iot enabled technologies - data preparation and … advanced analytics choose your database based on...
TRANSCRIPT
IoT Enabled TechnologiesDelivering Actionable Insights for the Telecom Industry
Joe Pusztai
VP, Solution Marketing
Datawatch
Syed HodaChief Marketing Officer
ParStream
2
Strong IoT momentum with CEO’s & CIO’s
Predictions 2015IoT software platforms will become the
rage in 2015 and drive IoT Adoption
Top 10 Strategic Technology Trends for 2015
1. Computing Everywhere
2. Internet of Things
3. 3-D Printing
4. Advance, Pervasive Analytics
5. Context-Rich Systems
6. Smart Machines
7. Cloud Computing
8. Software Defined Infrastructure
9. Web-scale IT
10. Risk-Based Security
2015 Tech Predictions
1. Digital transformation
2. Internet of Things
3. Convergence of big data with consumer data
4. Hybrid cloud
5. Collaboration
6. Predictive analytics will lead big data
7. Mobile wearable technology
8. A Platform and orchestration is needed
9. Networked Economy
10. The end of apps
Better IoT data collection and analysis would deliver more value
• 70% say they would make better, more meaningful decisions with improved data
• 86% would increase the ROI of their IoT investment
IoT not delivering full potential because of data challenges
• 86% of stakeholders in business roles say data is important to their IoT project
• Only 8% are fully capturing and analyzing IoT data in a timely fashion
• 94% face challenges collecting and analyzing IoT data
IoT projects vary widely – but all have challenges
• 53% are using IoT projects to optimize existing businesses and 47% as a strategic
business investment
• 96% have faced challenges with their IoT projects (#1 process, #2 users, #3 data)
Global IoT Survey: Key Findings
What does IoT really mean …
Intelligent connections that capture real-time events which enable
companies to transform their Sense and Respond capabilities driving
speed, efficiency, and quality.
The key to generating value from IoT data: Actionable Insights
Data
REAL-TIME DATA INGESTION + IMMEDIATE QUERIES
= ACTIONABLE/TIMELY INSIGHTS
Action
Devices
Devices
Devices
Rules or On-
demand InsightsA
ggre
gation
Imagine a world…
Where IoT analytics enable an energy company to…
30TBAnalyze Data
in Real-time
15%Increase
Efficiency
$18K/hr; $158M/yrGenerate Operational/
Economic Benefits
(20,000 Wind Turbines; 10 GW Capacity; .3 Capacity Factor; $40/MW-hour)
7
IoT analytics has a set of distinct requirements
8
Big DataData is growing faster and bigger
because of more sensors
10B+ rows
5TB+
Fast DataData streamed from sensors
requires fast ingestion
1M+ rows
per sec
Edge AnalyticsIoT data is mostly generated
at the ‘Edges’ of the network
100+
Locations
Real-Time InsightsUse cases require near
Real Time Analytics
<1 sec query
response
time
Existing products don’t fulfill IoT requirements
9
Product
Requirements
Columnar
Databases
Vertica,
Redshift
Row-based
Databases
Oracle,
Informix...
Value
Stores
Cassandra,
MongoDB
Hadoop
Batch
Cloudera,
Hortonworks
Hadoop
Streaming
Spark / Shark
Storm
BIG DATACapacity –
FAST DATA Import – –
EDGEAnalytics Capability – – – – –
REAL TIME Insights – – – – –
INTEGRATEDPlatform – – –
IoT DATAStorage Structure – – –
See details in backup
ParStream and Datawatch introduce the first analytics platform built for IoT
10
IoT Data Collection Platforms Enterprise Data Sources
ParStream DB
Geo-
Distributed
Analytics
Alerts +
Action
Time
Series
Advanced
Analytics
Choose your database based on your use-case
11
< 1..10 ms
1 min
1 hr
10..100 ms
1 sec
10 min
4 hrs
Resp
on
se T
Ime
Big Data
Massively parallel (MPP)
Real-Time
Hadoop / Cassandra / ImpalaOLTP
Reporting
In-Memory DB
Gigabyte Terabyte Petabyte
OLAP
Batch-Analytics
Real-Time IoT
Analytics
Stream-Analytics
Operations
Analytics
Complex Event
Processing
High
Low
ParStream is uniquely positioned for Real-time Analytics in IoT
12
REAL-TIME
IMPORT
REAL-TIME
QUERYING
FLEXIBLE
ANALYTICS
Small Form Factor / Low TCO
Bil
lio
ns
of
Re
co
rds
Thousands of Columns
ParStream has the fastest query response times
13
Environment: Single EC2 XL node with 15 GB RAM, 2 TB disk on Amazon AWS.
