ibm big data summit 2012 · • failed rab attempt counts for • lack of ul channelization code...
TRANSCRIPT
Agenda
Success Criteria for Big Data
Guidance for Quickstart
Typical Problem Areas
A few IBM Best Practice Examples
stream computing
Four years ago, we started
working with organizations
to build a smarter planet.
Through thousands of client
engagements, we learned that
analytics is fundamental
to success.
• From business initiative to business imperative
• From enterprise data to big data
• From advancing single organizations to
transforming entire industries
Since then, analytics has
continued to evolve:
Imagine the Possibilities of Harnessing your Data Resources
Retailer reduces time to run
queries by 80% to optimize inventory
Stock Exchange cuts
queries from 26 hours to 2 minutes on 2 PB
Government cuts acoustic
analysis from hours to 70 Milliseconds
Utility avoids power failures
by analyzing 10 PB of data in minutes
Telco analyses streaming network data to reduce
hardware costs by 90%
Hospital analyzes streaming
vitals to intervene 24 hours earlier
6
7 7
Public wind data is available on 284km
x 284 km grids (2.5o LAT/LONG)
More data means more accurate and
richer models (adding hundreds of
variables)
- Vestas wind library at 2.5 PB: to grow to over 6 PB in
the near-term
- Granularity 27km x 27km grids: driving to 9x9, 3x3 to
10m x 10m simulations
Reduced turbine placement
identification from weeks to hours
Perspective: The Vestas Wind library 7
University of Ontario
Institute of Technology
(UOIT) Detects Neonatal
Patient Symptoms Sooner
• Performing real-time analytics using
physiological data from neonatal babies
• Continuously correlates data from medical
monitors to detect subtle changes and alert
hospital staff sooner
• Early warning gives caregivers the ability to
proactively deal with complications
Significant benefits:
• Helps detect life threatening conditions up
to 24 hours sooner
• Lower morbidity and improved patient care
Capabilities Utilized:
Stream Computing
“Helps detect life threatening conditions up to 24 hours sooner”
Asian Health Bureau
reduces diagnostic
errors
• Telemedicine imaging diagnostics service
to improve rural healthcare
• Automatically sifts and analyzes large
collections looking for anomalies and
disease
• Makes it possible for radiologists and
Pathologists to analyze:
1000s of patient images
Significant improvements
expected:
• Reduction in diagnostic errors
• Improved outcomes by leveraging
physicians treating similar cases
“Over 80% of healthcare data is medical imaging”
Capabilities Utilized:
BigInsights
10
Dublin City Centre Increases
Bus Transportation
Performance
10
• Public transportation awareness solution
improves on-time performance and provides
real-time bus arrival info to riders
• Continuously analyzes bus location data to
infer traffic conditions and predict arrivals
• Collects, processes, and visualizes location
data of all bus vehicles
• Automatically generates transportation
routes and stop locations
Results
• Monitoring 600 buses across 150 routes
• Analyzing 50 bus locations per second
• Anticipated to Increase bus ridership
Capabilities Utilized
Stream Computing
Pacific Northwest Smart
Grid Demonstration
Project
Capabilities Utilized
Stream Computing
Data Warehouse Appliance
Results
• Demonstrates scalability from 100 to 500K homes while retaining 10 years’ historical data
• 60k metered customers in 5 states
• Accommodates ad hoc analysis of price fluctuation, energy consumption profiles, risk, fraud detection, grid health, etc.
12
Asian Operator - Real-time
predictive analytics to reduce
customer churn
Reduce time for launching promotions
Real Time Rewarding for eligible subscribers
Centralised promotion management system
Measure the financial impact of campaigns
Analyse customer behaviour for trend analysis
CDRs can be analyzed within 30 seconds
Anticipated improved campaign response rates by 25%
Forecasting churn reduction of 15-20%
Stream Computing
13
Indian telco reduces billing costs & improves customer satisfaction
Capabilities: Stream Computing & Analytic Accelerators
Real-time mediation and analysis of 6B CDRs per day
Data processing time reduced from 12 hrs to <1 sec
(Hardware cost reduced to 1/8th)
Where do Enterprises get Stuck with Big Data?
