page
VOLTDB AND HPE VERTICA:
BUILDING AN IOT ARCHITECTURE
FOR FAST + BIG DATA 
page© 2016 VoltDB
OUR SPEAKERS
2
Chris Selland
VP Business Development
Hewlett Packard Enterprise
Dennis Duckworth, Director
Product Marketing
VoltDB
page© 2016 VoltDB page
AGENDA
•  Introduction
•  What is the Internet of Things?
•  HPE Vertica and IoT
•  VoltDB and IoT
•  Demo of VoltDB and HPE Vertica
•  Q&A
3
Building an IoT
Architecture for Big &
Fast Data
Chris Selland
VP Business Development
What is the Internet of Things?
“The Internet of Things is the network of physical objects that
contains embedded technology to communicate and sense or
interact with the objects' internal state or the external
environment.”1
Consumer “things”
Personal sensing with remote
monitoring and control
1 Gartner. IT Glossary – Internet of
Things.
Industrial “things”
From individual sensors to
entire power plants
Business outcomes
through big data analytics
How can value be created from IoT data?
Up to $11 trillion per year in potential economic impact by 20251
•  Increased revenue
•  Optimized operations
•  New products and services
•  Workforce productivity
•  Reduced risks
•  Reduced operating costs
•  Optimized maintenance
•  Optimized network throughput
Smart citiesManufacturing
Consumer
Smart homes
Healthcare
Oil & gas
Aviation
Energy & utilities
Transportation
Retail
Industries
that can benefit
Insurance
Agriculture
1 McKinsey Global Institute. The Internet of Things: Mapping the Value Beyond the Hype, June 2015.
Connected devices
source of contextual data
IoT is not the future. It’s already here.
Business Outcomes
•  Understand driver behaviors and
patterns
•  To improve efficiency of the
company and drivers
Business Outcomes
•  Safety and well-being of citizens
•  Analyzing safety threats
•  Accurate identification and scene
analysis
Business Outcomes
•  Increases customer value with
equipment monitoring, event
mitigation, & energy performance
services
•  Helps customers reduce energy
& maintenance costs
7
Equipment Manufacturers Enterprises Service-based Businesses
Integrated sensor and telemetry
analytics in HP Fleet vehicles
Monitors 2000+ cameras and
performs video analytics
Analytics enhance the Trane
Intelligent Services business
Successes in IoT
HPE Vertica and IoT
8
Computing at the Edge – “shifting left” for strategic advantage
Goal
–  Processing streaming data as close
to the sensor as possible creates
new opportunities
Advantage
–  Processing data streams in real time,
before the data is stored for additional
analysis, creates advantages
–  Example: Transformer or turbine thermal
runaway, requires immediate action
Result
–  Fast action prior to data storage
prevents data obsolescence and lost
opportunities/alerts
HPE Confidential | Share under NDA 9
“Things” generate data and
need control
Edge IT, data center and cloudOperations technology
Early analytics
and compute
Deep analytics
and compute
Data is sensed,
Things controlled
Data acquired
and aggregated
The Edge
Opportunity
Accelerate insight by moving compute
from the data center to the Edge
Data flow
Control flow
Augment the data center
with processing at the edge
The Vertica Real-Time Analytics Engine
Native
High
Availability
Standard
SQL
Interface
Column
Orientation
Auto
Database
Design
Advanced
Compression
MPP
Massive
Parallel
Processing
Leverages BI, ETL,
Hadoop/MapReduce
and OLTP investments
No disk I/O bottleneck
simultaneously load &
query
Native DB-aware
clustering on low-cost
x86 Linux nodes
Built-in redundancy
that also speeds up
queries
Automatic setup,
optimization, and DB
management
Up to 90% space
reduction using 10+
algorithms
ü  50x – 1000x faster
than traditional
RDBMS
ü  Scales from TB to
PB with industry-
standard hardware
ü  Simple integration
with existing ETL
and BI solutions
ü  SQL-99+ compliant
ü  Ultimate deployment
flexibility
ü  Extended advanced
analytics
ü  24/7 Load & Query
Confidential 10
Taking IoT analytics to the next level – Integrated Machine Learning
Native Machine Learning algorithms run in
database (Vertica)
•  K-means (anomalies)
•  Linear Regression (risks, trends)
•  Logistic Regression
Train predictive analytics models in the
datacenter, and easily run at the edge
Utilizes the same hardware as the
