Smart Enterprise Big Data Bus
ForThe Modern Responsive Enterprise
Anand Venugopal
Sr. Director - StreamAnalytix
Larry Pearson
V.P. of Marketing
Recorded version available at http://bit.ly/1FND9fe
The Journey So Far…
• Impetus Technologies – Leading Big Data Solutions Provider
• StreamAnalytix – Enterprise Class Streaming Analytics platform
• Early Access Program: Aug 2014
• Pilot Projects: October 2014 – February 2015
• GA Launch: February 2015
• Today – Sharing one of the key insights and use-case patterns from pilot
Recorded version available at http://bit.ly/1FND9fe
Smart Enterprise Big Data Bus
Agenda
Why ? Case Study
Emerging Big Data Landscape and its Challenges
What ? Solution Characteristics
Components Required
How ? Implementation Using StreamAnalytix
Technology Stack
Q&A Summary
Follow up
Recorded version available at http://bit.ly/1FND9fe
Case Study - How It Began
Enterprise Data
Hub/Lake
Hadoop
App-1 App-2 App-3
Enterprise NoSQL
Healthcare giant is speeding up critical business processes by using a streaming analytics platform
for real-time data synchronization between their Hadoop platform and their enterprise NoSQL database
Recorded version available at http://bit.ly/1FND9fe
Emerging Enterprise Big Data Landscape
• Hadoop Stack – Becoming the Center of the Enterprise Data Universe
• Business Critical Applications Are Still ''Silos''
• Real-time Streaming Analytics Adoption Rapidly Increasing
Recorded version available at http://bit.ly/1FND9fe
Hadoop
Active
Archive
ETL
Single
Source of
Truth
BI +
Analytics
Streams/
Transactions
Hadoop – It Makes Coffee Too!
Recorded version available at http://bit.ly/1FND9fe
Hadoop Architecture View: Cloudera
Recorded version available at http://bit.ly/1FND9fe
Hadoop Architecture View: Hortonworks
Recorded version available at http://bit.ly/1FND9fe
Hadoop Architecture View: MapR
Recorded version available at http://bit.ly/1FND9fe
What’s Missing From These Pictures?
Recorded version available at http://bit.ly/1FND9fe
How Hadoop Connects To The Rest Of It All !
Enterprise Data
Hub/Lake
Hadoop
ERP CRM
Current
EDW/
ADW
App Server 1
App Server 2
Hub and Spoke
Architecture
The Integration Challenge – Developing, Maintaining, Managing the Hub-spoke System
HadoopActive
Archive
ETL
Single Source
of Truth
BI + Analytics
Streams/
Transactions
Recorded version available at http://bit.ly/1FND9fe
Emerging Enterprise Big Data Landscape
• Hadoop Stack – Becoming the Center of the Enterprise Data Universe
• Business Critical Applications Are Still ''Silos''
• Real-time Streaming Analytics Adoption Rapidly Increasing
Recorded version available at http://bit.ly/1FND9fe
Siloed Enterprise Applications
May Take Many Years To Integrate With Hadoop
Provisioning
Transaction
Processing
Billing
Customer
Service
Recorded version available at http://bit.ly/1FND9fe
The Data Integration Challenge Is Here To Stay
Enterprise Data
Hub/Lake
Hadoop
ERP CRM
Current
EDW/
ADW
App Server 1
App Server 2
Hub and Spoke
Architecture
The Integration Challenge – Developing, Maintaining, Managing the Hub-spoke System
Recorded version available at http://bit.ly/1FND9fe
Emerging Enterprise Big Data Landscape
• Hadoop Stack – Becoming the Center of the Enterprise Data Universe
• Business Critical Applications Are Still ''Silos''
• Real-time Streaming Analytics Adoption Rapidly Increasing
Recorded version available at http://bit.ly/1FND9fe
• Growth of Internet of Things (IoT) and
Sensor/ Machine Data Sources
• Context-sensitive Customer Service Sales
Web
Site
Billing
Customer
Service
The Modern Enterprise
Expected To Be ''Real-time'' Or ''Near Real-time''
Recorded version available at http://bit.ly/1FND9fe
• Mobile Location Based Offers
• Internet Advertisements
• Call-center Agent Interactions
The Modern Enterprise
Expected To Be ''Real-time'' Or ''Near Real-time''
Sales
Web
Site
Billing
Customer
