Transcript BI on the Grid: A Customer Perspective

BI on the Grid: A Customer
Perspective
March 23, 2009
Copyright © 2008 SAS Institute Inc. All rights reserved. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Agenda
 Selection of SAS E-Miner on a Grid as our
analytic solution
 Systems Architecture implications
 System Implementation
Copyright © 2008, SAS Institute Inc. All rights reserved. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
ISO Background
 ISO
• Started in 1971
• Leading source of information about risk
• Supplies data, analytics, and decision support services
• Originally focused on property & casualty insurance
• Now also serves healthcare and mortgage industries
and the general risk management function in all
industries
Copyright © 2008, SAS Institute Inc. All rights reserved. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
ISO Innovative Analytics (iiA) division
 iiA
• Formed in 2005
• Located in San Francisco
• Expertise: Statistical Modeling, Actuarial Science, Data
Management
• Mission: Apply data mining, predictive modeling, and
other advanced analytic techniques to create new
products and enhance existing products
Copyright © 2008, SAS Institute Inc. All rights reserved. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Situation in 2008
 Staff
• Rapid Growth – iiA staff numbered about 20 at the
beginning of the year with plans to grow to 35 by year
end
• Highly Technical – about 2/3 of staff are M.S. and Ph.D.
level statisticians and mathematicians. Most of the rest
are DBA’s and data analysts.
 Systems
• Infrastructure – had evolved over time – 12 analytic
servers plus analyst workstations.
• Tools – wide variety, no standardization -- SAS,
Statistica, R, etc.
Copyright © 2008, SAS Institute Inc. All rights reserved. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Needs
 Efficiency
• Capacity – Our analysts are highly skilled and valuable.
Tools must enable them to work at max capacity.
• Standardization – Difficult to do consistent
documentation, training, quality control, and product
launches with disparate tools.
 Scalability
• Rapid growth required a system that could expand
easily as the organization grows.
Copyright © 2008, SAS Institute Inc. All rights reserved. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Analytic Tool Selection
 Task force coordinated by ISO systems architect
and including iiA analytic staff and management.
 Broad market scan of possible analytic tools
 Narrowed list to 3 finalists
 Received detailed proposals and demos from
finalists along with more intensive investigation
by task force
Note: Since this was a new system, tool choice
would drive system architecture
Copyright © 2008, SAS Institute Inc. All rights reserved. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
SAS Enterprise Miner of the GRID
 Technical Factors
• Scalable – Easy to increase capacity with additional
commodity hardware
• Capable – Best at handling very large data sets for
analysis
• Parallel Processing – Explicit and built-in parallelism
 “Soft” Factors
• Most analysts already familiar with SAS
• “Legacy” SAS code could be run on new system
• Robust technical support system
• Comprehensive training available
Copyright © 2008, SAS Institute Inc. All rights reserved. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Customer Data
Secure
Data
Area
Sanitized
Data
Area
iiA Systems Block Diagram
SAS
ModelModels
Manager
SAS Analytic
SAS
Advanced
E-Guide E-Miner
SAS
Platform
JMP
GRID
Scoring
Platform
Text
Mining
Out to Customer
GIS
Analytics
3rd Party Data
Data Warehousing Environments
ISO Data
MS SQL
Server
2008
Copyright © 2008, SAS Institute Inc. All rights reserved. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
System Architecture
 ISO IT team worked with SAS to develop a
system architecture specific to iiA’s needs
• IBM Blade servers formed the basis of the GRID
• MS Windows-based due to ISO’s expertise supporting
this OS
• Since most users are remote from the server location,
established Virtual Machines to act as gateway to the
GRID for users
• Also built a new data management system based on
MS SQL Server 2008 that will eventually provide 18 TB
of data storage
Copyright © 2008, SAS Institute Inc. All rights reserved. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Users
Virtual Machines
“The Grid”
Copyright © 2008, SAS Institute Inc. All rights reserved. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Data Warehousing
Environments
Conclusion
 Just finished User Acceptance Testing of the
system
 Results look promising so far but now comes the
real test … 30+ users on the system!
Copyright © 2008, SAS Institute Inc. All rights reserved. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
QUESTIONS?
Copyright © 2008, SAS Institute Inc. All rights reserved. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Copyright © 2008, SAS Institute Inc. All rights reserved. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.