HortonworksInc.2011– 2016.AllRightsReserved

TheFutureofDataActionableIntelligence

D ATA I N M O T I O N

STORAG

ESTORA

GE

GROUP2GROUP1

GROUP4GROUP3

D ATA AT R E S T

INTERNETOF

ANYTHING

ConnectedDataArchitectureAcrossyourDataPlane

ispoweringActionableIntelligence

Anyandalldatafromsensors,machines,

geolocation,clicks,files,social.

Securepoint-to-pointandbi-directionaldataflows

Collectandcuratealldata.

DataPowersHighwaySafety

8 ©HortonworksInc.2011– 2016.AllRightsReserved8 ©HortonworksInc.2011– 2016.AllRightsReserved

ConnectedDataDrivestheConnectedCarInsurancePremiums

Warranties

RecallsPricingModels

DesignInnovation

AutonomousDriving

ConnectedCity Infotainment

Sensors

ScheduledMaintenance

PredictiveMaintenance

RouteOptimization

INSURANCECOMPANIES

GOVERNMENTAGENCIES

INFOTAINMENTPROVIDERS

SOFTWARECOMPANIES

AUTOMAKERS

9 ©HortonworksInc.2011– 2016.AllRightsReserved

NewDataParadigmOpensUpNewOpportunity

2.8zettabytesin2012

44zettabytesin2020

N E W

1 zettabyte (ZB) = 1 million petabytes (PB); Sources: IDC, IDG Enterprise, and AMR Research

Clickstream

ERP,CRM,SCM

Web&social

Geolocation

InternetofThings

Serverlogs

Files,emails

Transformeveryindustryviafullfidelityofdataandanalytics

Opportunity

T R A D I T I O N A L

LAGGARDS

LEADERS

AbilitytoConsumeData

EnterpriseBlindSpot

10 ©HortonworksInc.2011– 2016.AllRightsReserved

Whatdisruptedthedatacenter?

?

Data?

