2. Overwhelmed by Data ?
Big
Data
Sensor –
Machine
based data
Clik Stream
Data
Mobile
Information
Geo-‐‑
localization
Data
Sentiment
Data
3. 80% of Telecom Data will land on
Hadoop
• Hadoop is like a data warehouse but can store a huge
amount of data, different kinds of data and perform
more flexible analyses
• Hadoop is open source and run on industry standard
hardware : It’s ½ more economical than conventional
data warehouse
• Hadoop provides more cost effective Storage,
Processing & Analysis
• Hadoop delivers a foundation for profitable growth :
Gain value from all your data by asking bigger questions
7. From Data Warehouse to Hadoop
The Challenge
• Many Data-warehouses
are out of capacities
• Running out of budget
before running out
relevant data
• Older data archived in
dark with no access
Data Warehouse
The Solution
• Hadoop for data storage
& data processing : Parse,
clean, apply structure &
transform
• Free framework and data
warehouse
• Retain all data for Analysis
Data Warehouse
Hadoop
Operational (44%)
Analytics (11%)
ETL Processing (44%)
Operational (50%)
Analytics (50%)
Storage & Processing
Cost is
1/10
8. Disrutive innovation on Big Data
Traditional
Data base
Data
Warehouse
MPP
Analytics
No SQL
Data Base
Hadoop
Structured
Data Types
Any types including
unstructured
Pre-‐‑defined, fixed, required on
write
Schema
Required on read
Store First.. Ask question
later
No or limited data processing
Processing
Processing coupled with data
Parallel processing / Scale
out
Enterprise grad
Mission critical
Physical
Infrastructure
Commodity is an option
Much cheaper option
9. How Can I Use Hadoop ?
Extraction,
Transformation
& Storage
Risk Modeling
Fraud
Detection
Customer
Churn
Analysis
Ads Targeting
Point of Scale
Transactional
Analysis
Quality Search
Social Analysis
Real Time
Analytics