Download External overview for field use

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Database wikipedia , lookup

Entity–attribute–value model wikipedia , lookup

Microsoft Jet Database Engine wikipedia , lookup

Extensible Storage Engine wikipedia , lookup

Microsoft SQL Server wikipedia , lookup

SQL wikipedia , lookup

Open Database Connectivity wikipedia , lookup

Relational model wikipedia , lookup

Clusterpoint wikipedia , lookup

Object-relational impedance mismatch wikipedia , lookup

Database model wikipedia , lookup

Functional Database Model wikipedia , lookup

Transcript
SAS and Teradata Partnership Overview
October, 2007
Headlines
SAS and Teradata Enter into Strategic Partnership
BusinessIntelligence.com – 10/8/07
“In-database analytics is a key development that promises to improve
efficiency and effectiveness of business analytic solutions. It will
decrease data movement and increase performance, thus enabling
IT to better respond to the decision support needs of business
decision makers.”
Dan Vesset, Vice President, Business Analytics, IDC
2
SAS - Teradata CEO’s Announcing Strategic Partnership
October, 8th 2007
Mike Koehler, president and chief executive officer (CEO) of Teradata
Jim Goodnight, chief executive officer (CEO) of SAS
3
Session Topic:
Benefits of SAS and Teradata Partnership
 Jeff Mudd – Senior Account Executive
Support All Operating Divisions within DHHS
 Maximize Value via SAS Enterprise License Agreement (ELA)
 CMS SAS users supporting key initiatives:
Center for Medicaid Management (CMM)
Office of Actuary (OAC)
Office of Clinical Standards and Quality (OCSQ)
Office of Research and Development (ORDI)
Program Integrity / Program Safeguard
 SAS & Teradata integration leveraged across most SAS offerings
 What does this announcement and integration mean for CMS ?
4
IT Infrastructure to Support CMS Mission
MACs
FFS
PLANs
Part C & D
States
Medicaid
QICs
Enterprise Data Center (EDC)
Industry
Partners
Streamlined
MMA
Applications
HIGLAS
Modernized FFS
Claims Applications
Medicaid
Related
Applications
Medicare
Appeals
System
Web Services
QIO Related
Applications
Integrated Data Repository
Medicare
Beneficiary
Medicare
Part A & B
Claims
Medicare
Part D Claims
Medicaid
Beneficiary
Enrollment
Standard Interfaces (EDI)
SSA
OPM
RRB
Standard Interfaces (Portal)
Standard Front End
e-Health
Initiatives
e-Gov
Initiatives
Medicaid
Claims
Source: CMS
RDDC
CMS Internal Users
•Internal Extracts
•Public Use Files
HHS Opdivs
(FDA, NIH…)
FHA Partners
(VA, DOD…)
QIO’s,
Program
Integrity
5
Agenda
• Marketplace Challenges
• Partnership Vision and Benefits
• Outside Perspective
• Joint Development Roadmap
• SAS and Teradata Center of Excellence
• Summary
6
Market Trends on Decision Making
Stayed the
same,
29%
“…the number of
decisions you make
daily has...”
“… the amount of
data available to
you is...?”
Increased,
68%
More Decisions
68% say number of
daily decisions has
increased over last year
“… the complexity
of the business
decisions you are
making has ...?”
Decreased ,
4%
Less complex,
9%
More
complex,
45%
More Complexity
45% say decisions are
more complex
About the
same, 46%
Increasing,
4%
Increasing
slightly,
40%
Not
increasing,
4%
Tripling, 8%
More Data
96% say data is increasing
52% say data is doubling
or tripling every year
Doubling,
44%
Source: The 2006 Teradata Report on Enterprise Decision-Making
7
Technology Barriers to Decision Making
Limitations:
> Insufficient analytical capabilities and usage
> Inability to capture value from growing volumes of data
> Data movement and data quality issues
> Data redundancy & high infrastructure costs
> Database architectures that do not scale
Requirements:
>
>
>
>
>
Faster analytic answers, faster time to market
Reduced data movement and latency issues
Enhanced effectiveness - analysts focused on higher value tasks
Improved data quality and data consistency
Lower total cost of ownership and investment protection
8
SAS and Teradata
Strategic Partnership Vision and Value
Already Hundreds of Shared Customers Globally!
Partnership vision
• To create breakthrough
customer value
> True scalable analytic
solutions
> Improved time to value
> Reduced technology
infrastructure costs
Enhanced
Performance
Enhanced
Productivity
•Entertainment
Company
• Financial Services
Company
> Reduces analytic
processing from 36 to
one hour
> Opportunity to
redeploy 500 SAS
analysts from data
prep to customer and
fraud analysis
More Choice
Lower TCO
• Major North America
Insurer