database itself
•  Lower cost solution
•  No transfer of data required
•  Large data sets lead to more accurate
models
11
Taking IoT analytics to the next level – High Performance Messaging
Vertica Streaming Adapter for Apache
Kafka
Kafka replaces custom data ingest
solutions with a robust open-source
implementation
Enables distributed data pipelines for
high throughput and low latency ingest
into Vertica
•  Time to insight reduced from hours to
seconds (micro-batch loading)
•  Easy handling of data bursts (10m
messages per minute)
12
Taking IoT analytics to the next level – In-database Sensor Data Analytics
Powerful time series and window
functions (gap filling, interpolation) for
data quality management at the edge1
Live Aggregate Projections
(personalized billing on demand)
Log text analytics and pattern
matching (SIEM – security, intrusion
detection)
Geospatial analytics (location based
services, asset management)
13
1 HPE internal testing of Vertica on EL4000 shows it can:
-  Repair 5.7 million readings per second
-  Load, repair and analyze a reading every 282 nanoseconds
www.hpe.com/events/bdc
Big Data Conference 2016
Boston August 29- September 1
page
BUILDING AN IOT ARCHITECTURE
FOR FAST + BIG DATA 
DENNIS DUCKWORTH
DIRECTOR OF PRODUCT MARKETING
page© 2016 VoltDB
VOLTDB DOESN’T MAKE THE APPS, WE MAKE THE APPS…
16
• Real-time intelligence and context for richer interactions
• Allows unique decisions on each individual event or person
• Analyze and act on real-time/streaming data
• More efficient use of hardware
• 100X faster than traditional transactional databases
• World record performance in the cloud (YCSB)
• Millisecond response times
• Hundreds of thousands to millions of transactions per
second
• High-speed data ingestion
• Simpler apps, easier to test and maintain
• Easier to program with SQL + Java
• Seamless ecosystem integration
• Data is always consistent and correct, never lost
Smarter
Faster
Simpler
1/10
of the Resources Needed	
  
100X
Traditional Transactional DB	
  
100%
Consistent, Correct	
  
page© 2016 VoltDB
Low  Complexity  
  
Rich,  Smart  
   Value of Individual Data Item Value of Data Collection
DataValue
Data Warehouse
Hadoop, etc.NoSQL
Data in Motion Data at Rest
Fast Data Big Data
Feeds, Collectors
CEP
CEP + DB
DataInteraction
FAST DATA + BIG DATA
page© 2016 VoltDB
Streaming
Analytics
-  Filtering
-  Windowing
-  Aggregation
-  Enrichment
-  Correlations
Deep
Analytics
-  Machine Learning
-  Statistical correlations
-  Multi-dimensional analysis
-  Predictive/Prescriptive analytics
Operational
Interactions /
Transactions
-  Context-aware
-  Personal
-  Real-time
+
Big Data
FAST DATA + BIG DATA
Fast Data
page© 2016 VoltDB
CUSTOMERS USING VOLTDB FOR IOT TODAY
•  Smart Meters (over 60 million meters under
VoltDB management)
•  Smart Energy GB
•  Smart Metering in Japan (Shikoku and Hokkaido)
•  Wearables
•  Home Security
•  Equipment Service Automation
page© 2016 VoltDB
EQUIPMENT SERVICE AUTOMATION:
LEADING STORAGE EQUIPMENT VENDOR USING IOT
page© 2016 VoltDB 21
#Simplify
page© 2016 VoltDB
Proven
ü  Jepsen Tested
Simpler
ü  Fewer components and clusters needed
ü  Easier to write, test and maintain applications
Familiar
ü  SQL and Java
Glue
Code
Glue
Code
Zookeeper
“DIY” STREAMING DATA INFRASTRUCTURE VS VOLTDB
page© 2016 VoltDB
EXAMPLE: LARGE TELCO IOT PLATFORM
23
Challenges:
•  High volume and velocity of data from smart
devices
•  Complexity (multiple ingest points, apps,
databases)
•  Performance – need to automate action on
inbound data at the velocity of the feeds
Notes
•  Applications have their own database
•  Each database replicates to Cassandra
and Hadoop
•  In-memory grid used to maintain logic and
publish ‘state’ back and forth
•  Rules engine for intra-day data to trigger
actions (e.g., ‘turn lights on’)
•  PostgreSQL used for dimension data
page© 2016 VoltDB
BEFORE
24
AFTER
Results
ü  Simplified system architecture
ü  Single ingest point for high-velocity feeds of inbound data
ü  Faster time to value
page© 2016 VoltDB
BEFORE
25
OR POSSIBLY
Results
ü  Simplified system architecture
ü  Single ingest point for high-velocity feeds of inbound data
ü  Faster time to value
page© 2016 VoltDB
DEMO
26
page© 2016 VoltDB
DEMO
27
page© 2016 VoltDB
QUESTIONS?