11 ©HortonworksInc.2011– 2016.AllRightsReserved

ModernDataApplications

Polygot Persistence

SQLNoSQL

NewSQLSearch

Graph

At-Rest In-Motion

AnalyticsDataVariety

Integration

DataLake Federation

OptimizationStorage,Compute

DistributedComputing

CommodityHardware

Cloud

HybridDistributedComputing

12 ©HortonworksInc.2011– 2016.AllRightsReserved

TraditionalDataArchitectures

13 ©HortonworksInc.2011– 2016.AllRightsReserved

Next Generation AnalyticsIterative & ExploratoryData is the structure

IT TeamDelivers DataOn Flexible

Platform

BusinessUsers

Explore andAsk Any Question

Analyze ALL Available Information

Whole population analytics connects the dots

Traditional AnalyticsStructured & Repeatable

Structure built to store data

BusinessUsers

DetermineQuestions

IT TeamBuilds System

To AnswerKnown Questions

13

Available Information

AnalyzedInformation

Capacity constrained down sampling of available information

Carefully cleanse all information before any analysis

AnalyzedInformation

Analyze information as is & cleanse as needed

AnalyzedInformation

ModernDataApplications

14 ©HortonworksInc.2011– 2016.AllRightsReserved

Next Generation AnalyticsIterative & ExploratoryData is the structure

Traditional AnalyticsStructured & Repeatable

Structure built to store data

14

?AnalyzedInformation

Question

DataAnswer

Hypothesis

StartwithhypothesisTestagainstselecteddata

Data leads the way Explore all data, identify correlations

Data

Correlation

All Information

Exploration

Actionable Insight

Analyzeafterlanding… Analyzeinmotion…

ModernDataApplicationsHasTwoThemes

15 ©HortonworksInc.2011– 2016.AllRightsReserved

Next Generation AnalyticsIterative & ExploratoryData is the structure

Traditional AnalyticsStructured & Repeatable

Structure built to store data

15

?AnalyzedInformation

Question

DataAnswer

Hypothesis

StartwithhypothesisTestagainstselecteddata

Data leads the way Explore all data, identify correlations

Data

Correlation

All Information

Exploration

Actionable Insight

ModernDataApplicationsHasTwoThemes

4. AppDev5. Delivery

1. Requirements2. DataSourcing3. ETL

4. AppDev5. LandMoreData6. Iterate

1. Landrawdata2. DataScience/ML3. DataSourcing4. Requirements

16 ©HortonworksInc.2011– 2016.AllRightsReserved

RDBMS

Sales

NoSQL

Unstructured

Visualization&Dashboards

BusinessAnalytics

DataMarts

DataMarts Archive

StatisticsOLAP

EDW

FileServer

ClickstreamLogs

Web&SocialLogs

AudioVideo

LogsLogs

Logs

Geolocation

JSON

ETL

POS CRM ERP

ECM

Filter

AppServer

MessageBus

Documents

17 ©HortonworksInc.2011– 2016.AllRightsReserved

RDBMS

Sales

NoSQL

Unstructured

Visualization&Dashboards

BusinessAnalytics

DataMarts

DataMarts Archive

StatisticsOLAP

EDW

FileServer

ClickstreamLogs

Web&SocialLogs

AudioVideo

LogsLogs

Logs

Geolocation

JSON

ETL

POS CRM ERP

ECM

Filter

AppServer

MessageBus

Documents

à Tooexpensiveandslowasdatagrowthkeepsaccelerating

à Tooslowtogetthedatapreparedforanalytics

à Analyticsisonlyleveragingalimiteddataset

à Colddatabecomesarchivedandisnolongerusableforanalytics

à DataingestisrigidandslowfornewIoAT datatypes

à Limitedrealtimeinsights

TraditionalDataArchitectureChallengeswithBigData

18 ©HortonworksInc.2011– 2016.AllRightsReserved

RDBMS

Sales

NoSQL

Unstructured

Visualization&Dashboards

BusinessAnalytics

DataMarts

DataMarts Archive

StatisticsOLAP

EDW

FileServer

ClickstreamLogs

Web&SocialLogs

AudioVideo

LogsLogs

Logs

Geolocation

JSON

ETL

POS CRM ERP

ECM

Filter

AppServer

MessageBus

Documents

What’sApacheHadoop?

20 ©HortonworksInc.2011– 2016.AllRightsReserved

HadoopArchitecture

DataAccessEngines

DistributedReliableStorage

DistributedComputeFrameworkResourceManagement,DataLocalityDataOperatingSystem

Batch Interactive Real-time

Governance&

IntegrationSecurity

Applications

DeployAnywhere

21 ©HortonworksInc.2011– 2016.AllRightsReserved

HadoopDataPlatformArchitecture

StoreandprocessallofyourCorporateDataAssets

YARN:DataOperatingSystem

DATA MANAGEMENT

Providelayeredapproachto

securitythroughAuthentication,Authorization,Accounting,andDataProtection

SECURITY

Accessyourdatasimultaneouslyinmultipleways(batch,interactive,real-time)

DATA ACCESS

Loaddataandmanageaccording

topolicy

GOVERNANCE & INTEGRATION

ENTERPRISEMGMT&SECURITY

EmpowerexistingoperationsandsecuritytoolstomanageHadoop

PRESENTATION&APPLICATION

Enablebothexistingandnewapplicationtoprovidevaluetotheorganization

Providedeploymentchoiceacrosson-premise,appliance,virtualized,cloud

DEPLOYMENTOPTIONS

Deployandeffectivelymanagetheplatform

OPERATIONS

22 ©HortonworksInc.2011– 2016.AllRightsReserved runson

ETL

RDBMSImport/Export

DistributedStorage&ProcessingFramework

SecureNoSQL DB

SQLonHBase

NoSQL DB

WorkflowManagement

SQL

StreamingDataIngestion

ClusterSystemOperations

SecureGateway

DistributedRegistry

ETL

Search&Indexing

EvenFasterDataProcessing

DataManagement

MachineLearning

HadoopEcosystem

23 ©HortonworksInc.2011– 2016.AllRightsReserved

OpenEnterpriseHadoopCapabilities

YARN : Data Operating System

DATA ACCESS SECURITYGOVERNANCE & INTEGRATION OPERATIONS

1 ° ° ° ° ° ° ° ° °

° ° ° ° ° ° ° ° ° °

°

N

Data Lifecycle & Governance

FalconAtlas

AdministrationAuthenticationAuthorizationAuditingData Protection

RangerKnoxAtlasHDFSEncryptionData Workflow

SqoopFlumeKafkaNFSWebHDFS

Provisioning, Managing, & Monitoring

AmbariCloudbreakZookeeper

Scheduling

Oozie

Batch

MapReduce

Script

Pig

Search

Solr

SQL

Hive

NoSQL

HBaseAccumuloPhoenix

Stream

Storm

In-memory

Spark

Others

ISV Engines

Tez Tez Slider Slider

DATA MANAGEMENT

HortonworksDataPlatform

DeploymentChoiceLinux Windows On-Premise Cloud

HDFS Hadoop Distributed File System

24 ©HortonworksInc.2011– 2016.AllRightsReserved

HORTONWORKSDATAPLATFORM

DATAMGMT

HDP2.2Dec2014

HDP2.1April2014

HDP2.0Oct2013

HDP2.2Dec2014

HDP2.1April2014

HDP2.0Oct2013

2.2.0

2.4.0

2.6.0

OngoingInnovationinApache

HDFSYARNMapReduceHadoopCore

WhatisApacheHadoop?

Yahoo!2006

HortonworksOct2011

Yahoo!startfocusonmultipleHadoopapps&clustersContributesHadooptoApache

2008

HDP1.0Oct2012

ApacheHadoopv2YARN

GooglepublishesGFS&MapReduce papers2004-2005

HDP2.4March2016 2.7.1

HDP2.2Dec2014HDP2.3July2015 2.7.1

HDP 2.5Aug2016

2.7.3

25 ©HortonworksInc.2011– 2016.AllRightsReserved

Apache Hadoop = Storage + Compute

storage storage

storage storage

HadoopDistributedFileSystem(HDFS)

CPU RAM

YetAnotherResourceNegotiator(YARN)

26 ©HortonworksInc.2011– 2016.AllRightsReserved

`

+ /directory/structure/in/memory.txt

Resource management + schedulingDisk, CPU, Memory

CoreNameNode

HDFS

ResourceManagerYARN

Hadoop daemon

User application

NN

RM

DataNodeHDFS

NodeManagerYARN

Worker Node

27 ©HortonworksInc.2011– 2016.AllRightsReserved

Hadoop Distributed File System (HDFS)

Fault Tolerant Distributed Storage• Dividefilesintobigblocksanddistribute3copiesrandomly acrossthecluster• ProcessingDataLocality

