nasscom gic conclave 2013 - session 3 b - analytics as a service - oliver ratzesberger
DESCRIPTION
TRANSCRIPT
Analytics as a ServiceBigData in Private Clouds
Oliver Ratzesberger
VP Information Analytics & Innovation
@ratzesberger
Oliver Ratzesberger – VP Analytics & Innovation• 20 years in Large scale Data Warehouse
• 7 years at eBay – Analytics PlatformTeradataHadoop
100PB of infrastructure – largest commercial database sized for >50PB of raw data
• At Sears Holdings/MetaScale since October 2011Transforming a legacy icon into an Analytical Competitor
What is BigData?PetaBytes of information
Hundreds of Millions of CustomersComplex/Semi/Unstructured Data
NoSQL/MapReduce/MPP/HadoopData Science & Data Visualization
Advanced Algorithms & Predictive TechnologiesNatural Language & Image Processing
Sensor DataSentiment Analysis
BigData at SHC/MetaScale
3.5PB EDW(w/pCloud) 2.5PB Hadoop
>15 Million requests per day
Consolidating all Data Marts into a Single Version of the Truth
Simplicity
Occam’s Razor:“simpler explanations are …
generally better than more complex ones”
The simple solution is easy to explain, implement,
and maintain
Design for the Unknown
“Of design for analytics platforms - Perfect is Wasteful”
Friction to change & code weight are the antithesis of agility
Time to Market ( is everything …)
Are your Analytical needs getting stuck in traffic?
The Iceberg Problem
Physical Data Marts
are like Icebergs: 90% of their cost is ‘hidden’
A ‘free’ Physical Data Mart is too expensive to justify its
existence
HR
Stores
FinanceInternational
Finance
Online
Loyalty
CRM
Marketing
IT
Supply Chain
Scrum – Adopting an Agile Methodology
Amount of Change
Competing Priorities in Technology
What is DevOps?• Blend of
Agile Development AND
Agile Operations
• Software development methods that stress
communication and collaboration
• Developing the 1st line of code with Operations in mind
The Foundation
Technology Platform Storage and processing platforms, Teradata & Hadoop, and data interconnect services
Analytics as a Service (A3S)Reusable, powerful, and integrated analytics services that automates the actions in an analytics environment. This enables rapid deployment of a high-quality feature rich collaborative analytics environment that will empower users to be radically more self sufficient, be more productive, and achieve better results.
Insights PlatformAdvanced analytics products with out of the box segmentation, trending, alerting, experimentation, etc. capabilities supporting extremely large data sets
Serv
ices
, Tra
inin
g, S
uppo
rt
Dev
elop
er P
latfo
rm
Example Prototype developed in pCloud
Daily Summary
Triple Intersection
Duple Intersection
KPIs / Segment IntersectionSegment Intersection Populations
KPIs / segmentSegment populations
Daily Detail
Segment Definition
Export Segments
Define Logic - submit to engineTag members with attributes
Member Details / Segment
pCloud enabled Developer Platform
ODE - Advanced Platform Analytics
pCloud Consumption – On Demand Analytics
Elastic Capacity
Leverage pCloud by the hour to support spikes in demand
Develop new prototypes
Active Management – decide when to use pCloud
SEARS HOLDING CORPORATION COPYRIGHT 2012 22
Separating GOOD from BAD
SEARS HOLDING CORPORATION COPYRIGHT 2012 23
Consistent Simplicity
SEARS HOLDING CORPORATION COPYRIGHT 2012 24
Data Science - When the AVERAGE is useless
Questions?
Oliver Ratzesberger
VP Information Analytics & Innovation
@ratzesberger