•  Use the chat window to type in your questions
•  Download our ebook,
Architecting for the Internet of Things
(learn.voltdb.com/EbookIoT.html)
•  Try VoltDB yourself
Free trial of the Enterprise Edition:
www.voltdb.com/Download
•  Email us at: info@voltdb.com
28

VoltDB and HPE Vertica Present: Building an IoT Architecture for Fast + Big Data

  • 1.
    page VOLTDB AND HPEVERTICA: BUILDING AN IOT ARCHITECTURE FOR FAST + BIG DATA 
  • 2.
    page© 2016 VoltDB OURSPEAKERS 2 Chris Selland VP Business Development Hewlett Packard Enterprise Dennis Duckworth, Director Product Marketing VoltDB
  • 3.
    page© 2016 VoltDBpage AGENDA •  Introduction •  What is the Internet of Things? •  HPE Vertica and IoT •  VoltDB and IoT •  Demo of VoltDB and HPE Vertica •  Q&A 3
  • 4.
    Building an IoT Architecturefor Big & Fast Data Chris Selland VP Business Development
  • 5.
    What is theInternet of Things? “The Internet of Things is the network of physical objects that contains embedded technology to communicate and sense or interact with the objects' internal state or the external environment.”1 Consumer “things” Personal sensing with remote monitoring and control 1 Gartner. IT Glossary – Internet of Things. Industrial “things” From individual sensors to entire power plants
  • 6.
    Business outcomes through bigdata analytics How can value be created from IoT data? Up to $11 trillion per year in potential economic impact by 20251 •  Increased revenue •  Optimized operations •  New products and services •  Workforce productivity •  Reduced risks •  Reduced operating costs •  Optimized maintenance •  Optimized network throughput Smart citiesManufacturing Consumer Smart homes Healthcare Oil & gas Aviation Energy & utilities Transportation Retail Industries that can benefit Insurance Agriculture 1 McKinsey Global Institute. The Internet of Things: Mapping the Value Beyond the Hype, June 2015. Connected devices source of contextual data
  • 7.
    IoT is notthe future. It’s already here. Business Outcomes •  Understand driver behaviors and patterns •  To improve efficiency of the company and drivers Business Outcomes •  Safety and well-being of citizens •  Analyzing safety threats •  Accurate identification and scene analysis Business Outcomes •  Increases customer value with equipment monitoring, event mitigation, & energy performance services •  Helps customers reduce energy & maintenance costs 7 Equipment Manufacturers Enterprises Service-based Businesses Integrated sensor and telemetry analytics in HP Fleet vehicles Monitors 2000+ cameras and performs video analytics Analytics enhance the Trane Intelligent Services business Successes in IoT
  • 8.
  • 9.
    Computing at theEdge – “shifting left” for strategic advantage Goal –  Processing streaming data as close to the sensor as possible creates new opportunities Advantage –  Processing data streams in real time, before the data is stored for additional analysis, creates advantages –  Example: Transformer or turbine thermal runaway, requires immediate action Result –  Fast action prior to data storage prevents data obsolescence and lost opportunities/alerts HPE Confidential | Share under NDA 9 “Things” generate data and need control Edge IT, data center and cloudOperations technology Early analytics and compute Deep analytics and compute Data is sensed, Things controlled Data acquired and aggregated The Edge Opportunity Accelerate insight by moving compute from the data center to the Edge Data flow Control flow Augment the data center with processing at the edge
  • 10.
    The Vertica Real-TimeAnalytics Engine Native High Availability Standard SQL Interface Column Orientation Auto Database Design Advanced Compression MPP Massive Parallel Processing Leverages BI, ETL, Hadoop/MapReduce and OLTP investments No disk I/O bottleneck simultaneously load & query Native DB-aware clustering on low-cost x86 Linux nodes Built-in redundancy that also speeds up queries Automatic setup, optimization, and DB management Up to 90% space reduction using 10+ algorithms ü  50x – 1000x faster than traditional RDBMS ü  Scales from TB to PB with industry- standard hardware ü  Simple integration with existing ETL and BI solutions ü  SQL-99+ compliant ü  Ultimate deployment flexibility ü  Extended advanced analytics ü  24/7 Load & Query Confidential 10
  • 11.
    Taking IoT analyticsto the next level – Integrated Machine Learning Native Machine Learning algorithms run in database (Vertica) •  K-means (anomalies) •  Linear Regression (risks, trends) •  Logistic Regression Train predictive analytics models in the datacenter, and easily run at the edge Utilizes the same hardware as the database itself •  Lower cost solution •  No transfer of data required •  Large data sets lead to more accurate models 11
  • 12.