• NotJuststoragebutcomputation

10110100101001001110011111100101001110100101001011001001010100110001010010111010111010111101101101010110100101010010101010101110010011010111010

0

Logical File

1

2

3

4

Blocks

1

Cluster

1

1

2

22

3

3

34

44

28 ©HortonworksInc.2011– 2016.AllRightsReserved

HDFSStorageArchitecture- Before

Before• DataNode isasinglestorage• Storageisuniform- OnlystoragetypeDisk

• Storagetypeshiddenfromthefilesystem

Alldisksassinglestorage

29 ©HortonworksInc.2011– 2016.AllRightsReserved

CloudStorage

HDFSStorageArchitecture- Now

New Architecture• DataNode isacollectionofstorages• Supportdifferenttypesofstorages

– Disk,SSDs,Memory

Block Storage Policies– DescribeshowtostoredatablocksinHDFS

Collectionoftierstorage

30 ©HortonworksInc.2011– 2016.AllRightsReserved

It Looks Like a File System

31 ©HortonworksInc.2011– 2016.AllRightsReserved

Batch Processing in HadoopMapReduceBatch Access to DataOriginal data access mechanism for Hadoop

• FrameworkMadefordevelopingdistributedapplicationstoprocessvastamountsofdatain-parallelonlargeclusters

• ProvenReliableinterfacetoHadoopwhichworksfromGBtoPB.But,batchoriented– Speedisnotit’sstrongpoint.

• EcosystemPortedtoHadoop2torunonYARN.SupportsoriginalinvestmentsinHadoopbycustomersandpartnerecosystem.

DataNode1

Mapper

Dataisshuffledacrossthenetwork

&sorted

MapPhase Shuffle/Sort ReducePhase

MapReduce JobLifecycle

SayingthatMapReduceisdeadispreposterous- Wouldlimitsustoonlynewworkloads- ALLHadoop clustersusemapreduce

- ProvenatEnterpriseScale

DataNode2

Mapper

DataNode3

Mapper

DataNode1

Reducer

DataNode2

Reducer

DataNode3

Reducer

YARN:DataOperatingSystem

Interactive Real-TimeBatch

32 ©HortonworksInc.2011– 2016.AllRightsReserved

What is MapReduce?Break a large problem into sub-solutionsMap

• Iterate over a large # of records

• Extract something of interest fromeach record

Shuffle

• Sort Intermediate results

Reduce

• Aggregate, summarize, filter or transform intermediate results

• Generate final output

MapProcess

MapProcess

MapProcess

MapProcess

Data

DataData

Data

DataData

DataData

DataData

Data

DataData MapProcess

ReduceProcess

ReduceProcess

Data

Read&ETL

Shuffle&Sort Aggregation

Data

DataData

Data

Data

Data

Data

Data

33 ©HortonworksInc.2011– 2016.AllRightsReserved

1st GenHadoop:CostEffectiveBatchatScale

HADOOP1.0BuiltforWeb-ScaleBatchApps

SingleAppBATCH

HDFS

SingleAppINTERACTIVE

SingleAppBATCH

HDFS

SiloscreatedfordistinctusecasesSingleApp

BATCH

HDFS

SingleAppONLINE

34 ©HortonworksInc.2011– 2016.AllRightsReserved

HadoopemergedasfoundationofnewdataarchitectureApacheHadoopisanopensourcedataplatformformanaginglargevolumesofhighvelocityandvarietyofdata

• BuiltbyYahoo!tobetheheartbeatofitsad&searchbusiness

• DonatedtoApacheSoftwareFoundationin2005withrapidadoptionbylargewebproperties&earlyadopterenterprises

• Incrediblydisruptivetocurrentplatformeconomics

TraditionalHadoopAdvantages

ü Managesnewdataparadigmü Handlesdataatscaleü Costeffectiveü Opensource

TraditionalHadoopHadLimitations

Batch-onlyarchitectureSinglepurposeclusters,specificdatasetsDifficulttointegratewithexistinginvestmentsNotenterprise-grade

Application

StorageHDFS

Batch ProcessingMapReduce

35 ©HortonworksInc.2011– 2016.AllRightsReserved

YARNextendsHadoopintodatacenterleaders

YARNThe Architectural Center of Hadoop

• Common data platform, many applications

• Support multi-tenant access & processing

• Batch, interactive & real-time use cases

• Supports 3rd-party ISV tools

(ex. SAS, Syncsort, Actian, etc.)

YARN Ready Applications Facilitates ongoing innovation and enterprise adoption via ecosystem of new and existing “YARN Ready” solutions

YARN : Data Operating System

BATCH, INTERACTIVE & REAL-TIME DATA ACCESS

1 ° ° ° ° ° ° ° ° °

° ° ° ° ° ° ° ° ° °

°

N

HDFS Hadoop Distributed File System

DATA MANAGEMENT

Batch

MapReduce

Script

Pig

Search

Solr

SQL

Hive

NoSQL

HBaseAccumuloPhoenix

Stream

Storm

In-memory

Spark

Others

ISV Engines

Tez Tez Slider Slider

36 ©HortonworksInc.2011– 2016.AllRightsReserved

WhatdoesiOS 6andWindows3.1haveincommon?