Service
Recorded version available at http://bit.ly/1FND9fe
Provisioning
Machine Data
Processing
Billing
Customer
Service
Enterprise Data
Hub/Lake
Hadoop
Real-time Responses Need Hi-speed Any-to-Any
Data Synchronization & Analytics Over Streaming Data
Developing and managing all the peer-to-peer data interfaces and the hub-spoke model together would be very hard
Recorded version available at http://bit.ly/1FND9fe
Emerging Enterprise Big Data Landscape
• Hadoop Stack – Becoming the Center of the Enterprise Data Universe
• Business Critical Applications Are Still ''Silos''
• Real-time Streaming Analytics Adoption Rapidly Increasing
Recorded version available at http://bit.ly/1FND9fe
Smart Enterprise Big Data Bus
Agenda
Why ? Case Study
Emerging Big Data Landscape and its Challenges
What ? Solution Characteristics
Components Required
How ? Implementation Using StreamAnalytix
Technology Stack
Q&A Summary
Follow up
Recorded version available at http://bit.ly/1FND9fe
Provisioning
Machine Data
Processing
Billing
Customer
Service
Enterprise Data
Hub/Lake
Hadoop
A Hi-speed Big Data Bus Architecture Would Be An
Efficient Any-to-Any Data Synchronization Mechanism
Recorded version available at http://bit.ly/1FND9fe
Provisioning
Machine Data
Processing
Billing
Customer
Service
Enterprise Data
Hub/Lake
Hadoop
A Hi-speed Big Data Bus Architecture Would Be An
Efficient Any-to-Any Data Synchronization Mechanism
Recorded version available at http://bit.ly/1FND9fe
Provisioning
Machine Data
Processing
Billing
Customer
Service
Enterprise Data
Hub/Lake
Hadoop
A Hi-speed Big Data Bus Architecture Would Be An
Efficient Any-to-Any Data Synchronization Mechanism
Recorded version available at http://bit.ly/1FND9fe
Provisioning
Machine Data
Processing
Billing
Customer
Service
Enterprise Data
Hub/Lake
Hadoop
The Hi-speed Big Data Bus Would Also Need To Support
''On The Wire'' Computation For Data Analytics, Transforms, CEP
Transformation
Analytics, Alerting
Recorded version available at http://bit.ly/1FND9fe
Capability List for the ''Smart Enterprise Big Data Bus''
• Ingest
• Parse
• Filter
• Transform
• Move
• Store
• Read
• Synchronize
• Analyse
• Predict
• Alert
• Visualise
AT SCALE, AND FAST !
Provisioning
Machine Data
Processing
Billing
Enterprise Data
Hub/Lake
Hadoop
Transformation
Analytics,
AlertingCustomer
Service
Recorded version available at http://bit.ly/1FND9fe
Is this real ?? Case Studies
Recorded version available at http://bit.ly/1FND9fe
Is this real ?? Case Studies
Read-Write Adapters
Stream Processing Services provided
by the ''Smart Enterprise Big Data Bus''
include UI for Work-flow Orchestration,
Management and Monitoring
''Stations'' in the Data Transit System
Reliable, Fault-tolerant,
Elastic Scalable Distributed Stream
Processing and Transport Fabric
Recorded version available at http://bit.ly/1FND9fe
ESB (vs. Smart Enterprise Big Data Bus)
• Were architected for a different workload in a different era
• Designed for light weight remote service invocations – not as a
heavy throughput full peer-to-peer data transfer mechanism
• No compute / analytics capability on the wire
• Expensive vertical scaling vs. distributed elastic scale-out with
commodity hardware
• Monolithic workflows vs. independent control and elastic
scalability of each stage in a workflow based on compute needs
Recorded version available at http://bit.ly/1FND9fe
Smart Enterprise Big Data Bus
Agenda
Why ? Case Study
Emerging Big Data Landscape and its Challenges
What ? Solution Characteristics
Components Required
How ? Implementation using StreamAnalytix
Technology Stack
Q&A Summary
Follow up
Recorded version available at http://bit.ly/1FND9fe
Smart Enterprise Big Data Bus Implementation
• Kafka
• Rabbit MQ
• Apache Storm
• Operators
• RT Dashoards
• Websockets
• CEP
• Filter
• Indexer
• NoSQL
• HDFS, Hbase
• PMML
AT SCALE, FAST, EASY !