• Major Bank
> Reviewing new SAS BI
opportunity with
Teradata due to data
integration
and partnership
> Eliminates data
redundancy and
reduces IT costs
by integrating SAS
and Teradata
9
SAS and Teradata Background
SAS
Teradata
• Global reach, local presence
• Leader in Data Warehousing
> Global presence
> 43,000 customer installations
> Hundreds of local user groups
> $1.9B revenue in 2006
• Market leader for:
> Business intelligence
> Data quality and integration
> Data mining and analytics
> Horizontal/Vertical Solutions
• Breadth and depth of tools
> General/specialized analytics
> 40+ countries worldwide
> 850+ global customers
> $1.6B revenue in 2006
• Enterprise view
> Smarter, faster decision making
> Analytical technologies and
solutions
• Highest performing technology
> Parallel environment
> Speed and scalability
> Relational view of data
10
The SAS and Teradata Strategic Partnership
• Teradata is SAS’ first RDBMS partner for its
“In-Database” initiative
• Joint Product Roadmap
• Dedicated R&D for Optimizing SAS for Teradata
• The SAS and Teradata Center of Excellence
• SAS and Teradata Executive Customer Advisory
Board
11
Outside Perspective
"In-Database Analytics is a key development that promises to
improve efficiency and effectiveness of business analytics
solutions…”
> Dan Vesset, Vice President, Business Analytics, IDC
"The ability to achieve a higher level of integration with both
SAS and our Teradata platforms will provide both increased
operational efficiencies and enhanced information analytics...”
> Mark Halloran, CIO, Medco Health Solutions, Inc.
“As a result of integrating SAS and Teradata technologies, we
have reduced overall processing time to run our forecasting
model from 36 hours to 1 hour and 15 minutes.”
> Thomas Tileston, Vice President of Business Decision Support, Warner
Home Video
12
SAS® In-Database Processing for Teradata
Current
Capabilities
Future
Option
Analytic
Modeling
Analytic
Modeling
SAS
SAS
Data
SAS
Scoring
Data
SAS
Modeling
Teradata
EDW
SAS
Scoring
Teradata
EDW
13
Joint R&D Roadmap
4Q 2007
1st Half 2008
2nd Half 2008
2009+
Phase 0
Phase 1
Phase 2
Futures
• Optimize current tools
and solution approach
• Begin integration of
SAS functions into
Teradata
• Optimize SAS
functions within
Teradata
• Complete ELT support
• Certify Retail
applications on Teradata
• Certify joint data
mining solution
> SAS/Enterprise Miner
> Teradata ADS
Generator
> SAS/Teradata
investments
• Improved data
integration options
• Integrate SAS AML
Analytics with Teradata
• SAS Credit Risk
database on Teradata
• SAS Risk solution on
Teradata
> Map SAS to
Teradata Financial
Services Logical
Data Model
• Continued evaluation of
industry solutions on
Teradata
• Enhance Teradata
Demand Chain Mgmt
• Managed SAS/Teradata
environment
Benefits
• Improve SAS end-user
productivity
• Maximize current
SAS/Teradata
investments
• Minimize data
movement
• Deploy more SAS
models in Teradata
• Faster data access
• SAS analytics in
Teradata
• Improved data
integration
• Optimize SAS/Teradata
investments
• Minimize data
movement
• Tighter application
integration
14
The SAS and Teradata Center of Excellence
Business Value:
A strategic dedicated team of solution & technical architects that can consult with
customers and assist them in developing a roadmap to optimize & improve the
efficiency of their technology platform’s performance.
Domain Knowledge and Best Practices:
>
>
>
>
>
SAS products, applications, and solutions
Data integration, business intelligence, and analytical software deployment
Teradata data warehousing, client tools, and data modeling
Financial Services, Insurance, Healthcare, Retail and Communications
IT infrastructure (servers, storage, database, network)
Deliverables include:
>
>
>
>
Architecture assessments and roadmap recommendations
Proof-of-concepts
Benchmarking and sizing analysis
Customized consulting services
15
Summary
Together, the SAS and Teradata partnership delivers:
> A compelling and robust business intelligence and analytics
environment from two industry leaders
> Solutions that allow companies to focus on higher value
business opportunities
– Expands the use of analytics to increase competitive advantage
– Delivers top-line and bottom-line growth faster
> A reduction in the complexity and cost for decision making
– Reduced data movement and latency issues
– Improved data quality and data consistency
– Lower total cost of ownership and investment protection
16
Highlights from Initial SAS