37 ©HortonworksInc.2011– 2016.AllRightsReserved

HadoopBeyondBatchwithYARN

SingleUseSysztemBatchApps

MultiUseDataPlatformBatch,Interactive,Online,Streaming,…

Ashiftfromtheoldtothenew…

HADOOP 1

MapReduce(cluster resource management

& data processing)

Data FlowPig

SQLHive

Others

API,Engine,

andSystem

YARN(Data Operating System: resource management, etc.)

Data FlowPig

SQLHive

OtherISV

Apache Yarn as a Base

System

Engine

API’s

1 ° ° ° ° °

° ° ° ° ° N

HDFS (redundant, reliable storage)

1 ° ° ° ° ° ° ° ° °

° ° ° ° ° ° ° ° ° N

HDFS (redundant, reliable storage)

BatchMapReduce

Tez Tez

MapReduce as the BaseHADOOP 2

38 ©HortonworksInc.2011– 2016.AllRightsReserved

ArchitectureEnabledbyYARNAsinglesetofdataacrosstheentireclusterwithmultipleaccessmethodsusing“zones”forprocessing

1 ° ° ° ° ° ° °

° ° ° ° ° ° ° °

° ° ° ° ° ° ° n

SQLHive

InteractiveSQLQueryforAnalytics

Pig

Script-basedETLAlgorithmexecutedinbatchtoreworkdatausedbyHiveandHBaseconsumers

• Maximize compute resources to lower TCO

• No standalone, silo’d clusters

• Simple management & operations

…all enabled by YARN

StreamProcessingStorm

Identify&actonreal-timeevents

NoSQLHbase

Accumulo

Low-latencyaccessservingupawebfrontend

39 ©HortonworksInc.2011– 2016.AllRightsReserved

HadoopWorkloadEvolution

SingleUseSystemBatchApps

MultiUseDataPlatformBatch,Interactive,Online,Streaming,…

Ashiftfromtheoldtothenew… MultiUsePlatformData&Beyond

HADOOP 1

YARN

HADOOP 2

1 ° ° ° °

° ° ° ° N

HDFS (redundant, reliable storage)

1 ° ° °

° ° ° N

HDFS

MapReduce

HADOOP.Next

YARN ‘

1 ° ° ° ° ° °

° ° ° ° ° ° N

HDFS (redundant, reliable storage)

DATA ACCESS APPS

Docker

MySQLMR2 Others(ISV Engines)

Multiple(Script, SQL, NoSQL, …)

MR2 Others(ISV Engines)

Multiple(Script, SQL, NoSQL, …)

Docker

Tomcat

Docker

Other

40 ©HortonworksInc.2011– 2016.AllRightsReserved

Gartner:WhatisHadoop?

à CommonApacheProjects– ALL=7(6)– Exceptfor1=3(5)– Exceptfor2=4(4)² About14CommonProjects

à UncommonProjects– Only1=9(1)– Only2=7 (2)– Only3=6 (3)² About22UncommonProjects

http://blogs.gartner.com/merv-adrian/2015/07/02/now-what-is-hadoop/

ODPi

ODPi

ODPi

ODPi

ODPi ODPi ODPi

41 ©HortonworksInc.2011– 2016.AllRightsReserved

HORTONWORKSDATAPLATFORM

Hado

op&YAR

N

Flum

e

Oozie

HDP2.3isApacheHadoop;not“basedon”Hadoop

Pig

Hive

Tez

Sqoo

p

Clou

dbreak

Amba

ri

Slider

Kafka

Knox

Solr

Zookeepe

r

Spark

Falcon

Ranger

HBase

Atlas

Accumulo

Storm

Phoe

nix

4.10.2

DATAMGMT DATAACCESS GOVERNANCE&INTEGRATION OPERATIONS SECURITY

HDP2.2Dec2014

HDP2.1April2014

HDP2.0Oct2013

HDP2.2Dec2014

HDP2.1April2014

HDP2.0Oct2013

0.12.0 0.12.0

0.12.1 0.13.0 0.4.0

1.4.4 1.4.4 3.3.23.4.5

0.4.00.5.0

0.14.0 0.14.0 3.4.6 0.5.0 0.4.00.9.30.5.2

4.0.04.7.2

1.2.1 0.60.0 0.98.4 4.2.0 1.6.1 0.6.0 1.5.21.4.5 4.1.02.0.0

1.4.0 1.5.1 4.0.0

1.3.1

1.5.1 1.4.4 3.4.5

2.2.0

2.4.0

2.6.0

2.7.1 1.4.6 1.0.0 0.6.0 0.5.02.1.00.8.2 3.4.61.5.25.2.1 0.80.0 0.5.01.7.04.4.0 0.10.0 0.6.10.7.01.2.10.15.0HDP2.3Oct2015 4.2.0

0.96.1

0.98.0 0.9.1

0.8.1

1.4.1 1.1.2

2.7.3 1.4.6 1.3.0 0.9.0 0.6.02.4.00.10.0 3.4.61.5.25.5.1 0.91.0 0.7.01.7.04.7.0 1.0.1 0.10.00.7.01.2.1+2.1***0.16.0

HDP2.5*2H2016

4.2.01.6.2+2.0** 1.1.2

2.7.1 1.4.6 1.2.0 0.6.0 0.5.02.2.10.9.0 3.4.61.5.25.2.1 0.80.0 0.5.01.7.04.4.0 0.10.0 0.6.10.7.01.2.10.15.0HDP2.4Mar2016 4.2.01.6.0 1.1.2

Zepp

elin

OngoingInnovationinApache

0.6.0

*HDP2.5– ShowscurrentApachebranchesbeingused.FinalcomponentversionsubjecttochangebasedonApachereleaseprocess.