OTP data set with 150 Million records. Query set based on customer use-cases.
RedShift
1 second
10seconds
22seconds
31seconds
38seconds
98seconds
Edge analytics delivers real-time insights by minimizing network traffic
14
20 Billion Rows
40 records found
ParStream
ParStream Geo-Distributed Server
3
rows
14
rows
5
rows
12
rows
6
rows
ParStream ParStream ParStream ParStream
40 Rows
ApplicationApplication
Centralized
storage
40 records found
4B
rows
4B
rows
4B
rows
4B
rows
4B
rows
Traditional Analytics Edge Analytics
ParStream introduces EdgeAnalyticsBox• Specifically designed to enable edge analytics
• Ruggedized for use in real-world edge analytics applications
such as oil/drilling sites, cell phone towers, wind farms, etc.
Industry-leading Product Recognition
15
ParStream is the most
reliable System in our Data
Center
CTO, etracker
"ParStream’s ability to
analyze terabytes of data
with sub-second response
time helps us generate
significant value."
President, Envision Energy
ParStream enabled us to
scale internationally - TCO
is much lower than with
Hadoop
VP Eng, Searchmetrics
#1
Big Data
Startup
Cisco Entrepreneurs in Residence 2014 IoT Excellence Award
About Datawatch
NASDAQ: DWCH
Pioneer in real-time visual data discovery
and data preparation
Global operations and support
US, UK, Germany, France, Australia,
Singapore, Philippines
Extensive global customer base
99 of the Fortune 100
12 of the 15 largest financial institutions
Resold and embedded by leading vendors
Visual Data Discovery enabling meaningful and timely decisions for IoTanalytics.
• Single, integrated platform
• Modular approach
– Start on the Desktop and add
Server capabilities when
ready
– Deploy any or all capabilities
• Complements with your
existing investments
– Deploy with other BI tools,
ETL and data warehouse
technologies
Visual Analytics Platform
Datawatch Architecture
AutomationService
ContentRepository
VisualizationEngine
In MemoryCache
Prepare & Design
Top IoT Challenge in Telecom
From 2.4 billion in 2012 to an
estimated 18 billion by 2022
(22% CAGR)
Largely driven by connected
consumer electronics
New services such as
streaming media & all-you-can-
eat LTE are putting huge strains
on network performance
Real-time monitoring of
bottlenecks and demand spikes
are essential to avoid outages
and QoS degradation…
M2M Traffic Explosion
Top Operational Requirements in Telecom
• Real-time performance monitoring to:
– Ensure maximum uptime
– Anticipate and provision for peak demand
– Provide improved customer service
– Increase customer loyalty
– Schedule preventative maintenance
• Real-time usage monitoring & alerting
– Subscriber bill shock remains a chronic problem
– Proactive alerting of roaming, excess data charges, connected devices
are becoming vital to customer service
• Mobile Device Management
– High availability & security of mobile devices and applications are
mission-critical to an increasing number of businesses
Technology Requirements for IoT Analytics
Visual Data Discovery
Streaming Data Visualization
Time Series Data + Multiple Time Horizons
Predictive & Advanced Analytics
Complex File Formats
Real-time Geospatial & Location
1/ Visual Data Discovery
• Easy for users to author,
customize and share
• Interactive exploration &
visually filter results
• Quickly identify
anomalies and outliers
with large or in-motion
datasets
• Rich palette of
visualizations for static
and time series data
Data at Rest
2/ Streaming Data Visualization
Database Distributed or
Hybrid Database
In-Memory
Database
Streaming Analytics
3/ Time Series + Time Horizons
• Traditional BI only looks at coarse
buckets of time:
– Year > Quarter > Month > Week > Day
• Events are continuous and have varying
analysis requirements:
– Second, millisecond, microsecond
– Time windows
– Time slices
– Playback
• Visibility of all time horizons -> complete
situational awareness:
– Now (streaming)
– Intra-day
– Historic
4/ Predictive & Advanced Analytics
• Text Analytics– Entity Extraction
– Sentiment Analysis
• Predictive Analytics– Regression
– Clustering
– Machine learning
• Many IoT Use Cases– Predictive maintenance
– Subscriber churn
– Smart logistics
– Clinical pattern detection
Modeled and
transformed
for analysis
5/ Complex Data & File Formats
• Real-world data is multi-structured
• Often no metadata – must be induced from the data
• Blurred line between “Data in Motion” & “Data at Rest” (e.g. log
file ingestion at one-second intervals)
26
Log Files
HTML, XML JSON
PDFs
6/ Real-Time Geospatial & Location
• Real-time (streamed) plotting
• Geo-coded street-level maps
• Customized SVG files
• Time-series playback
HealthcareRetail
Logistics
Utilities
The Next Wave of Business Transformation
Source: Industrial Analytics: The Next Wave of Business Transformation
Gartner, March 2014
Visualization: Goals -> Design
If Goal Is… Visual Design Includes… Examples
Profitability
High-density displays, multiple data
sources, sophisticated models,
streaming data, detection of “profit-
making events”
Trading desk applications,
marketing automation
ControlDomain-specific data sources, limited
analysis features, positive feedback
loops (alerts and actions)
Energy grid monitoring,
digital oilfield, industrial and
financial process control
applications,
Persuasion
Audience-specific, visually polished
presentation: attractive colors, “story
telling”, infographics
Executive / KPI dashboards,
customer-facing marketing
applications
Knowledge
As many data sources as is required to
gain insight, highly-granular data,
ultimate analysis control (alt.
hierarchies, filters, time-series
playback), statistical discovery
Customer churn analysis,
sentiment analysis, product
traceability, strategy planning
applications
A Blueprint for Your Integrated IoT Analytics Platform
RulePoint
CEP
real-time
analytics
Visual & search
based data
discoverySensor Data
Stream
Real-time
actions
Real-time/Historical
reporting
Discover &
Deploy
Derive
Connectivity to “Things” (sensors, actuators)
IoT PlatformDevice-level protocols,
transport, security
Real-time
alerts
DatawatchVisual
Analytics
ParstreamAnalytical Database
Success Factors for your IoT Program
Find OT + IT convergence opportunities with dollarized business value (e.g. true per-subscriber profitability)
Leverage new generation of low cost sensors to create new data sources and deliver holistic system-wide view
Define “build or buy” criteria for every element of your IoT blueprint:
Purpose-built vs home-grown device & transport platforms
Commercial vs open-source for databases, event processing, predictive, visualization
Cloud vs on-prem vs hybrid deployment
Do not treat IoT initiatives as “IT projects”:
OT factors much larger than IT
Flexibility and rapid insights are more important than transactional integrity
Need cross-functional buy-in from non-traditional stakeholders
Sales, Marketing, Operations, Product Development
New mix of skills e.g. an individual “data scientist” rarely has the entire breadth of skills and knowledge
32
Next Steps
www.parstream.com
Download the IoT Industry Survey at:http://sites.parstream.com/parstream-iot-survey-whitepaper
• Get in touch www.datawatch.com/contact-us/
• Discover more about Datawatch solutions:
• Explore: www.datawatch.com/explore/
• Evaluate: www.datawatch.com/free-trial/