Recorded version available at http://bit.ly/1FND9fe
StreamAnalytix Architecture Block Diagram
Recorded version available at http://bit.ly/1FND9fe
StreamAnalytix – Enterprise Big Data Bus – Creation
Recorded version available at http://bit.ly/1FND9fe
Data Fabric - 2
Recorded version available at http://bit.ly/1FND9fe
Monitoring Screen
Recorded version available at http://bit.ly/1FND9fe
Real-Time Dashboard
Recorded version available at http://bit.ly/1FND9fe
Sample Deployment/Hardware Specification
External systems
Data Store
Integration layer
Processing and Analytics layer
Storm
RHEL 4 cores, 16GB RAM
vm06dev222
x 1
ElasticSearch
RHEL 8 cores, 32GB RAM
vm06dev218
x 1
PostgreSQL
RHEL 4 cores, 8GB RAM
vm06dev222
x 1
Meta Information Store
Zookeeper
x 1
Kafka+ZK
RHEL 4 cores, and 8GB RAM
vm06dev220
x 1
Free Switch
x 2
Graphite
x 1
Web UI Layer
RHEL 4 cores, and 2GB RAM
vm07eng29, vm07eng31
IR(Platform)
x 1
RHEL 4 cores, and 2GB RAM
vm07dev97
RHEL 4 cores, 16GB RAM
vm06dev222
RHEL 4 cores, 4GB RAM
vm06dev224
Tomcat(Query Server)
RHEL 4 cores, 8GB RAM
vm06dev218
x 1
Tomcat(Admin+Log UI)
RHEL 4 cores, 16GB RAM
vm06dev222
x 1
Couchbase
RHEL 8 cores, 32GB RAM
vm06dev224
x 1
RabbitMQ
RHEL 4 cores, and 4GB RAM
vm06dev224
x 1
www.streamanalytix.com
Email: inquiry@streamanalytix.com
?
Recorded version available at http://bit.ly/1FND9fe

Smart Enterprise Big Data Bus for the Modern Responsive Enterprise- StreamAnalytix Webinar

  • 1.
    Smart Enterprise BigData Bus ForThe Modern Responsive Enterprise Anand Venugopal Sr. Director - StreamAnalytix Larry Pearson V.P. of Marketing Recorded version available at http://bit.ly/1FND9fe
  • 2.
    The Journey SoFar… • Impetus Technologies – Leading Big Data Solutions Provider • StreamAnalytix – Enterprise Class Streaming Analytics platform • Early Access Program: Aug 2014 • Pilot Projects: October 2014 – February 2015 • GA Launch: February 2015 • Today – Sharing one of the key insights and use-case patterns from pilot Recorded version available at http://bit.ly/1FND9fe
  • 3.
    Smart Enterprise BigData Bus Agenda Why ? Case Study Emerging Big Data Landscape and its Challenges What ? Solution Characteristics Components Required How ? Implementation Using StreamAnalytix Technology Stack Q&A Summary Follow up Recorded version available at http://bit.ly/1FND9fe
  • 4.
    Case Study -How It Began Enterprise Data Hub/Lake Hadoop App-1 App-2 App-3 Enterprise NoSQL Healthcare giant is speeding up critical business processes by using a streaming analytics platform for real-time data synchronization between their Hadoop platform and their enterprise NoSQL database Recorded version available at http://bit.ly/1FND9fe
  • 5.
    Emerging Enterprise BigData Landscape • Hadoop Stack – Becoming the Center of the Enterprise Data Universe • Business Critical Applications Are Still ''Silos'' • Real-time Streaming Analytics Adoption Rapidly Increasing Recorded version available at http://bit.ly/1FND9fe
  • 6.
    Hadoop Active Archive ETL Single Source of Truth BI + Analytics Streams/ Transactions Hadoop– It Makes Coffee Too! Recorded version available at http://bit.ly/1FND9fe
  • 7.
    Hadoop Architecture View:Cloudera Recorded version available at http://bit.ly/1FND9fe
  • 8.