    Taking IoT analyticsto the next level – High Performance Messaging Vertica Streaming Adapter for Apache Kafka Kafka replaces custom data ingest solutions with a robust open-source implementation Enables distributed data pipelines for high throughput and low latency ingest into Vertica •  Time to insight reduced from hours to seconds (micro-batch loading) •  Easy handling of data bursts (10m messages per minute) 12
  • 13.
    Taking IoT analyticsto the next level – In-database Sensor Data Analytics Powerful time series and window functions (gap filling, interpolation) for data quality management at the edge1 Live Aggregate Projections (personalized billing on demand) Log text analytics and pattern matching (SIEM – security, intrusion detection) Geospatial analytics (location based services, asset management) 13 1 HPE internal testing of Vertica on EL4000 shows it can: -  Repair 5.7 million readings per second -  Load, repair and analyze a reading every 282 nanoseconds
  • 14.
    www.hpe.com/events/bdc Big Data Conference2016 Boston August 29- September 1
  • 15.
    page BUILDING AN IOTARCHITECTURE FOR FAST + BIG DATA  DENNIS DUCKWORTH DIRECTOR OF PRODUCT MARKETING
  • 16.
    page© 2016 VoltDB VOLTDBDOESN’T MAKE THE APPS, WE MAKE THE APPS… 16 • Real-time intelligence and context for richer interactions • Allows unique decisions on each individual event or person • Analyze and act on real-time/streaming data • More efficient use of hardware • 100X faster than traditional transactional databases • World record performance in the cloud (YCSB) • Millisecond response times • Hundreds of thousands to millions of transactions per second • High-speed data ingestion • Simpler apps, easier to test and maintain • Easier to program with SQL + Java • Seamless ecosystem integration • Data is always consistent and correct, never lost Smarter Faster Simpler 1/10 of the Resources Needed   100X Traditional Transactional DB   100% Consistent, Correct  
  • 17.
    page© 2016 VoltDB Low Complexity     Rich,  Smart     Value of Individual Data Item Value of Data Collection DataValue Data Warehouse Hadoop, etc.NoSQL Data in Motion Data at Rest Fast Data Big Data Feeds, Collectors CEP CEP + DB DataInteraction FAST DATA + BIG DATA
  • 18.
    page© 2016 VoltDB Streaming Analytics - Filtering -  Windowing -  Aggregation -  Enrichment -  Correlations Deep Analytics -  Machine Learning -  Statistical correlations -  Multi-dimensional analysis -  Predictive/Prescriptive analytics Operational Interactions / Transactions -  Context-aware -  Personal -  Real-time + Big Data FAST DATA + BIG DATA Fast Data
  • 19.
    page© 2016 VoltDB CUSTOMERSUSING VOLTDB FOR IOT TODAY •  Smart Meters (over 60 million meters under VoltDB management) •  Smart Energy GB •  Smart Metering in Japan (Shikoku and Hokkaido) •  Wearables •  Home Security •  Equipment Service Automation
  • 20.
    page© 2016 VoltDB EQUIPMENTSERVICE AUTOMATION: LEADING STORAGE EQUIPMENT VENDOR USING IOT
  • 21.
    page© 2016 VoltDB21 #Simplify
  • 22.
    page© 2016 VoltDB Proven ü Jepsen Tested Simpler ü  Fewer components and clusters needed ü  Easier to write, test and maintain applications Familiar ü  SQL and Java Glue Code Glue Code Zookeeper “DIY” STREAMING DATA INFRASTRUCTURE VS VOLTDB
  • 23.
    page© 2016 VoltDB EXAMPLE:LARGE TELCO IOT PLATFORM 23 Challenges: •  High volume and velocity of data from smart devices •  Complexity (multiple ingest points, apps, databases) •  Performance – need to automate action on inbound data at the velocity of the feeds Notes •  Applications have their own database •  Each database replicates to Cassandra and Hadoop •  In-memory grid used to maintain logic and publish ‘state’ back and forth •  Rules engine for intra-day data to trigger actions (e.g., ‘turn lights on’) •  PostgreSQL used for dimension data
  • 24.
    page© 2016 VoltDB BEFORE 24 AFTER Results ü Simplified system architecture ü  Single ingest point for high-velocity feeds of inbound data ü  Faster time to value
  • 25.
    page© 2016 VoltDB BEFORE 25 ORPOSSIBLY Results ü  Simplified system architecture ü  Single ingest point for high-velocity feeds of inbound data ü  Faster time to value
  • 26.
  • 27.
  • 28.
    page© 2016 VoltDB QUESTIONS? • Use the chat window to type in your questions •  Download our ebook, Architecting for the Internet of Things (learn.voltdb.com/EbookIoT.html) •  Try VoltDB yourself Free trial of the Enterprise Edition: www.voltdb.com/Download •  Email us at: info@voltdb.com 28