**Spark1.6.2+Spark2.0– HDP2.5supportinstallationofbothSpark1.6.2andSpark2.0.Spark2.0isTechnicalPreviewwithinHDP2.5.

***Hive2.1isTechnicalPreviewwithinHDP2.5.

42 ©HortonworksInc.2011– 2016.AllRightsReserved

NextGenerationDataVendorsInvestmentfortheEnterprise

VerticalIntegration with YARN and HDFSEnsure engines can run reliably and respectfully in a YARN based cluster

Provision, Manage & Monitor

AmbariZookeeper

Scheduling

Oozie

Loaddataandmanageaccordingtopolicy

Providelayeredapproachto

securitythroughAuthentication,Authorization,Accounting,andDataProtection

SECURITYGOVERNANCE

Deployandeffectivelymanagetheplatform

° ° ° ° ° ° ° ° ° ° ° ° ° ° °

Script

Pig

SQL

Hive

JavaScala

Cascading

Stream

Storm

Search

Solr

NoSQL

HBaseAccumulo

BATCH, INTERACTIVE & REAL-TIME DATA ACCESS

In-Memory

Spark

Others

ISV Engines

1 ° ° ° ° ° ° ° ° ° ° ° ° ° °

YARN: Data Operating System(ClusterResourceManagement)

HDFS (Hadoop Distributed File System)

Tez Slider SliderTez Tez

OPERATIONS

Horizontal Integration for Enterprise ServicesEnsure consistent enterprise services are applied across the Hadoop stack

43 ©HortonworksInc.2011– 2016.AllRightsReserved

Whatdodistributionsdo?

à Defineastackofcomponents• RichandlatestsetofApacheProjects(opensource&opencommunity)withoutlockin

à VerticalandHorizontalintegrationofcomponents• Vertical:BestSpeedandScale• Horizontal:OpenEnterpriseReady

à ProvisionandUpgradestack• Robust,EasyandAnywhere

à Acceleratetimetovalue(easyofuse)• NewFaceofHadoopwithUis fromAmbari,AmbariViews,Ranger,Falcon,Atlas

à PartnerEcosystem• RichandDeep

à Support• Industry’sbest,SmartSense andinfluencecommunity

HadoopOperations&Tools

45 ©HortonworksInc.2011– 2016.AllRightsReserved

How Do You Operate a Hadoop Cluster?

Apache™ Ambari isaplatformtoprovision,manageandmonitorHadoopclusters

46 ©HortonworksInc.2011– 2016.AllRightsReserved

Ambari Core Features and Extensibility

Install&Configure

Operate,Manage&Administer

Develop

Optimize&Tune

Developer

DataArchitect

Ambariprovidescoreservicesforoperations,developmentandextensionspointsforboth

ExtensibilityFeatures

Stacks,Blueprints&RESTAPIs

CoreFeatures

InstallWizard&Web

Web,OperatorViews,Metrics&Alerts

UserViews

UserViews

ViewsFramework&RESTAPIs

ViewsFramework

ViewsFramework

How?

ClusterAdmin

47 ©HortonworksInc.2011– 2016.AllRightsReserved

Newuserinterfaceenablesfast&easySQLdefinitionandexecution.

48 ©HortonworksInc.2011– 2016.AllRightsReserved

DataWorkerandDevOpsTooling

à Apluggablewaytoprovideacommonuserexperienceacrossmultipleuserpersonas.

Ambari Views

HDP

SystemAdmin/operators

DataWorkers

ApplicationDevelopers

AMBAR I

à Singlepointofentryforallusers.

à OpenCommunity

49 ©HortonworksInc.2011– 2016.AllRightsReserved

New User Views for DevOps

CapacitySchedulerViewBrowseandmanageYARNqueues

Tez ViewViewinformationrelatedtoTez jobsthatareexecutingonthecluster

50 ©HortonworksInc.2011– 2016.AllRightsReserved

NewUserViewsforDevelopment

PigViewAuthorandexecutePigScripts.

HiveViewAuthor,executeanddebugHive

queries.

FilesViewBrowseHDFSfilesystem.

51 ©HortonworksInc.2011– 2016.AllRightsReserved

ApacheZeppelin

• Web-basednotebookfordataengineers,dataanalystsanddatascientists• Bringsinteractivedataingestion,data

exploration,visualization,sharingandcollaborationfeaturestoHadoopandSpark

• Moderndatasciencestudio• ScalawithSpark• PythonwithSpark• SparkSQL• ApacheHive,andmore.

HadoopDataAccess

53 ©HortonworksInc.2011– 2016.AllRightsReserved

Access patterns enabled by YARN

YARN: Data Operating System

1 ° ° ° ° ° ° ° ° °

° ° ° ° ° ° ° ° °

°

°N

HDFS Hadoop Distributed File System

Interactive Real-TimeBatch

Applications BatchNeeds to happen but, no timeframe limitations

InteractiveNeeds to happen at Human time

Real-Time Needs to happen at Machine Execution time.