    Hadoop Architecture View:Hortonworks Recorded version available at http://bit.ly/1FND9fe
  • 9.
    Hadoop Architecture View:MapR Recorded version available at http://bit.ly/1FND9fe
  • 10.
    What’s Missing FromThese Pictures? Recorded version available at http://bit.ly/1FND9fe
  • 11.
    How Hadoop ConnectsTo The Rest Of It All ! Enterprise Data Hub/Lake Hadoop ERP CRM Current EDW/ ADW App Server 1 App Server 2 Hub and Spoke Architecture The Integration Challenge – Developing, Maintaining, Managing the Hub-spoke System HadoopActive Archive ETL Single Source of Truth BI + Analytics Streams/ Transactions Recorded version available at http://bit.ly/1FND9fe
  • 12.
    Emerging Enterprise BigData Landscape • Hadoop Stack – Becoming the Center of the Enterprise Data Universe • Business Critical Applications Are Still ''Silos'' • Real-time Streaming Analytics Adoption Rapidly Increasing Recorded version available at http://bit.ly/1FND9fe
  • 13.
    Siloed Enterprise Applications MayTake Many Years To Integrate With Hadoop Provisioning Transaction Processing Billing Customer Service Recorded version available at http://bit.ly/1FND9fe
  • 14.
    The Data IntegrationChallenge Is Here To Stay Enterprise Data Hub/Lake Hadoop ERP CRM Current EDW/ ADW App Server 1 App Server 2 Hub and Spoke Architecture The Integration Challenge – Developing, Maintaining, Managing the Hub-spoke System Recorded version available at http://bit.ly/1FND9fe
  • 15.
    Emerging Enterprise BigData Landscape • Hadoop Stack – Becoming the Center of the Enterprise Data Universe • Business Critical Applications Are Still ''Silos'' • Real-time Streaming Analytics Adoption Rapidly Increasing Recorded version available at http://bit.ly/1FND9fe
  • 16.
    • Growth ofInternet of Things (IoT) and Sensor/ Machine Data Sources • Context-sensitive Customer Service Sales Web Site Billing Customer Service The Modern Enterprise Expected To Be ''Real-time'' Or ''Near Real-time'' Recorded version available at http://bit.ly/1FND9fe
  • 17.
    • Mobile LocationBased Offers • Internet Advertisements • Call-center Agent Interactions The Modern Enterprise Expected To Be ''Real-time'' Or ''Near Real-time'' Sales Web Site Billing Customer Service Recorded version available at http://bit.ly/1FND9fe
  • 18.
    Provisioning Machine Data Processing Billing Customer Service Enterprise Data Hub/Lake Hadoop Real-timeResponses Need Hi-speed Any-to-Any Data Synchronization & Analytics Over Streaming Data Developing and managing all the peer-to-peer data interfaces and the hub-spoke model together would be very hard Recorded version available at http://bit.ly/1FND9fe
  • 19.
    Emerging Enterprise BigData Landscape • Hadoop Stack – Becoming the Center of the Enterprise Data Universe • Business Critical Applications Are Still ''Silos'' • Real-time Streaming Analytics Adoption Rapidly Increasing Recorded version available at http://bit.ly/1FND9fe
  • 20.
    Smart Enterprise BigData Bus Agenda Why ? Case Study Emerging Big Data Landscape and its Challenges What ? Solution Characteristics Components Required How ? Implementation Using StreamAnalytix Technology Stack Q&A Summary Follow up Recorded version available at http://bit.ly/1FND9fe
  • 21.
    Provisioning Machine Data Processing Billing Customer Service Enterprise Data Hub/Lake Hadoop AHi-speed Big Data Bus Architecture Would Be An Efficient Any-to-Any Data Synchronization Mechanism Recorded version available at http://bit.ly/1FND9fe
  • 22.
    Provisioning Machine Data Processing Billing Customer Service Enterprise Data Hub/Lake Hadoop AHi-speed Big Data Bus Architecture Would Be An Efficient Any-to-Any Data Synchronization Mechanism Recorded version available at http://bit.ly/1FND9fe
  • 23.
    Provisioning Machine Data Processing Billing Customer Service Enterprise Data Hub/Lake Hadoop AHi-speed Big Data Bus Architecture Would Be An Efficient Any-to-Any Data Synchronization Mechanism Recorded version available at http://bit.ly/1FND9fe
  • 24.