Shortage Skilled
Resources
Not able to
understand the
potential of Big
Data
One-Size-Fits-All
Vague Goals
(e.g. lack of use
cases)
Lack Executive
Support
Processing
Bottlenecks
14
Besides factors to be managed for each project:
•Budget, time, scope, resources, detailed specification, solution, project management
In Big Data projects you should specifically spend an additional effort on:
1.Data privacy / security – Compliance is key !!!
2.Clear Business Scope: Successful Projects are initiated by business but not by IT
3.Appointment of Data Stuarts, Data Analysts as fix job role in project and business
organization
4.Understand data as the true asset but not the systems hosting data
5.Understand the data
6.Data volume and velocity push into a new dimension processes will change
7.Data variety (by constantly new sources and formats) requires a build-in flexibility from
start on
8.Attention to Ease of Use to satisfy user of solution environment
Specific Success Factors for Big Data Projects
23.10.2012 15 © Copyright IBM Corporation 2012
Guidance for Quickstart
1. Start with Data Compliance Issues (privacy, security, data sources, …)
3. Specify Business Requirements and Sucess Criteria
4. Validate Business Benefit (Use Cases)
5. Specify selected Use Cases in detail (LoB + IT together)
0. If possible try to attend a live-demo to feel the new capabilities of Big Data
6. Map Business und Use Case Requirements to Functional Requirements
7. Define Solution Architecture
9. Provide Business Case
2. Start little (specific scope) expand later (often POC mature for production)
8. Capability Check
10. Start Implementation Project
Big Data is a strategic but not a tactical move
Source: http://blog.digital.telefonica.com/?press-release=telefonica-launches-telefonica-dynamic-insights-a-new-global-big-data-business-unit
- New Business Unit for Big Data
„Telefónica Dynamic Insights“
- Partnership with GfK
- The first product, ‘Smart Steps’,
will use fully anonymised and
aggregated mobile network data
to enable companies and public
sector organisations to measure,
compare, and understand what
factors influence the number of
people visiting a location at any
time.
Note:
GfK is one of the world’s largest research
companies, with more than 11,500 experts
working to discover new insights into the way
people live, think and shop
Press Release 09.Oct 2012
Big Data is a strategic but not a tactical move
http://www.fiercemobilecontent.com/story/verizons-hillier-discusses-data-privacy-and-future-mobile-marketing/2012-10-05
Press Release 05.Oct 2012
New Devision at Verizon:
„Precision Marketing“
„We realized we had a latent asset”
“We looked at the clusters of
demographic makeup for each of their
events and found out interesting
things about the types of consumers
that attended their events--from what
type of event, the time of day, their
record and other environmental
conditions that were occurring in their
market” (for a sport stadium)
“Data is the New Oil” – It is up to you to refine it !
In contrast to oil we know already where to look
for data. Very often we own it already !
• Top Handsets
By:
Voice Minutes of Use
Voice Call Counts
Occurrence on Network
Voice Dropped Call Counts
Voice Access Attempts Failures
SMS Counts
Period (Trend)
Network (2G/3G)
Breakdown:
Manufacturer
Model
Handset
20
• Voice Call Attempts
• Successful Voice Call Counts
• Successful (Zero Duration)Voice Call Counts
• Voice Call Minutes
• Voice Call Erlangs
• Average Call Duration
• SMS Counts
Data By:
Cell ID
LAC
Market
Region
Handset Manufacturer
Handset Model
Network (2G/3G)
Period (Trend)
Direction
Breakdown:
National/International
Roaming
Service Usage
• Voice Call Drop Counts
• Voice Call Drop Percentages
• Voice Call Access Failures
• Voice Call Access Failures Percentages
• SMS Failure Counts
• SMS Failure Percentages
Data By:
Cell ID
LAC
Market
Region
Handset Manufacturer
Handset Model
Network (2G/3G)
Period (Trend)
Direction
Service Quality
• Total Subscribers
• Subscribers &Congested Cell Sites
• Subscribers and Failing Handsets
Subscribers
• Voice Accessibility UMTS
• Data Accessibility UMTS
• Dropped Speech
• Failed RAB Attempt Counts for
• Lack of UL Channelization Code
• Failed DL ASE
• Lack of UL Hardware
• Failed RRC/RAB Establishments After Admission Control Count
• Rejected RRC/RAB Establishments after Admission Control Count
• Top N Congested Cell Sites
Data By:
• Cell ID
• LAC
• Market
Radio Quality
Example: KPIs
• Region
• Direction
• Period (Trend)