54 ©HortonworksInc.2011– 2016.AllRightsReserved

Apache Hive: SQL in Hadoop

• Created by a team at Facebook

• Provides a standard SQL interface to data stored in Hadoop

• Quickly find value in raw data files

• Proven at petabyte scale

• Compatible with ALL major BI tools such as Tableau, Excel, MicroStrategy, Business Objects, etc…

SensorMobile

WeblogOperational

/MPP

SQLQueries

Page 55 © Hortonworks Inc. 2011 – 2014. All Rights Reserved

Hive Architecture

User issues SQL query

Hive parses and plans query

Query converted to MapReduce and executed on Hadoop

2

3

Web UI

JDBC / ODBC

CLIHiveSQL

1

1HiveServer2 Hive

MR/Tez Compiler

Optimizer

Executor

2

Hive

MetaStore(MySQL, Postgresql,

Oracle)

MapReduce, Tez or Spark Job

Data DataData

Hadoop 3Data-local processing

56 ©HortonworksInc.2011– 2016.AllRightsReserved

Hive and the Stinger Initiative

BaseOptimizationsGeneratesimplifiedDAGsIn-memoryHashJoins

VectorQueryEngineOptimizedformodernprocessor

architectures

TezExpresstasksmoresimply

EliminatediskwritesPre-warmedContainers

ORCFileColumnStore

HighCompressionPredicate/FilterPushdowns

YARNNext-genHadoopdataprocessing

framework

+ +

QueryPlannerIntelligentCost-BasedOptimizer

PerformanceOptimizations100x+fastertimetoinsightDeeperanalyticalcapabilities

57 ©HortonworksInc.2011– 2016.AllRightsReserved

Stinger.next andSub-SecondSQL

Emergence of LLAP brings Sub-Second SQL response times within reach with Hive.

BATCH & INTERACTIVE BATCH & INTERACTIVE BATCH, INTERACTIVE & SUB-SECONDSPEED

DELIVERY

SQL

UPDATES

ENGINES

STINGERD E L I V E R E D

PROGRESSD E L I V E R E D FINALVERSION

HDP 2.1VERSION

0.13VERSION

HDP 2.3VERSION

1.2.1

SQL:2003+ SQL:2011 SUBSET

READ-ONLY SQL INSERT/UPDATE/DELETE

MR, TEZ MR, TEZ

F U T U R ES T I N G E R N E X T

COMPLETE ACID SUPPORT INCLUDING MERGE

COMPREHENSIVE SQL:2011 BASED ANALYTICS

MR, TEZ, LLAP

DELIVERED IN DEVELOPMENT

TieredDataStorage

Stinger.next Phase3

YARN:ContainerizedApplications

58 ©HortonworksInc.2011– 2016.AllRightsReserved

DataTypes SQL Features File Formats Latest Additions…Numeric CoreSQLFeatures Columnar ScalableCrossProduct

FLOAT/DOUBLE Date,Time andArithmeticalFunctions ORCFile PrimaryKey/Foreign KeyDECIMAL INNER,OUTER,CROSSandSEMIJoins Parquet Non-EquijoinINT/TINYINT/SMALLINT/BIGINT DerivedTableSubqueries

TextTechPreview:Proc.Extensions(PL/SQL)

BOOLEAN Correlated+ UncorrelatedSubqueries CSV FutureString UNIONALL Logfile ACIDMERGE

CHAR/VARCHAR UDFs, UDAFs,UDTFs Nested/Complex MultiSubquerySTRING CommonTableExpressions Avro Comparisontosub-selectBINARY UNIONDISTINCT JSON INTERSECT andEXCEPT

Date, Time AdvancedAnalytics XMLDATE OLAPandWindowingFunctions CustomFormatsTIMESTAMP CUBE andGroupingSets OtherFeaturesIntervalTypes NestedDataAnalytics XPath Analytics

ComplexTypes NestedDataTraversalARRAY LateralViewsMAP ACIDTransactionsSTRUCT INSERT/UPDATE/DELETEUNION

ApacheHive:JourneytoSQL:2011Analytics

LegendExisting

Future

NewwithHive2.0

59 ©HortonworksInc.2011– 2016.AllRightsReserved

Stor

age

Columnar Storage

ORCFile Parquet

Unstructured Data

JSON CSV

Text Avro

Custom

Weblog

Engi

ne

SQL Engines

RowEngine VectorEngineSQ

L

SQL Support

SQL:2011 Optimizer HCatalog HiveServer2

Cac

he

Block Cache

LinuxCache

Dis

tribu

ted

Exec

utio

n

Hadoop 1

MapReduce

Hadoop 2

Tez Spark

Vector Cache

LLAP

Persistent Server

Historical

Current

In-development

Legend

Apache Hive: Modern Architecture

60 ©HortonworksInc.2011– 2016.AllRightsReserved

ApacheTezisacriticalinnovationoftheStingerInitiative.

• Along with YARN, Tez not only improves Hive, but improvesallthingsbatchand interactiveforHadoop;Pig,Cascading…

• More Efficient Processing than MapReduce

• Reduceoperationsandcomplexityofbackendprocessing• AllowsforMapReduceReducewhichsavesharddiskoperations• Implementsa“service”whichisalwayson,decreasingstarttimesofjobs• AllowsCachingofDatainMemory

YARN

Dev

Cascading/ Scalding

WhyisTez Important?

