OpenShift Commons Paris - Choose Your Own Observability Adventure
Tmw20101 hayden.j and spaar
1. HOW TO: Move from Data Silos to
Enterprise-wide Data Analytics
Stefan Spaar & Jim Hayden
2. The Possibilities are “Unlimited”
● Unlimited, Flat-Rate Mobile
Voice and Data Services
● Simple, All-Inclusive Pricing &
Predictable Bills
● No Contracts & No Long-Term
Service Commitments
● High-Quality Feature Rich
Devices
● Access to High-Quality
Nationwide 3G and 4G LTE
Networks
● Low-Cost Provider
3. Big Data Paradigm Shift
IT
Structures the
data to answer
that question
IT
Delivers a platform to
enable creative
discovery
Business Users
Explores what questions
could be asked
Business Users
Determine what
question to ask
Monthly sales reports
Profitability analysis
Customer surveys
Brand sentiment
Product strategy
Maximum asset utilization
Big Data Approach
Iterative & Exploratory Analysis
Traditional Approach
Structured & Repeatable Analysis
4. Adopting Variety, Velocity & Volume
Persistent Data In-Motion Data
Traditional
Data
Combination of
Non-traditional/
traditional data
Reuse Warehouse
Data
Filters incoming
data
Real-time
Big Data
Data Warehouse
Variety
Velocity
Volume
6. Big Data Analytics Methodology
• Create a comprehensive 360
o
view of customer in order to monetize our data assets.Goal
• Combine multiple Big Data sources to allow for analytics along any dimension.Process
• Incrementally leverage data produced from ROI based initiatives based on value added.Strategy
7. Big Data – “Goldmine”
• Location Determine the latitude and longitude of your customer at any time..
• Travel Patterns Identify frequent routs that your customer traverses.
• Application Use Distinguish the applications that customers most frequently use.
• Calling Habits Associate call types and call destinations for customers.
• Perceived Service Quality Understand the customer experience with Cricket service.
Customer Behavior
• Music Tastes Characterize customer preferences with music (Muve).
• Browsing Patterns Identify the web sites that customer most frequent.
• Interests Extrapolate customer interests based on search histories.
Customer Preferences
• Social Circles Realize how individuals interact with one another.
• Customer Sentiment Evaluate customer opinions of services or products they purchase.
• Influencers Highlight those individuals that persuade the habits of others.
• Brand Loyalty Determine the brands that our customers choose.
Social Media
8. TEOCO’s Role at Cricket
• What:
Optimize service delivery
costs &margin
• Benefits:
Cost, time and resource
reduction; achieved over 5x
ROI
• What:
Optimize network availability
& performance
• Benefits:
Maximize performance,
capacity and quality
• What:
Optimize RAN network
performance
• Benefits:
Maximize coverage, capacity
and quality
OSS/BSS
Solutions
Big Data
Customer
Analytics
Insights
• Who is using what service?
• How much is being spent?
• When was last use?
• How often used?
• What are common attributes
attributes of customers for
behavior X?
• What are the most popular
services, devices, plans?
• End-to-end network health
• What elements, services,
devices were affected by
network errors?
• What services are seeing
high error rates?
• What services, devices,
customers were affected by
network errors?
• What are the most common
errors?
• Where did errors happen?
• Where are the heavy use
hotspots & deadspots?
• Where is subscriber X, and
where has he been?
• Billions of usage recs XDRs --
Data, SMS, MMS, AAA,
2G/3G/4G Data, Music,
Roaming, etc.
• Customer info
• Product, service & bundles
• Rate plans
• Market
• Hundreds of millions of
events, errors, alarms
• 2G, 3G & 4G network
infrastructure from 3
vendors, Muve Music
servers, PDSNS, etc.
• Billions of 2G/3G/4G
network mobile
measurements from RNCs
Data Sources
Usage Analytics Performance Mgmt RAN Optimization
10. Call & Texting Behavior by Age
-
100
200
300
400
500
600
<18
18to24
25to34
35to44
45to54
55to64
65to74
>75
Average #Texts by Age
Band
-
100
200
300
400
500
600
700
<18
18to24
25to34
35to44
45to54
55to64
65to74
>75
Average #Calls by Age
Band
0%
50%
100%
150%
200%
250%
300%
-
100
200
300
400
500
600
700
<18 18 to 24 25 to 34 35 to 44 45 to 54 55 to 64 65 to 74 >75
Call:Text Ratio by Age Band
Avg SMS Avg Calls Calls/SMS Ratio
14. Future Applications: Subscriber Location
Pattern Analysis
Subscriber 1
Subscriber 2
Subscriber 3
Home: location 837
Work: location 482
Classic 9 - 5
Home: location 919
Work: location 1537
night worker
Home: location 275
Work: location 278
non-standard workweek,
multiple jobs
Location Day of Week/Time of Day Summaries
15. Future Applications: Mobile Advertising geo-temporal
Predict future location of subscriber relative to 3rd party locations
16. Lessons Learned & Next Steps
• Incremental approach beats Big Bang
• Prioritize use cases based on ROI/perceived value
• Engage departmental sponsors
• Don’t get hung up on technology
• Experiment using Analytics Sandbox
• The value of exploratory analytics is harder to
quantify