    Provisioning Machine Data Processing Billing Customer Service Enterprise Data Hub/Lake Hadoop TheHi-speed Big Data Bus Would Also Need To Support ''On The Wire'' Computation For Data Analytics, Transforms, CEP Transformation Analytics, Alerting Recorded version available at http://bit.ly/1FND9fe
  • 25.
    Capability List forthe ''Smart Enterprise Big Data Bus'' • Ingest • Parse • Filter • Transform • Move • Store • Read • Synchronize • Analyse • Predict • Alert • Visualise AT SCALE, AND FAST ! Provisioning Machine Data Processing Billing Enterprise Data Hub/Lake Hadoop Transformation Analytics, AlertingCustomer Service Recorded version available at http://bit.ly/1FND9fe
  • 26.
    Is this real?? Case Studies Recorded version available at http://bit.ly/1FND9fe
  • 27.
    Is this real?? Case Studies Read-Write Adapters Stream Processing Services provided by the ''Smart Enterprise Big Data Bus'' include UI for Work-flow Orchestration, Management and Monitoring ''Stations'' in the Data Transit System Reliable, Fault-tolerant, Elastic Scalable Distributed Stream Processing and Transport Fabric Recorded version available at http://bit.ly/1FND9fe
  • 28.
    ESB (vs. SmartEnterprise Big Data Bus) • Were architected for a different workload in a different era • Designed for light weight remote service invocations – not as a heavy throughput full peer-to-peer data transfer mechanism • No compute / analytics capability on the wire • Expensive vertical scaling vs. distributed elastic scale-out with commodity hardware • Monolithic workflows vs. independent control and elastic scalability of each stage in a workflow based on compute needs Recorded version available at http://bit.ly/1FND9fe
  • 29.
    Smart Enterprise BigData Bus Agenda Why ? Case Study Emerging Big Data Landscape and its Challenges What ? Solution Characteristics Components Required How ? Implementation using StreamAnalytix Technology Stack Q&A Summary Follow up Recorded version available at http://bit.ly/1FND9fe
  • 30.
    Smart Enterprise BigData Bus Implementation • Kafka • Rabbit MQ • Apache Storm • Operators • RT Dashoards • Websockets • CEP • Filter • Indexer • NoSQL • HDFS, Hbase • PMML AT SCALE, FAST, EASY ! Recorded version available at http://bit.ly/1FND9fe
  • 31.
    StreamAnalytix Architecture BlockDiagram Recorded version available at http://bit.ly/1FND9fe
  • 32.
    StreamAnalytix – EnterpriseBig Data Bus – Creation Recorded version available at http://bit.ly/1FND9fe
  • 33.
    Data Fabric -2 Recorded version available at http://bit.ly/1FND9fe
  • 34.
    Monitoring Screen Recorded versionavailable at http://bit.ly/1FND9fe
  • 35.
    Real-Time Dashboard Recorded versionavailable at http://bit.ly/1FND9fe
  • 36.
    Sample Deployment/Hardware Specification Externalsystems Data Store Integration layer Processing and Analytics layer Storm RHEL 4 cores, 16GB RAM vm06dev222 x 1 ElasticSearch RHEL 8 cores, 32GB RAM vm06dev218 x 1 PostgreSQL RHEL 4 cores, 8GB RAM vm06dev222 x 1 Meta Information Store Zookeeper x 1 Kafka+ZK RHEL 4 cores, and 8GB RAM vm06dev220 x 1 Free Switch x 2 Graphite x 1 Web UI Layer RHEL 4 cores, and 2GB RAM vm07eng29, vm07eng31 IR(Platform) x 1 RHEL 4 cores, and 2GB RAM vm07dev97 RHEL 4 cores, 16GB RAM vm06dev222 RHEL 4 cores, 4GB RAM vm06dev224 Tomcat(Query Server) RHEL 4 cores, 8GB RAM vm06dev218 x 1 Tomcat(Admin+Log UI) RHEL 4 cores, 16GB RAM vm06dev222 x 1 Couchbase RHEL 8 cores, 32GB RAM vm06dev224 x 1 RabbitMQ RHEL 4 cores, and 4GB RAM vm06dev224 x 1
  • 37.