°1 ° ° ° ° ° ° ° °

° ° ° ° ° ° ° ° °

°

°°

° ° ° ° ° ° °

° ° ° ° ° ° N

HDFS (Hadoop Distributed File

System)

Scripting

Pig

SQL

Hive

Tez Tez

Applications

Tez

YARN:DataOperatingSystem

Interactive Real-TimeBatch

61 ©HortonworksInc.2011– 2016.AllRightsReserved

ApacheTez

Hive– MapReduce Hive– Tez

SELECT a.state, COUNT(*), AVG(c.price) FROM a

JOIN b ON (a.id = b.id)JOIN c ON (a.itemId = c.itemId)

GROUP BY a.state

SELECTa.state

JOIN(a,c)SELECTc.price

SELECTb.id

JOIN(a,b)GROUPBYa.state

COUNT(*)AVG(c.price)

M M M

R R

M M

R

M M

R

M M

R

HDFS

HDFS

HDFS

M M M

R R

R

M M

R

R

SELECTa.state,c.itemId

JOIN(a,c)

JOIN(a,b)GROUPBYa.state

COUNT(*)AVG(c.price)

SELECTb.id

Tez avoidsunneededwritestoHDFS

Tez allowsReducer-onlyjobswithintheDAG

62 ©HortonworksInc.2011– 2016.AllRightsReserved

Sub-secondQueriesinHive:LLAP(LiveLongandProcess)

à Persistentdaemons– Savestimeonprocessstartup(eliminatescontainerallocationandJVMstartuptime)– AllcodeJITed withinaqueryortwo

à Datacachingwithanasync I/Oelevator– Hotdatacachedinmemory(columnaraware,soonlyhotcolumnscached)– Whenpossibleworkscheduledonnodewithdatacached,ifnotworkwillberuninothernode

à OperatorscanbeexecutedinsideLLAPwhenitmakessense– Large,ETLstylequeriesusuallydon’tmakesense– UsercodenotruninLLAPforsecurity

à Workingoninterfacetoallowotherdataenginestoreadsecurelyinparallel

à Betain2.0

63 ©HortonworksInc.2011– 2016.AllRightsReserved

Hive2withLLAP:ArchitectureOverview

Deep

Storage

HDFS S3+OtherHDFSCompatibleFilesystems

YARNCluster

LLAPDaemon

QueryExecutors

In-MemoryCache

LLAPDaemon

QueryExecutors

In-MemoryCache

LLAPDaemon

QueryExecutors

In-MemoryCache

LLAPDaemon

QueryExecutors

In-MemoryCache

QueryCoordinators

Coord-inator

Coord-inator

Coord-inator

HiveServer2(Query

Endpoint)

ODBC/JDBC SQL

Queries

64 ©HortonworksInc.2011– 2016.AllRightsReserved

HiveWithLLAPExecutionOptions

AM AM

T T T

R R

R

T T

T

R

M M M

R R

R

M M

R

R

Tez Only LLAP + Tez

T T T

R R

R

T T

T

R

LLAP only

65 ©HortonworksInc.2011– 2016.AllRightsReserved

Scripting Data Pipeline & ETLApache Pig

• Dataflowengineandscriptinglanguage(PigLatin)• Allowsyoutotransform dataanddatasets

Advantages over MapReduce• Reducestimetowritejobs• Communitysupport• PiggybankhasasignificantnumberofUDF’stohelpadoption• TherearealargenumberofexistingshopsusingPIG

YARN:DataOperatingSystem

Interactive Real-TimeBatch

66 ©HortonworksInc.2011– 2016.AllRightsReserved

PigLatin

• Pigexecutesinauniquefashion:oDuringexecution,eachstatementisprocessedbythePiginterpreteroIfastatementisvalid,itgetsaddedtoalogicalplan builtbytheinterpreter

oThestepsinthelogicalplandonotactuallyexecuteuntilaDUMPorSTOREcommandisused

67 ©HortonworksInc.2011– 2016.AllRightsReserved

WhyusePig?

• Maybewewanttojointwodatasets,fromdifferentsources,onacommonvalue,andwanttofilter,andsort,andgettop5sites

68 ©HortonworksInc.2011– 2016.AllRightsReserved

Resource Management

Storage

Elegant Developer APIsDataFrames, Machine Learning, and SQL

Made for Data ScienceAll apps need to get predictive at scale and fine granularity

Democratize Machine LearningSpark is doing to ML on Hadoop what Hive did for SQL on Hadoop

CommunityBroad developer, customer and partner interest

Realize Value of Data Operating SystemA key tool in the Hadoop toolbox

ApacheSparkenthusiasm

Applications

SparkCoreEngine

ScalaJava

Pythonlibraries

MLlib(Machinelearning)

SparkSQL*

SparkStreaming*

SparkCoreEngine

69 ©HortonworksInc.2011– 2016.AllRightsReserved

Apache Spark & Apache Hadoop Perfect Together

General Purpose Data Access Engineforfast,large-scaledataprocessing

Designed for Iterative, In-Memorycomputationsandinteractivedatamining

Expressive Multi-Language APIsforJava,Scala,PythonandR

Built-in LibrariesEnabledataworkerstorapidlyiterateoverdatafor:ETL,MachineLearning,SQLandStreamprocessing

YARN

ScalaJava

PythonR

APIs

Spark Core Engine

Spark SQL

Spark StreamingMLlib GraphX

1 ° ° ° ° ° ° ° ° °

° ° ° ° ° ° ° ° ° °

°

NHDFS

70 ©HortonworksInc.2011– 2016.AllRightsReserved

Apache Projects Enable Access Patterns

Various open source projects have incubated in order to meet these access pattern needs