and Teradata R&D Integration
Experiments …
• Based on a multi-node Teradata machine
• Move work from a SAS procedure into the
database
• Move SAS formats to the database
• Move a scoring application into the database
• Executing SAS stored processes from the
database server
18
Experiment #1
• Make PROC FREQ more database aware
19
Descriptive Statistics – Your basic Crosstab
proc freq data=credit_data;
table state * credit_score;
20
Proc Freq Today
Teradata
SAS® Session
proc freq
table state*credit;
SQL
Access Engine
21
SAS/ACCESS
MVA SAS®
Proc Freq
SAS/ACCESS to Teradata
• Multiload support
• Fastload support
• Fastexport support
• Multi-statement insert
•TPT support (near future)
• Implicit and Explicit SQL support
SAS I/O Supervisor
Access Engine
22
Proc Freq Today
Teradata
SAS® Session
proc freq
table state*credit;
Request All Rows
SQL
select state, credit
Access Engine
23
Proc Freq Today – BIG data pull
Teradata
SAS® Session
proc freq
table state*credit;
Request All Rows
SQL
select state, credit
Access Engine
24
Future: Proc Freq
Teradata
SAS® Session
proc freq
table state*credit;
SQL
Access Engine
25
Future: Proc Freq – Smarter SQL
Teradata
SAS® Session
proc freq
table state*credit;
select count(*),
state, credit from …
group by state, credit
SQL
Access Engine
26
Future: Proc Freq – Smarter SQL
Teradata
SAS® Session
proc freq
table state*credit;
select count(*),
state, credit from …
group by state, credit
SQL
Access Engine
27
Future: Proc Freq – Smarter SQL
Traditional
Freq
Test
Rows
Returned
Push Down
Elapsed
Time
Rows
Returned
Elapsed
Time
1
9,000,000
55
51
2
Total
9,000,000
55
51
2
28
Experiment #2
• Running SAS Formats on the database server
29
Bin States in Regions – There is always a
catch.
proc freq data=credit_data;
format state $region.;
table state * credit_score;
30
Region Format – Something a User would do
proc format;
value $region
'AL'='South'
'AK'='West'
'AS'='Other'
'AZ'='West'
'AR'='South'
'CA'='West'
'CO'='West'
...
31
User formats require managed database
objects
Teradata
SAS® Session
region_fmt.xml
SQL
proc format;
value $region
'AL'='South'
'AK'='West'
'AS'='Other'
'AZ'='West'
'AR'='South'
'CA'='West'
'CO'='West'
...
SAS Format Library
putc()
32
Proc Freq Tomorrow
Teradata
SAS® Session
proc freq
table state*credit;
SQL
select count(*),
putc(state, "$region.")
as region, credit from…
group by region, credit
Access Engine
SAS Format Library
putc()
33
Proc Freq Tomorrow
Teradata
SAS® Session
proc freq
table state*credit;
SQL
select count(*),
putc(state, "$region.")
as region, credit from…
group by region, credit
Access Engine
SAS Format Library
putc()
34
Proc Freq Tomorrow – Smarter SQL
Calling SAS provided UDF’s
Traditional
Freq
Test
Rows
Returned
Push Down
Elapsed
Time
Rows
Returned
Elapsed
Time
1
9,000,000
55
51
2
2 w/
format
9,000,000
151
51
11
3-Sort
10,000,000
592
3-Freq
0
5
109
3
4
1,499,600
29
51
3
5
9,980,000
180
15,767
11
39,479,600
1,012
15,818
32
Total
35
Experiment #3
• Move the Scoring work to the Database…
36
Scoring Today
BIG data pull and push
Teradata
SAS® Session
data out;
set cust;
<<score code>>
Request All Rows
SQL
select * from cust
Access Engine
37
Scoring Tomorrow – Score at the data
Teradata
SAS® Session
proc sql
create table out as
(select a,b,c,
score(a,b,c) as score…
SQL
create table out as
(select a,b,c,
score(a,b,c) as score…
Access Engine
SAS TSPL Library
score()
38
Scoring – 6,000,000 rows
Insert into sas_usr.myresults
Sel clage,clno,debtinc,loan,"VALUE",job,reason
,ninq,delinq,mortdue,yoj,derog,id
,sas_usr.udf_i_bad(clage,clno,debtinc,delinq,derog,j
ob,loan,mortdue,ninq,reason,"VALUE",yoj) as score
From hmeq;
224
225
227
228
229
230
231
data reslib.results (bulkload=yes);
set scorelib.hmeq_score;
*------------------------------------------------------------*;
* EM SCORE CODE;
* VERSION: 5.2;
* GENERATED BY: carynt\sasled;
* CREATED: 09AUG2006:13:57:12;
39
In Database Scoring vs.
External Scoring
800,000
750000
Obs / Second Scored
700,000
625000
600,000
500,000
500000
400,000
375000
300,000
250000
200,000
Optimized I/0
125,000
100,000
16,400
77,040
0
1
2
3
4
5
6
Number Teradata Nodes
External Process
In place via UDF
40
Experiment #4
• Executing SAS stored process server from DBMS
41
SAS Stored Process - Simple Values
Call sasStoredProcess(“calcRisk”,balance)
Select …
/*
Any sas code
*/
SAS
$20,000.00
Foundation
Teradata
account
42
Experiment Results
• Reduced Data Pull
• Execute SAS Specific Syntax on the DBMS
• Leverage MPP Scalability of Teradata
• Processing Options
43