Today, they can all run on a single cluster on a single set of data because of YARN

All powered by a broad open community

YARN: Data Operating System

1 ° ° ° ° ° ° ° ° °

° ° ° ° ° ° ° ° °

°

°N

HDFS Hadoop Distributed File System

InteractiveSolr

SparkHivePig

Real-TimeHBase

AccumuloStorm

BatchMapReduce

Applications

Kafka

71 ©HortonworksInc.2011– 2016.AllRightsReserved

ConnectedDataArchitecture

©HortonworksInc.2011– 2016.AllRightsReserved

DatainMotion

DataatRest

DeepHistoricalAnalysis

DATA C ENTER

StreamAnalytics

EdgeData

DatainMotion

MachineLearning

C LOUD EdgeData

EdgeAnalytics

DataatRest

TransformationalApplicationsRequireConnectedData

EnableModernApplicationsonaConnectedDataArchitecture

ContinuousInsights

DeliverinsightsfromALLdata,origintorest

EnterpriseReady

ManagementSecurity

Governance

DeploymentFlexibility

OnPremiseCloudHybrid

OpenInnovation

ArchitectureCommunityEcosystem

HandsonLabOverview

HDP2.4Sandbox

à ProvidesFreepreconfiguredHDP– RunsinaVirtualMachineor

AzureHortonworks.com/sandbox

à EasytoUse– Operations

• Ambari– DevandDevOps

• AmbariUserViews– WebNotebook

• Zeppelin

à Workswith60+FreetutorialHortonworks.com/tutorials

DataDiscoveryLab• Elefante WineCompanyhasafleetofover100trucks.

• Thegeolocationdatacollectedfromthetruckscontainseventsgeneratedwhilethetruckdriversaredriving.

• Thecompany’sgoalwithHadoopistoMitigateRisk:o Understandcorrelationsbetweenmilesdrivenandeventso Computetheriskfactorforeachdriverbasedonmileage&events

o LabEnvo Sandbox2.4

o LabDoco URL:http://goo.gl/14OAato LoadDatao QueryDatao ProcessData

Elefante Wine Current Challenges

The CompanyElefante Wine is a boutique wine fulfillment company with a large fleet of trucks. It delivers wine in a highly-regulated industry with stringent transportation requirements.

The SituationRecently a number of driver violations led to fines and increased insurance rates

The Challenges• Rising Operational Costs• Driver Safety• Risk Management• Logistics Optimization

© Hortonworks Inc. 2012 Professional Services

Elefante WineCompanyhasalargefleetoftrucksinUSA

Atruckgeneratesmillionsofeventsforagivenroute;aneventcouldbe:

§ 'Normal'events:starting/stoppingofthevehicle

§ ‘Violation’events:speeding,excessiveaccelerationandbreaking,unsafetaildistance

Companyusesanapplicationthatmonitorstrucklocationsandviolationsfromthetruck/driverinreal-timetocalculaterisk

Route?Truck?Driver?

Analystsqueryabroadhistorytounderstandiftoday’sviolationsarepartofalargerproblemwithspecificroutes,trucks,ordrivers

Elefante Wine Risk and Driver Safety Challenges

Trucksoutfittedwithnewsensorsgeneratinglargevolumesofnewdata:

• Location

• Speed

• DriverViolations

Needtobeintegratereal-time&historicaldata

Increase safety and reduce liabilitiesAnticipate driver violations BEFORE they happen and take precautionary actionsFindpredictivecorrelationsindriverbehavioroverlargevolumesofreal-timedata

Difficult to deliver timely insights to the right people and systems to take action

Data DiscoveryUncover new findings

Predictive Analytics Identify your next best action

Better Understandingof the Past

Better Prediction of the Future

What’sourgoal?

à Solution:– CollectadditionaldataviasensorsintruckstobetterunderstandRiskFactors

à How:– Quicklystorenewsensordatainacommonrepository– Preparethedataforanalysis– Explorethedata– CalculateRisk– Generateareport

Move Data Into Hadoop

Geolocation.csv

trucks.csv

Geolocation_stage Geolocation

Trucks_stage Trucks

csv

csv ORC

ORCSQL

SQL

move

LOAD

Geolocation

Trucks

ORC

ORC

SQL

SQL

PIGorSparkRiskCalculation

Truck_mileage

ORC

Avg_mileage

ORC

DriverMileage

ORC

RiskFactor

ORC

Events

ORC

Trucking Risk Analysis – Hadoop ELT

CalculateRisk

GettingStartedResources

85 ©HortonworksInc.2011– 2016.AllRightsReserved

developer.hortonworks.com

86 ©HortonworksInc.2011– 2016.AllRightsReserved

HortonworksNourishestheCommunityHORTONWORKS

COMMUN I T Y CONNECT IONHORTONWORKS PARTNERWORKS

https://community.hortonworks.com

87 ©HortonworksInc.2011– 2016.AllRightsReserved

Thankyou!

[email protected]

@racoss

88 ©HortonworksInc.2011– 2016.AllRightsReserved

ProtectingtheElephantintheCastle…..Kerberos,

WireEncryption

HDFS Encryption

ApacheRangerNetworkSegmentation,

Firewalls

LDAP/AD

ApacheKnox